Please select what you would like included for printing: Copy the text below and then paste that into your favorite email application Mary was a graduate of Meadowdale High School and had attended Chaminade Julienne High School she participated in folkloric dancing with the Hispanic Brotherhood Center of Dayton She worked at Johnson Electric for a number of years until her health was compromised Mary was a member of Corpus Christi Church as a child As a young person she participated in the Dayton Hispanic Ministry where she grew to love The Virgin Guadalupe This past year she was attending Our Lady of Grace Church Mary shared her love of Disney with her and was excited to take her to Disney World with Michael on several occasions Mary snorted when she laughed and had a smile that was contagious The doctors and nurses soon caught on to her style of humor and enjoyed being with her while helping her as a patient Mary had a special place in her heart for those she loved her own and even the stray cats that showed up at the house Although Mary struggled with her health this past year Visitation will be held from 10:00 am - 12:00 pm on Saturday where the funeral service will begin at 12:00 pm. Mary's service will be livestreamed and can viewed by clicking the link below To share a memory of Mary or to leave a special message for her family Enter your phone number above to have directions sent via text Click to watch This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply Service map data © OpenStreetMap contributors OK (KOKH) — Them Kasey Alert issued for Isabell Jean Marmolejo has been canceled The Oklahoma Highway Patrol has issued a Kasey Alert on behalf of the Del City Police Department for 19-year-old Isabell Jean Marmolejo Marmolejo who has developmental delays was last seen walking in Del City It is not known as to what she was last seen wearing Officials said Marmolejo could become violent without her medication BEXAR COUNTY Texas – A man was arrested Tuesday for allegedly assaulting three two-year-old children at a far Northwest Bexar County preschool according to the Bexar County Sheriff’s Office was reported to have “used force” on three children at Excelled Montessori Preschool Surveillance video shows Marmolejo grabbing a two-year-old child by the arm dragging and aggressively slamming the child on a chair Marmolejo then appeared to be seen on video slapping another student on the back of the head with his hand He then walked toward the back of the classroom causing that child to hit their head and also fall to the floor Marmolejo is accused of yelling at the third child He also dragged and slammed that student onto a chair Surveillance video then showed Marmolejo picking up and kicking several items around the classroom “I did not think what happened in the classroom was that bad.” He said that “his emotions got the best of him.” Marmolejo faces three counts of injury to a child and is waiting to be booked in Bexar County jail If anyone has additional information on Marmolejo, BCSO asks to call 210-335-6000 or email bcsotips@bexar.org Copyright 2024 by KSAT - All rights reserved Madalynn Lambert is a Content Gatherer at KSAT-12 She graduated from The University of Texas at Austin in 2024 with a degree in journalism and minors in global and science communication Email Newsletters KSAT RSS Feeds Contests and Rules Contact Us KSAT Internships Careers at KSAT Closed Captioning / Audio Description Public File Current EEO Report Terms of Use Privacy Policy Do Not Sell My Info FCC Applications Copyright © 2025 KSAT.com is managed by Graham Digital and published by Graham Media Group « Back This website is using a security service to protect itself from online attacks. The action you just performed triggered the security solution. There are several actions that could trigger this block including submitting a certain word or phrase, a SQL command or malformed data. You can email the site owner to let them know you were blocked. Please include what you were doing when this page came up and the Cloudflare Ray ID found at the bottom of this page. He is currently the Director of Partnerships at the EAAMO research initiative (www.eaamo.org) a Postdoctoral Fellow at the Harvard School of Engineering and Applied Sciences and a Research Fellow at Input Output Global (www.iog.io) Francisco was a Career Development Fellow in Computer Science at Balliol College at the University of Oxford in Computer Science from the University of Oxford Francisco is a dedicated researcher specializing in applied modeling striving to enhance resource accessibility for underserved communities His work focuses on addressing critical challenges in public health and improving access to financial services with blockchain technology Francisco is committed to fostering interdisciplinary collaborations between academics Francisco's primary goal is to further research into and develop practical AI tools aimed at addressing the needs of Mexico's indigenous populations He plans to achieve this by enhancing the scope of the EAAMO Summer of Science program that he launched in collaboration with multiple Mexican State Research Councils in 2022 This program supports female students from indigenous communities in Mexico as they design and execute research projects aimed at improving their local communities SEAS symposium explores engineering contributions of Indigenous peoples Fellow Francisco Marmolejo-Cossío participated in a recent SEAS Symposium on Indigenous approaches to engineering Subscribe to our email list for the latest news and commentary from the Berkman Klein Center and our community Unless otherwise noted this site and its contents are licensed under a Creative Commons Attribution 3.0 Unported license An Endangered Missing Advisory has been activated for a missing 19-year-old woman from Del City It is unknown what Marmolejo was wearing when she was last seen OHP said Marmolejo can become violent when without medication Residents are asked to not approach her if spotted and general news updates from News 9 delivered right to your inbox This website is using a security service to protect itself from online attacks The action you just performed triggered the security solution There are several actions that could trigger this block including submitting a certain word or phrase You can email the site owner to let them know you were blocked Please include what you were doing when this page came up and the Cloudflare Ray ID found at the bottom of this page Robert Samarron SAN ANTONIO Texas – A preschool teacher accused of roughing up three toddlers not only has been arrested but also lost his job In a written statement sent to KSAT 12 News, a spokesperson for Excelled Montessori Plus said the school fired Alfredo Marmolejo, 29, immediately after learning about the allegations Marmolejo was arrested Tuesday on a charge of injury to a child An arrest affidavit spells out the details of the case The affidavit says Bexar County Sheriff’s deputies were called to the school and reviewed surveillance video which shows Marmolejo roughly handling three different children It said he was seen grabbing a two-year-old girl by her arm The affidavit also details how Marmolejo hit a two-year-old boy in his head causing that boy to fall and hit his head on a table The single charge against Marmolejo stems from the child who hit his head The parents of one child who was not involved in the incident said they were in the dark about what happened Colin and Rachel Odesanya spoke to KSAT 12 News outside the school and said that although they did receive notice about an incident there it left them with more questions than answers “Just a general email saying something happened,” Colin Odesanya said I just really wanted to know what happened.” In an emailed statement sent on behalf of the school Teresa Bernal said parents of the students directly involved were given details She said the school was limited in what it could share with everyone else although a message was sent out schoolwide that the school self-reported the incident immediately to the authorities and is cooperating with the investigation It also said that the highest priorities at Excelled Montessori are the safety wellbeing and trust of students and their families “We have reached out to all our families to reassure them of our commitment to the wellbeing of their children and encouraged them to reach out to us if they have any questions,” the statement said The Bexar County Sheriff’s Office said that while this appears to be a single incident parents who fear their children may have been harmed should contact the department Man arrested for assaulting three children at far Northwest Bexar County preschool, BCSO says Katrina Webber joined KSAT 12 in December 2009 but after living in Gulf Coast states for the past decade It's not unusual to find her singing karaoke or leading a song with her church choir when she's not on-air Home » Gold & Blue Magazine » Recent J.D graduate accepts prestigious federal clerkship ’24) accepted a 2026-2027 clerkship with the Hon López is currently a law clerk for the Hon District Judge in the Western District of Texas She will complete two years with Briones before starting her new clerkship in Fall 2026. “Ms. López’s success in securing these prestigious clerkships is a testament to her exceptional abilities,” said St. Mary’s Law Dean Patricia Roberts, J.D. “We are immensely proud of Berenis and confident that her achievements will inspire future generations of St Her clerkships with Judge Briones and Judge Ramirez showcase the exceptional opportunities available to St Mary’s Law students and highlight the value of our Judicial Clerkship Mentorship Program.” López benefited from her participation in the St. Mary’s Law’s Judicial Clerkship Mentorship Program. In 2022, the Hon. Marina Garcia Marmolejo (M.A./J.D. ’96) approached Roberts about growing clerkship opportunities from her court The program offers a select number of students the chance to work closely with the judge and her clerks to prepare for a post-grad judicial clerkship Mary’s Law in the federal law clerk community.” “In addition to receiving significant help with the clerkship application process, students in the program enroll in two advanced legal writing seminars in which they perform duties similar to those of actual judicial clerks,” Assistant Dean for Career Strategy Robin Thorner, J.D., said “This unique hands-on exposure to the work of a judicial law clerk is a tremendous advantage for our students.” Mary’s Law graduates and students have secured judicial clerkships Mary’s Judicial Clerkship Mentorship Program obtained subsequent clerkships following their initial terms a Jurist in Residence at the School of Law stressed the importance of the Judicial Clerkship Mentorship Program especially for first-generation students who receive employment and experience after graduation Garcia Marmolejo said she is “immensely proud of Berenis This achievement is a testament to Berenis’ dedication determination and intellect; it also underscores the value and impact of the Judicial Clerkship Mentorship Program.” López will concentrate on legal research and writing The duties of a clerk also may include conducting legal research proofreading the judge’s orders and opinions communicating with counsel regarding case management and procedural requirements and assisting the judge during courtroom proceedings. Ramirez is the first Latina woman to hold the position She has appellate jurisdiction over district courts in Texas but the judge and her law clerks travel to New Orleans to the Fifth Circuit to hear cases set for oral argument. “It is truly an honor to serve in this role for an esteemed jurist like Judge Ramirez,” López said López is the child of immigrants from Mexico She is a first-generation college graduate she was a member of the nation’s top-ranked St Mary’s National Moot Court Team. Before graduating in May 2024 López was named a 2024 Presidential Award recipient by St The Presidential Award is presented to students who exemplify leadership “A clerkship at the United States Court of Appeals can be a steppingstone to a similar position in Washington, D.C., with a justice at the United States Supreme Court,” said Vincent Johnson, J.D., Professor of Law and the Katherine A. Ryan Distinguished Chair for Global, Comparative and International Law Mary’s has a long history of sending many of its best graduates into positions with excellent state and federal judges.” Wanting to show how religious-affiliated universities partnered with market research group YouGov to see how graduates of Catholic institutions fare against their secular school counterparts Although her dream of attending law school was delayed to better serve her clients after working for Gov Ann Richards and with the Great Springs Project along with other new and revamped degrees in Public Health further the University's mission of teaching holistic health Nadia Abdulbaki has always been fascinated with the medical field and helping others since childhood That's why the Biology major and member of the St Mary's University Honors Program hopes to go into the medical field after graduation Ben Hodge’s growth as an entrepreneur has roots in his family tree that has branched into his second year as a student in the Online MBA for Professionals program at the St Mary’s University Greehey School of Business Texas (KVIA) -- Two Borderland educators won a major teaching award today Clint High School teacher Joshua Marmolejo just received the 2024-25 Texas Milken Educator Award Marmolejo leads the Clint HS P-TECH automotive technology program Marmolejo is Clint ISD's first Milken Educator Award recipient A representative for Milken Educator Awards surprised Marmolejo with his award at a school assembly today known as the "Oscars of Teaching," has established a national network of over 3,000 educators in its 40-year history "Marmolejo is the second and final Texas Milken Educator for the 2024-25 season and 59th recipient overall since the state joined the initiative in 2000," officials say "Nancy Chavira of Johanna O’Donnell Intermediate School in Fabens ISD was recognized with the Award earlier this morning KVIA ABC 7 is committed to providing a forum for civil and constructive conversation Please keep your comments respectful and relevant. You can review our Community Guidelines by clicking here If you would like to share a story idea, please submit it here Terms of Service | Privacy Policy | Community Guidelines | KVIA-TV FCC Public File | FCC Applications | Do Not Sell My Personal Information Wind gusts of 30+ mph along with afternoon and evening blowing dust, could reduce visibility. Weather Maps RadarBorderland teachers honored with prestigious Milken Educator Award and $25,000 cash prizeby Renne Estrada Texas (KFOX14/CBS4) — Two teachers in the Borderland received the Milken Educator Award also known as the "Oscars of teaching" along with a $25,000 cash prize Nancy Chavira and Joshua Marmolejo were the two educators who joined a group of over 3,000 Milken Educators spanning the Awards’ 40-year history Recipients for the award are sought out in their early- to mid-career for what they have achieved and for the promise of what they will accomplish given the resources and opportunities afforded by the Award according to a statement from the Milken Educator Award program Chavira is a fifth-grade math and science lead teacher at Johanna O’Donnell Intermediate School She teaches in a 50-50 classroom where students learn both Spanish and English She is the first-ever recipient in Fabens ISD history and the 58th recipient in the state since Texas joined the Milken Educator Awards in 2000 When Chavira attended a scheduled school assembly featuring Texas Education Agency Deputy Commissioner of Operations Alejandro Delgado she was unaware she was the guest of honor She was presented with the award and the cash prize Marmolejo is well-known on the CHS campus for heading the P-TECH automotive technology program fueled through a partnership with El Paso Community College he is the first-ever Milken Educator Award recipient in Clint ISD history Marmolejo leads the P-TECH automotive technology program at Clint High School A program fueled by a partnership with El Paso Community College “Joshua Marmolejo’s passion for his craft and leadership in the P-TECH automotive technology program are creating a roadmap for students to succeed in school and the workforce,” Foley said Joshua plays an important role in his professional learning community and serves students and families throughout the district We welcome Joshua into the Milken Educator Network where he will have the opportunity to expand his impact and change even more lives “Hailed as the ‘Oscars of Teaching,’ the Milken Educator Awards inspire educators students and entire communities to ‘Celebrate Elevate and Activate’ the K-12 teaching profession and encourage young capable people to pursue teaching as a career More than $75 million in individual financial prizes and more than $145 million have been invested in the Milken Educator Award national network overall,” according to the news release For more information, you can visit the Milken Educator Awards website Sign up to receive the top interesting stories from in and around our community once daily in your inbox What sparked your interest in public health? I led a qualitative study on maternal health in the Dominican Republic which revealed the inadequate quality standards at some public hospitals This experience ignited my interest in building public health strategies to enhance maternal care for women globally It motivated me to pursue a career focused on advancing women’s and maternal health equity What led you to choose Bloomberg School of Public Health andjoin the Department of Population The summer before my senior year of undergrad, I had the opportunity to participate in the Bloomberg School’s Diversity Summer Internship Program which introduced me to the various tracks and concentrations offered by the School and perinatal health concentration within the Department of Population This focus aligned perfectly with my goal of deepening my expertise in maternal and women’s health improvement How did your degree prepare you for current and future roles What aspects of the PFRH degree program did you find most useful My MSPH studies significantly enhanced my critical thinking I frequently rely on the communication and writing skills developed through health policy courses as well as the data analysis expertise gained from biostatistics and epidemiology courses These skills are essential to my work addressing complex U.S I frequently rely on the communication and writing skills I developed through health policy courses as well as the data analysis expertise gained from biostatistics and epidemiology courses. Describe your current position and responsibilities in a way that will inform current and prospective students about career opportunities in your field As a Research Consultant at Advisory Board (a subsidiary of UnitedHealth Group) health care leaders stay current with advancements in healthcare strategy My team specifically supports specialty care hospital leaders in developing long-term strategies for high quality women’s health Our work is guided by a combination of literature reviews and data analysis to deliver evidence-based recommendations How did you build your sense of community during your time as a PFRH student I made a deliberate effort to connect with my classmates by inviting them to coffee or lunch fostering meaningful relationships beyond the classroom I attended school-wide events and joined student groups such as the Latino Public Health Network which allowed me to engage with individuals outside my department These efforts helped me build a diverse and supportive community throughout my studies (which overlapped with the COVID-19 pandemic) What was your favorite thing about living in Baltimore while you were a student where I created some of my favorite memories and watching breathtaking sunsets by the waterfront Do you have any advice for prospective students You will get out of the experience what you put in The skills and work ethic you build in the program will propel you forward in your future career Johns Hopkins University This story is from Texas Monthly’s archives. We have left the text as it was originally published to maintain a clear historical record. Read more here about our archive digitization project spoke these words with a gruff indignancy befitting his almighty stature No elected official in the Rio Grande Valley was more loved and feared He was a man of the people but also a man of the Valley an embodiment of both its sunny disposition and its murky soul and his almost whispery singsong voice smoldered a formidable mystique would show up on the doorstep of his home to turn themselves in He won by the greatest majorities of any elected official in South Texas without ever soliciting campaign funds; instead wanting to give me money.” Wild rumors swirled wherever he kicked up dust Was it true that he had a network of spies on both sides of the border Was it true that his enemies sometimes wound up face down in the Rio Grande Residents of the Valley seemed wholly unbothered They preferred that their sheriffs be larger than life few wished to believe what was being suggested in Laredo’s federal courthouse on July 25 that a man like Brig Marmolejo could be bought by drug-dealing inmates at the Hidalgo County jail It was the kind of declaration one seldom heard from a politician in the bribe-infested Valley He won office in 1976 with law-and-order rhetoric and a record to back it up His pledge to the voters of Hidalgo County was that he would not be only tough but pure: “I will not seek favor with any group and I will enforce the law equally when and where required.” For several years Marmolejo lived up to that pledge when so many others could not the rare Valley official who seemed capable of honesty Brig Marmolejo had become the latest in a series of Valley public figures to be accused of federal crimes had pleaded guilty to selling confidential police information Zapata County sheriff Romeo Ramirez and county clerk Arnoldo Flores pleaded guilty to federal drug-related charges after being arrested in Operation Prickly Pear government sting that culminated in June with the conviction of Zapata County judge Jose Luis Guevara And a few weeks before Marmolejo’s trial who had once been the Hidalgo County district attorney pleaded guilty to one count of money laundering and one count of failing to file an IRS form after it was learned that he had been the “house counsel” for three drug organizatons including the massive Donna-based drug-distributing network of Ramon “El Lechero” Martinez On July 27 Marmolejo joined their corrupt ranks The Laredo jury found him guilty of eight counts of bribery For those who remembered the sheriff’s once-unassailable integrity the implications of his demise were unbearably grim—so much so that many of his supporters chose to buy his ludicrous conspiracy theories rather than see the pressures of the Valley for what they were Even those who had a stake in his conviction seemed to fall into brooding after the jury’s decision “The temptations must have been unbelievable,” said IRS special agent John Trevino one of the most active federal agents in the Marmolejo investigation I’m sure the offers kept coming until finally the sheriff couldn’t pass up the chance.” Added former assistant U.S the drug trafficker who bribed Marmolejo: “I have a real feeling of empathy for Brig now.” This was not just one sheriff’s story It was the story of borders easily crossed of temptations ferried and ethics bartered It was the ageless parable of the Rio Grande Valley promising a better life than can be found either in the brush country to the immediate north or in the squalor south of the Rio Grande The Valley’s promise is an empty one for few regions in Texas have a higher unemployment rate Yet this is by no means an unhappy part of the world The four cities in the Rio Grande Valley—Harlingen and McAllen—maintain an amiable bustle despite the wilting South Texas climate The multitude of conolias between Harlingen and McAllen do not share the beaten-and-left-for-dead look of small towns throughout East Texas and along the Gulf Coast The area’s inhabitants seem to carry on their lives in a collective swoon of contentment Whatever deprivation they suffer seems somehow cushioned by the rich alluvial soil and the dazzling oranges But those in the Valley who seek an end to their poverty by any means necessary do not look to the land Perhaps no border in the history of western civilization has spawned as much criminal conduct as the two-hundred-mile stretch of the Rio Grande that separates Zapata Here Confederate shipments of cotton penetrated the Union blockade during the Civil War; here guns found their destination during the Mexican War; and here the tequileros bootlegged their wares during Prohibition more than half of the Mexican marijuana smuggled into the United States enters through Texas’ four southernmost counties which loosely make up the Rio Grande Valley The marijuana is brought across the river on rafts and inner tubes by Mexican “river rats” or on planes that land on private airstrips at a cost of several thousand dollars per load The pot is purchased on credit by local distributors who warehouse their cargo in produce sheds before loading it onto eighteen-wheelers and driving it to a designated ranch farther upstate mark the merchandise they wish to purchase each buyer’s driver pulls up a vehicle buried in the soil or exchanged for Krugerrands where it is laundered in casas de cambio (currency exchange offices) and driven back across the border in armored vehicles as legitimate reported currency untraceable souls who financed the deals at the outset A brief driving tour of the Valley is all it takes to understand why the allure of drug smuggling is at least as powerful as the incentive to see crack in the ghettos The economic deprivation blankets the landscape of the colonias until a sudden incongruity takes shape: a string of lush yards at the southern end of Weslaco a bright red Mercedes whipping into a ramshackle block at the edge of Donna an opulent hacienda looming over the scrubland outside of Hargill “These are extremely poor counties down here,” says Brownsville-based assistant U.S “The shining houses on the hill are owned by drug dealers So who’s a poor local kid going to look up to?” Anyone who has seen the Rio Grande—scarcely more than a puddle in places and many miles of it completely deserted—can imagine how easily contraband crosses the border But the real explanation for the torrential flow of Mexican marijuana into Texas lies with the elected officials and law enforcement agents who live on either side of the river In Mexico policemen pay regular “fees” to authorities for the privilege of having a job and therefore cannot make ends meet unless they find creative ways of making extra money Things are not so cut and dried in the Rio Grande Valley With as little effort as it takes to turn a cheek or wave a hand or speak a few words into a telephone an otherwise honest man can double his month’s salary The goods travel quickly out of the area and become someone else’s problem And if rumors should circulate . . Everyone on the border knows someone involved in the drug trade: a neighbor “Politics for me is easy because I just always try to get along with everybody Except the feds,” says Brig Marmolejo He manages a mirthless laugh as he reposes in the living room of his Edinburg home (He eventually resigned under pressure on November 7.) But he no longer shows up for work and instead spends his days working on his ranch It is Marmolejo’s nature to be braced for the worst He greets strangers with an informal air and sprawls back into his seat as a way of encouraging guests to make themselves at home he is given to filling the silence with comically pessimistic chatter about the weather through small talk that he avoids more-revealing disclosures Over the years I’ve been investigated by everyone from private investigators to the Texas Rangers to the FBI.” Marmolejo sighs as he falls back heavily into his living room sofa and folds his arms just over his prodigious belly “The goddamned Republican federal agents,” he then adds citing in a single epithet the two most loathsome characteristics he can imagine “They sent them down here mostly to keep an eye on me.” Marmolejo’s easy tenor and welcoming expression play into all the stereotypes of the good-natured an image he does not altogether discourage his successes as a sheriff and a politician owe to his complicated nature Described by almost everyone as a “man of the people,” Marmolejo has four years of college education and tends to ridicule those who do not He is quick to befriend yet trusts almost no one having been reared some two hundred miles up the road in the agricultural hamlet of Yorktown He spent his youth working in cotton fields and today asserts with proletarian hostility that “there’s no fat people in Yorktown”—seemingly unaware of his own physique His amiability is both genuine and a front for darker preoccupations Yet it was not so long ago that Brigido Marmolejo seemed to be nothing less than a knight in shining armor Hidalgo County was still very much a dark-skinned region run by white-skinned patricians named Bentsen and Newhouse was the county’s first Hispanic sheriff in nearly a century but he was hopelessly out of his depth and certain to be defeated after his second term Hispanic leaders feared a political backlash They looked to 42-year-old Brig Marmolejo as their savior His had been the career of a peace officer who could do no wrong: at the Edinburg Police Department he became the youngest assistant chief in the Valley; at the Hidalgo County highway patrol which he directed; and at the Texas Alcoholic Beverage Commission where he spent more than five years chasing bootleggers but he had an able tutor in his brother-in-law veteran Valley campaign manager Jim Wilson he was thoroughly “anglocized,” and it pleased the power bosses that Marmolejo had a reputation as a head-cracking law-and-order cop who would keep the common folk in line His campaign slogan said it all: “A Big Man for a Big Job.” Marmolejo defeated Castaneda and three other candidates in the Democratic primary and coasted through the general election The new sheriff’s first term was all that any law-abiding voter could hope for Marmolejo called a press conference to announce that the county’s newly built jail was shoddily constructed and susceptible to escapes and that he would appoint a blue-ribbon panel to fix the mess He demanded (and received) increased funding for more jailers to replace the current inmate-guards “Even though they’re called trusties they can’t be trusted,” When more than two hundred protesting farm workers from Alabama and Georgia marched down to the border to block the Hidalgo-Reynosa bridge The message was clear: Anyone who raised a ruckus in Hidalgo County was bound for jail But the distinguishing event of Brig Marmolejo’s first term occurred in 1978 The county’s most powerful official at that time was district attorney Oscar McInnis Word reached the sheriff from one of his snitches at the Hidalgo County jail that McInnis was trying to get rid of the ex-husband of a woman he was seeing on the sly and that he had solicited the snitch’s help Marmolejo placed a wire on the inmate and sent him off to talk to McInnis The resulting taped conversation featured the Hidalgo County DA offering the inmate money to kill his girlfriend’s ex-husband Marmolejo faced pressure to abandon the investigation but he pressed forward until McInnis was indicted for solicitation of murder and disbarred in 1979 Yet there would linger a fallout over his clash with Oscar McInnis who pleaded guilty to a lesser charge and later had his law license reinstated retaliated with their own investigation of Marmolejo while local FBI agents made known their resentment of the sheriff for being excluded from his investigation which he accused of hiring thieves and drunkards “Brig earned the never-ending enmity of every one of those agencies,” says Jim Wilson they would need the invaluable assistance of Brig Marmolejo Throughout the eighties the Hidalgo County sheriff routed his election-year opponents and rapidly became the most popular politician in the region the role models were not the ones wearing badges One of the most respected men in Brig Marmolejo’s county was an Edinburg High School teacher’s aide named Juan Frank Garcia who had done federal time for smuggling marijuana and no one in Edinburg seemed preoccupied with the source of his wealth When Garcia threw a party for the Valley football coaches that included alcohol the normally timid Edinburg Daily Review was moved to denounce his poor taste “We got a lot of negative response from readers who liked Juan Frank and thought we were picking on him unfairly,” remembers the Review’s editor When Garcia was found shot to death in 1989—apparently by Colmbian cocaine traficantes who felt threatened by his growing distribution network—an embarrassed silence fell over the county Yet Garcia’s presence in the Rio Grande Valley paled in comparison with that of Ramon Martinez known as El Lechero (“the Milkman”) from his early days of delivering marinjuana out of a milk truck got his start in the early seventies transporting marijuana supplied by a former Mexican military officer Martinez’s distribution network had grown to unfathomable proportions Much of his pot was funneled to Oklahoma City owned more than thirty trafficking vehicles and established distribution links at ranches outside Inez and Corsicana El Lechero donated thousands of dollars in hush money to the families of convicted subordinates Using sham transactions he invested in real estate and laundered money through his horse-racing business Unimpeded by Ronald Reagan’s War on Drugs Ramon Martinez was well on his way to becoming the wealthiest man in Hidalgo County He purchased softball uniforms for the Donna police department’s team When the city’s lawn mowers were in disrepair It would have surprised only an outsider that when Martinez was brought into court for a bond hearing in the late eighties numerous prominent citizens—including Mayor Hector Casiano—showed up to lend their support He paraded around with his $16,000 belt buckle and his Anglo girlfriends through his horses and his fighting roosters Like other high-rolling traffickers in Hidalgo County Ramon Martinez waved his ill-gotten wealth like a matador’s cape And the county’s bull of a law enforcement official It is impossible to pinpoint exactly when Marmolejo discarded what his 1976 campaign literature cited as “the public obligation to enforce the law wherever and whenever it is broken.” Likely the sheriff himself does not know when—or Perhaps the reinstatement of former DA Oscar McInnis’ law license convinced him that justice was a fool’s pursuit in the Valley Perhaps his continual exposure to crooked cops and federal agents on the take corroded his own integrity Perhaps he saw the futility of the War on Drugs Perhaps he saw the money others were making and decided that some of it should be coming his way Marmolejo acknowledged that he had been offered bribes on numerous occasions throughout his eighteen-year tenure On one occasion a close friend involved in the drug trade offered the sheriff some money in exchange for some “assistance.” Marmolejo agreed to meet the friend at an Edinburg restaurant “I’ve got that money for you.” and I might as well go ahead and stay broke.” He did not take the money; nor did he arrest the man for attempting to bribe an officer The friend’s overture remained their little secret Yet close observers could see that something about the sheriff had changed “He became more lax in law enforcement,” says one of his former deputies “as if he wasn’t interested.” Though Marmolejo often claimed credit for drug busts that were actually made by city cops he and his deputies were seldom active in major federal cases Edinburgh Daily Review editor Gilbert Tagle recalls thinking “Too many guys out there are making big money Where once Brig Marmolejo had stood for law and order for getting tough and not trusting inmates he was now developing a reputation as a softy “Brig never cared who he was seen with,” admits Jim Wilson “He was always friendly to the bad guys they’d buy him a drink and he’d sit and laugh with them and pat them on the back.” Marmolejo would tell people that dopers were voters too They were big-time contributors who often threw pachangas It did not amuse federal agents to walk into a bar and see the sheriff boozing it up with traffickers who loudly bragged about their latest shipments what was Brig Marmolejo doing about the infamous Ramon Martinez “We couldn’t get near the guy,” Marmolejo says today—though Kerr County authorities had arrested El Lechero once and the feds had nailed him three times the Edinburg teacher’s aide linked with Colombian traficantes “He was a good friend of mine,” Marmolejo says without a moment’s hesitation But it was his friends he should have been worrying about In 1992 Marmolejo ran for a fifth term and won “He got lazy and cocky,” says Jim Wilson who did not run his brother-in-law’s campaign that year “He was worn down to a frazzle.” Whenever and however the transformation had begun by 1992 it was complete: The once-pugnacious sheriff now presided over the state’s most drug-infested county with a peculiar disinterest Several of the area’s big-time traffickers would eventually be arrested And since no federal pretrial facility exists in the Valley such prisoners would often wind up at the Hidalgo County jail word spread: If you wanted special treatment such as “contact visits” (when inmates may talk to visitors outside of the visitation room this could be arranged at the Hidalgo County jail The doling out of contact visits consumed an inordinate amount of Marmolejo’s time He personally met with families of inmates and showed himself to be un jefe muy simpático It was an excellent way to curry favor with constituents since many Valley residents had at least one family member tied up in the drug trade But the doling out of contact visits benefited Marmolejo in more tangible ways as well “It was commonly known that if you had the money you could get contact visits,” says a former Hidalgo County deputy But there were also times when I saw people turned away because they didn’t have the right amount of cash The term was ‘political contributions.’ ” another term for this type of conduct: “bribery.” In exchange for money Marmolejo was giving wealthy smugglers preferential treatment in his jail “In effect,” says assistant U.S the man who would ultimately prosecute Marmolejo ‘Bring your drug business to Hidalgo County we’ll make it fun for you in jail.’ ” the sheriff made jail fun for Homero Heraldo Beltran-Aguirre Unlike the other high-dollar traffickers in the Valley He was from Monterrey and kept most of his assets which included a multimillion dollar mansion He did not throw pachangas for politicians or hang out in Edinburg’s Echo hotel bar with other dealers and their defense attorneys Homero Beltran was Ramon Martinez’s chief supplier of marijuana and perhaps the single largest exporter of Mexican pot in the Rio Grande Valley Beltran’s source was said to be the Quintero family suspected in the 1985 murder of Drug Enforcement agent Enrique Camarena From 1984 until 1990 Beltran was the conduit through which tens of thousand of pounds of marijuana from the Quintero farms made its way onto El Lechero’s eighteen-wheelers and across the United States But the Quintero-Beltran-Martinez axis began to crumble in 1989 when El Lechero was arrested and charged in a 92-count federal indictment El Lechero agreed to a “debriefing.” He ratted on every major player in his network In April 1991 the information he supplied led directly to Beltran’s arrest in San Diego Homero Beltran analyzed his prospects and decided he’d better sing He knew other distributors and money launderers in the Valley and federal agents involved in those cases hung on his every word Arrangements were made to transfer Beltran from California to Marmolejo’s Hidalgo County jail Beltran happened to notice that another federal drug prisoner in the jail “he helped me out quite a bit with votes.” The trafficker and the sheriff were friends and the former exploited their relationship by convincing the latter to hire three of Gonzalez’s nephews as jailers who in turn allowed their uncle to have contact visits whenever he wished Gonzalez told Beltran that some of these visits were not merely contact and that he was paying Marmolejo several thousand dollars in return Gonzalez let it be known that he wasn’t the only federal inmate paying for private visits with women in the jail library Freddy Gonzalez would see what he could do—for a fee Thus did the marijuana exporter and the sheriff have breakfast together one fall morning in 1991 The two men hit it off; as Marmolejo later testified “He was a good person to have a conversation with.” From a profit standpoint Beltran proved to be a particularly good conversationalist that morning The deal they ultimately settled on gave Beltran three “private” contact visits every week; in return Marmolejo would be paid $5,000 per month and an additional $1,000 per visit no money would change hands between the two Beltran would use a trusted intermediary who managed his affairs in Monterrey The man who would deliver Beltran’s money Marmolejo has consistently maintained that every gift he received and every deal he entered into with Guardado was done to develop a connection with him who told me he was a policeman in Mexico,” Marmolejo says “He said his brother-in-law was in jail Basically I was just trying to work a relationship with him It’s a hell of a lot better knowing somebody from Monterrey or Reynosa or wherever when the sheriff first laid eyes on him in October 1991 Marmolejo admits that he didn’t attempt to confirm whether the drug trafficker’s brother-in-law was the policeman he claimed to be because anybody has a damn badge.” Yet for reasons unexplained the sheriff courted Guardado from October 1991 until July 1993 as if he were the only cop ever to cross the border the “Mexican policeman” did not assist Marmolejo in a single investigation was that Guardado frequently brought visitors to Beltran and the visits took place in the sheriff’s office they often brought candy and cabrito for the inmate and various deputies So elaborate were these mealtime visits that annoyed Hidalgo County officers began to refer to Beltran’s group as “the picnickers.” the children and the food were not present according to both the Beltrans and Hidalgo County officers these one-on-one meetings took place behind the sheriff’s closed door Beltran was enjoying conjugal visits with a second woman: 25-year-old Mariadalia Huerta-Solis Guardado faithfully transported Huerta-Solis across the border without the knowledge of Beltran’s wife whisking her into the jail on weekends and taking her directly to the sheriff’s office and let the two sweethearts have full run of the place “It was well known that the sheriff let couples have closed-door meetings in his office,” a former deputy says Marmolejo admits that he granted Beltran contact visits in his office but insists that the door was never closed and that he never received money for these favors When at the sheriff’s trial the Beltran camp produced a ledger detailing payments to Marmolejo the sheriff’s supporters scoffed at the ledger’s authenticity Yet it is well known that major drug dealers—including Ramon “El Lechero” Martinez—frequently keep books of their transactions as there is no other way to maintain an accounting of their cash-only business which was maintained by one of his daughters listed the daily expenses incurred by Guardado on his trips to Edinburg: hotels Marmolejo.” These notations correspond precisely with records gathered by the federal government for Beltran’s ledger to be falsified the conspirators to this scheme would have to include his family and the commissary staff at Brig Marmolejo’s jail Homero Beltran had made a mockery of the jailhouse; it had become his brothel Marmolejo had received thousands of dollars and now and again Beltran had sweetened the pot with a few juicy noncash morsels Guardado paid $5,000 for a Marmolejo campaign billboard featuring the slogan “It takes a mighty big man to fill these boots.” Three months later while Beltran and Guardado were visiting with Marmolejo in his office Beltran’s brother-in-law explained that the watches were late Christmas gifts for Marmolejo “I don’t wear a watch,” Marmolejo told Guardado and Beltran When the three other officers asked the sheriff whether they should accept the watches “It’s up to you.” The law says otherwise: Accepting a gift from an inmate is a state misdemeanor for police officers When Guardado offered to give the sheriff a 1989 Trans Am a man of his girth might find things a little tight inside a sports car Beltran’s brother-in-law then offered the TransAm to Marmolejo’s daughter Irma as a wedding gift “It’s up to you,” he said Marmolejo’s daughter took the car in November 1992 she took it to a used-car dealer and traded it in for another car worth $10,000 In early 1993 the relationship between Marmolejo and the Beltran camp went one fatal step further During one of the sheriff’s many chats with the trafficker Beltran happened to mention that there was a produce shed near Edinburg that he would like to sell The shed had been purchased in the name of a Beltran relative as a means of laundering drug money When Beltran said he had never actually seen the shed but knew that it was on Alamo Road “I believe I know right where you’re talking about,” he said “If I find anybody that wants to buy this shed even coming from un jefe muy simpático Beltran dangled an incentive: If Marmolejo helped sell the shed which was figured to run in the neighborhood of $20,000 Thereupon the sheriff set out to work on behalf of the Valley’s biggest marijuana supplier which was being leased by a lemon and lime dealer named Joe Garcia Guardado collected the month’s $3,000 rent from Garcia who was instructed to make the check out to Marmolejo The sheriff would later insist that the checks were made out in his name because Guardado didn’t have a bank account and that he reimbursed Guardado with $3,000 in cash that he happened to have stashed in his office desk Garcia didn’t quite know what to make of the arrangement Marmolejo mentioned that the shed was for sale “If you want to buy the shed,” the sheriff added I’ll put you in touch with him.” The sheriff’s accommodating ways had reached new heights Marmolejo arranged for a contact visit between Garcia and Beltran to facilitate the sale Then he visited an Edinburg appraiser and asked him to assess the value of the shed Marmolejo personally delivered the $450 fee a realtor—though only for a single client The energy that Brig Marmolejo devoted to assisting a convicted drug dealer’s business ventures wasn’t what one expected of a sheriff it was the same sort of energy he used to display in the early days—at the expense of crooks In April 1993 agents received a tip about the sale of a produce shed on Alamo Road produce vendor Joe Garcia acknowledged that he was buying it from a Hidalgo County inmate named Homero Beltran Sheriff Brig Marmolejo was involved in the deal A tingle went up the federal agents’ collective spine The agents wasted no time in interrogating Beltran “We’ve got you on another money-laundering deal Homero,” the inmate was told by assistant U.S “Tell us everything there is to know about this Edinburg produce shed or we’ll file new charges.” Beltran could see right through Surovic’s bluff The trafficker had all the leverage on his side He had been fishing for a modified-sentence deal from the feds for some time Now he could see what they were hungry for Since he didn’t need favors at the Hidalgo County jail anymore it made perfect sense to throw Brig Marmolejo overboard Beltran sent the feds away that first meeting saying he would have to discuss things with his attorney He remained coy during a second meeting with Surovic while at the same time dropping a few tantalizing hints he was apparently given reason to believe that a reduced sentence was imminent Homero Beltran then commenced to sing Marmolejo’s swan song Beltran ordered Guardado to cooperate with the feds who promptly equipped Guardado with a wire and his truck with a video camera he arranged for a meeting with Marmolejo to discuss business While the two men drove aimlessly through Edinburg in Guardado’s truck Guardado handed Marmolejo several thousand dollars which the sheriff had asked for so he could build a pavilion on his ranch for his daughter’s wedding reception Marmolejo stealthily snatched up the money and slid it down his boot the longtime sheriff of Hidalgo County didn’t have to hit up an alleged Mexican cop for money: He had any number of banker friends who frequently approved loans for him without so much as a credit check But this was a special kind of loan—one that as Guardado assured Marmolejo when the two men met again four days later word had reached Marmolejo that Joe Garcia had told federal agents about the produce shed He also knew that Beltran was dealing with federal prosecutors he received a disquieting phone call from a San Antonio associate who had something to tell him in person With chief jailer Salinas in the passenger seat Marmolejo drove north and met the man more or less halfway who had received some information from a federal agent and they’re gonna do whatever it takes to get you,” he said “There’s three million dollars to be made if you take the rap that you did certain favors for Homero Beltran and [a Beltran intermediary] will get a million.” The money was from Beltran who obviously was worried that Marmolejo might prevail in court and thereby jeopardize the trafficker’s reduced-sentence deal The source did his best to convince Marmolejo that the scenario Beltran offered was more reasonable than facing a trial “You can go to Mexico,” he said “or stay here and get the charges reduced to something else.” It was a vintage Rio Grande Valley deal: Take the money leave things in the hands of powerful friends and wait patiently while justice warps like plastic under the South Texas sun Oscar McInnis would have approved of the deal—which was the best offer Homero Beltran had ever sent Brig Marmolejo’s way The sheriff had long since abandoned his integrity but his swagger had not altogether deserted him in a voice that must have sounded about eighteen years younger a federal grand jury in McAllen indicted the sheriff on eight counts of racketeering A warrant was issued for the arrest of Marmolejo In a raid of Marmolejo’s property federal agent failed to find the $147,000 in bribes that the sheriff had allegedly received from Beltran over the last two years discover farm equipment on the sheriff’s ranch that belonged to Eleazar Morin a drug trafficker who had been in custody at the Hidalgo County jail When the secret federal audit of Marmolejo’s many real estate properties leaked out even the most jaded Valley insiders had to chuckle with amazement The fair market value of his properties totaled more than $900,000 That wasn’t a bad pile for a man of the people—particularly one making an annual salary of $46,305 was a dramatic and melancholy spectacle of justice deflowered Because the case amounted to a primer on how things work in the Valley every witness’ motives were inherently suspect A host of information was deemed inadmissible by U.S district judge George Kazen and kept from the jury the twelve Laredo jurors knew enough to know when to smirk at the testimony and so they did: at Beltran’s insistence that the prosecution had offered him no deals; at Guardado’s unctuous assurance that his drug-dealing broth-in-law had urged him simply “to tell the truth”; and at Marmolejo’s boast that “no one can bribe me.” along with several federal agents and members of Marmolejo’s immediate family watched with growing numbness as the body of evidence accumulated attorney Surovic had built a classic Al Capone–style racketeering case: no drugs defense attorney Tony Canales brought forth almost no witnesses instead urging the jurors to weigh the respected sheriff’s word against that of a dope dealer This rather unenergetic strategy incensed Marmolejo’s supporters who conjectured ridiculously that Canales had been pressured by IRS agents to roll over whatever hope the defense harbored seemed to fade the moment Brig Marmolejo took the stand The sheriff’s voice was smooth and impassive with its usual quality of easygoing fatalism But his testimony was a tangle of contradictions and nonsense you have to.” “I have a policy that I do favors but I don’t like to ask for favors.” “Since Juan Antonio Guardado always said just call me,’ I didn’t see anything wrong with borrowing from him but I needed the money . . .” It was as if the Hidalgo County sheriff had become the official mouthpiece of Rio Grande Valley double-talk the clearer it became that regardless of what Brig Marmolejo had or had not done his word could be trusted no more than that of the felon Homero Beltran Canales did everything but drop to his knees as he beseeched the jurors: “Was the sheriff a fool I think we all agree. . . . But is there anybody in this room who hasn’t befriended someone who goes sour Is there anyone in this room who hasn’t been had “Guardado pushed and pushed,” he moaned And a gift isn’t a part of the indictment.” Again Marmolejo’s attorney said it: “My client was a fool.” Ironically attorney when Marmolejo brought him the information that Oscar McInnis was soliciting a murder Canales knew the kind of man his client was back then Valley drug smugglers and crooked politicians and was himself far removed from the days when Brig Marmolejo stood for something other than the Valley’s tortured ethics Canales chose not to dredge up the old heroics No character witnesses had been called to bring forth a more gallant era The jury had already heard too much about behavior in the Valley to imagine great things of the rotund man who now sat glassy-eyed and red-faced while his lawyer fought to save him by casting him as a buffoon The jury took all of five hours to decide that Marmolejo was guilty of everything the federal government had accused him of With the Martinez network dismantled in Hidalgo County and federal investigators sweeping through Starr County with Zapata County expunged of its crooked leaders with trafficking being pushed upriver to Webb and El Paso counties it would be comforting to imagine the Valley in a newfound state of purity Yet there are signs that this will never be Two powerful drug-smuggling families have sprung up in Weslaco two of them identified as the corpses of Mexican reporters were discovered this past summer in the Hidalgo County town of Monte Alto the number of border trucking companies has increased by more than 700 percent while the opportunities for currency laundering in the free-trade zone have become seemingly limitless And even as new players in the drug trade spring up daily that justice does not come easily in the Rio Grande Valley The information provided by Ramon “El Lechero” Martinez and Homero Beltran all but ensures that the two Hidalgo County traffickers will someday be back on the streets They will return to find a new sheriff; and they will doubtless remember that once upon a time “There’s more corruption in South Texas than Brig Marmolejo,” says a veteran attorney and lifelong Valley resident “And Brig’s not necessarily the most corrupt among them.” a middle-aged Hispanic gentleman sat outdoors with his visitor sipping Mexican beer and talking about his old friend Brig Marmolejo while the sun retreated from view His ranch property was cactus-snaggled and thoroughly unmanicured Ribs and sweetbreads roasted on the barbecue nearby Nothing stirred except for the crackling of the fire who had once been a prominent elected official in Hidalgo County but for the most part he conversed in a tone of airy unflappability as if all was inevitable and all would be put right in the future It was a melody peculiar to this part of the world—and I had a pachanga for Brig here,” the man was saying “It was almost like not acknowledging that someone had died.” He changed his tone “The point was not to judge,” he said you don’t turn your back on someone who was weak ‘We don’t know enough about what really happened The gentleman chewed thoughtfully on a rib for a while under the Valley’s darkening glow he said softly in that voice of almost hypnotic fatalism: “The only purpose was to say ‘We’re still the same.’ ” We deliver! Get curated industry news straight to your inbox. Subscribe to Adweek newsletters “Unfinished Legacies,” uses AI and deepfake technology to re-create the images voices and likenesses of three victims of the epidemic who died after mistakenly taking pills laced with fentanyl: Subscribe now for unlimited access to exclusive insider reporting I want to receive emails from Adweek about products services and events that they feel may be of interest to me I want to receive emails from Adweek on behalf of carefully-selected third party partners about products By submitting your information you agree to Adweek's Terms of Use and Privacy Policy and you will begin receiving our newsletters I understand by creating an account, I agree to Adweek’s Terms of Use and Privacy Policy and that I may review and update my marketing preferences at any time A code has been sent to your email address By subscribing you agree to Adweek’s Terms of Use and Privacy Policy Automatic Renewal and Cancellation By subscribing, you agree to the terms of sale, including the Cancellation and Refund Policy You authorize Adweek.com to charge your credit/debit card at the annual subscription price at the amount noted above now and in time to automatically renew your Adweek.com subscription every year before the start of each new 12-month term unless you tell us to stop. Renewal rates are subject to change at any time with or without notice your service will continue without interruption The cancellation goes into effect at the start of the following billing cycle contact customer service at 844.674.8161 (U.S.) or 845.267.3007 (Outside U.S.) David Cohen is a freelance writer for ADWEEK Adweek is the leading source of news and insight serving the brand marketing ecosystem SAN ANTONIO - A preschool teacher is behind bars after he says his "emotions got the best of him." a teacher at the Excelled Montessori Preschool was arrested after allegedly being caught on video physically assaulting three two-year-old children who attend the school Bexar County Sheriff's Deputies arrived at the school on Tuesday after receiving reports of the abuse Authorities say that when Child Safe Investigators arrived at the scene they were presented with surveillance camera video taken from inside the classroom Deputies say the video captured shows Marmolejo grabbing a child by the arm pointing his finger in the child's face before picking the child up and aggressively slamming the child into a chair Marmolejo can then be seen slapping a different two-year-old in the back of the head with his hand The suspect then walked toward the back of the classroom The video ends with Marmolejo yelling at the first child without checking on the two children he had just assaulted before picking up items and kicking other items throughout the room Marmolejo told investigators that he "did not think what happened in the classroom was that bad," and that "his emotions got the best of him.” Deputies with the Sheriff's Top Offenders Program (STOP) arrested Marmolejo and took him to the Bexar County Jail He is currently charged with three counts of Injury To A Child Anyone with any additional information is asked to call (210) 335-6000 Home › Program Recipients › Eduardo Marmolejo Illinois • Line of Duty Death: December 17 Chicago Police Officers Eduardo Marmolejo and Conrad Gary were struck and killed by a South Shore Line commuter train on December 17 while pursuing a suspect across the tracks Marmolejo had been on the force with the 5th District for about 2.5 years Tunnel to Towers Foundation is a 501(c)(3) nonprofit registered in the US under EIN: 02-0554654 The donation platform that supports this campaign is currently undergoing maintenance. Please try again later, or make a general donation to the Tunnel to Towers Foundation now Learn how to describe the purpose of the image (opens in a new tab) Leave empty if the image is purely decorative 2024 – 28th December,2024GR gallery255 Bowery(between Houston & Stanton)New YorkNY “Enigmatic Glance“ delves into the neutral expression, reinterpreting it as a subtle form of emotional communication. The expressionless face, often the most natural and frequent in everyday life, is presented in this exhibition as a canvas capable of conveying a wide range of meanings. From this perspective, neutrality is not emptiness but a state rich with potential interpretation. The artists highlight the inherent beauty of stillness and invite curiosity about the emotions or stories that might lie beneath. Through their works, the exhibition reveals the delicate boundary between emotion and calm, offering a glimpse into the hidden rhythms and mysteries of facial expressions. Tania Marmolejo captures ambiguous female facial expressions characterized by large eyes, aiming to evoke emotional empathy in viewers. Her work invites audiences to connect deeply with the feelings conveyed. Masumi Yamamoto draws children adorned in ethnic clothing, infusing her illustrations with a mystifying beauty that explores the originality found across diverse cultures. Through her art, she navigates the enigmatic space between the world and her own experiences. TRNZ seeks to blur the boundaries between two realms by presenting familiar objects in unfamiliar contexts. His work transforms everyday situations through unique gestures, highlighting the distinctiveness of ordinary items. Tatsuhito Horikoshi focuses on melancholic expressions drawn from memory and imagination, creating poignant portraits of boys and girls that resonate with feelings of nostalgia and reflection. His work, influenced by manga storytelling techniques, captures the subtle nuances of emotional experiences within the context of societal expectations. Art Plugged is a contemporary platform inspired by a relationship with the broader arts communities. We provide our audience with curated insight into the world of art, from exhibitions to artist interviews and more. Art Plugged is a contemporary platform inspired by a relationship with the broader arts communities. We provide our audience with curated insight into the world of art, from exhibitions to artist interviews and more. We strive to showcase all mediums of art from all corners of the globe, so we have you covered whether you’re a casual art lover or a serious collector. Volume 5 - 2019 | https://doi.org/10.3389/fams.2019.00043 The classic coefficient of variation (CV) is the ratio of the standard deviation to the mean and can be used to compare normally distributed data with respect to their variability this measure has been widely used in many fields the CV is used to evaluate demographic heterogeneity and social aggregates such as race Data of this nature are usually not normally distributed and the distributional characteristics can vary widely more accurate and robust estimator variations of the classic CV are needed to give a more realistic picture of the behavior of collected data we empirically evaluate five measures of relative variability of finite sample sizes via Monte Carlo simulations Our purpose is to give an insight into the behavior of these estimators as their performance has not previously been systematically investigated To represent different behaviors of the data we considered some statistical distributions—which are frequently used to model data across various research fields we consider parameters of these distributions that lead to a similar range of values for the CV Our results indicate that CV estimators based on robust statistics of scale and location are more accurate and give the highest measure of efficiency we study the stability of a robust CV estimator in psychological and genetic data and compare the results with the traditional CV The coefficient of variation (CV) is a standardized, dimensionless measure of dispersion relative to a data set's average [1]. It enables the comparison of several datasets [2] with different units of measurement [3, p. 84]. Karl Pearson was likely one of the first researchers to propose this measure of relative statistical dispersion [4 “In dealing with the comparative variation of men and women (or very often of the two sexes of any animal) we have constantly to bear in mind that relative size influences not only the means but the deviations from the means idle to compare the variation of the larger male organ directly with the variation of the smaller female organ […] we may take as a measure of variation the ratio of standard deviation to mean the standard deviation being treated as the total variation in the mean it does not follow because we have defined in this manner our “coefficient of variation,” that is coefficient is really a significant quantity in the comparison […]; it may be only a convenient mathematical expression but I believe there is evidence to show that it is a more reliable test of “efficiency” […] than absolute variation.” The usual practice is to transform the data so that the transformed response assumes values in the real line and then apply the classic CV Empirical analyses suggested that using the coefficient of variation may lead to incorrect conclusions about the effects of heterogeneity Generally speaking, a sample with a standard deviation larger than the mean will produce CVs > 1 (see [23]) CVs ≈ 0 suggest that there is a high precision of the sample's central tendency; i.e. the variability of the location parameter is very low The CV is also known as the relative standard deviation (RSD) which is the result of multiplying the absolute value of the CV by 100; its interpretation The goal of the current study is to compare the performance of some CV estimators based on the classic approach to the CV estimator Here robust location and scale estimation are used We investigate the estimation problem by varying distributional parameters under several statistical distributions commonly used to model data in several research fields The rest of the paper is organized as follows: section 2 introduces the notations definitions and the CV estimators used in the study; section 3 presents the simulation study and results; section 4 reports a discussion of empirical applications; and section 5 presents some concluding comments Xn be independently and identically distributed observations of a random variable having an unknown cumulative distribution function (CDF) FX(x) the location parameter and σ=Var(X) This work focuses on the population parameter θ = σ/μ Based on the definition of population CV (θ) we note that the CV is a unit-free measure that quantifies the degree of variability relative to the mean It can be used in comparing two distributions of different types with respect to their variability Let θ^ be an estimator of θ the population CV some nonparametric and robust estimators are available to deal with such situations some basic results and notations are put forward This estimator also has a breakdown point of 0 A robust scale estimator of λ is the median absolute deviation about the median (MAD) and its finite, sample breakdown point is approximately 0.5 [61, 64, 65] Estimators of relative variability considered in this study We carried out a Monte Carlo simulation study to analyse the efficiency and robustness of the proposed estimators and to compare them with the Classic CV estimator The main objective is to recommend a good estimator for a population CV via simulations in order to overcome problems of the analytical intractability under a theoretical comparison approach We consider different distributions that represent a wide variety of probabilistic patterns of data and various degrees of non-normality obtained in different applications (i.e. In the context of the design and analysis of Monte Carlo experiments [72–74], we adopted a space-filling design composed of B = 10, 000 multidimensional input points representing sample sizes, distribution model (depending on parameters) and estimators of the population CV. To control the population CV of all distributions and make the properties of the estimators given in Table 1 comparable we adopted a reparameterization of these distributions as a possible observed range of the population CV in the unit interval Although it is outside the scope of this paper to discuss these alternatives it is indeed a discussion needed in future work Probability parametric distributions considered in this study Let E={CQV7,CQV8,CVMAD,CVMnAD} be the set of names that define the alternative estimators to the classic CV estimator given in Table 1 To assess the accuracy of θ^j with respect to the classic estimator of CV of a particular CV estimator is less than that of the classic estimator of the population CV it indicates that a particular CV performs like the population's CV and small values of log10(γ) suggest that a particular CV is more informative that the population's CV We implemented in R [59] the following procedure for the Monte Carlo simulation study: 1. Select a distribution fX from Table 2 where θ is the parameter of interest (population CV) 3. Calculate the estimators for the CV, as shown in Table 1 A similar experimental design to the Monte Carlo scheme was described by Vélez and Correa [79], Marmolejo-Ramos et al. [84] and Vélez et al. [85]. The number of simulation runs B is equal to 10,000. The samples of size n = {10, 25, 50, 100, 200} were generated from each distribution in Table 2 5 approximately 95% of the distribution falls within two standard deviations around the mean To evaluate the efficiency of the estimators given in Table 1 we considered scaled-contaminated normal distributions (variance inflation) CN(μ,σ,α,λ) the following finite mixture model CN was used to simulate data that contain outliers: we considered the level of contamination α = {5% Note that for α = 0% in Equation (6) we obtained the Normal distribution N(μ,σ2) By using combinations of μ and σ a total of 140 simulation scenarios were evaluated in each case The robustness of the estimators given in Table 1 was analyzed by considering symmetric contaminated normal distributions, HNt(μ,σ,α,ν), close to the Normal(μ, σ2) but with heavier-than-normal tails [86] The following contaminated model HNt was used for creating outliers and modeling data sets that exhibit heavy tails: where tν(·) is the Student's t-distribution with ν degrees of freedom [87] Note that for α = 0% in Equation (7) we obtained the Normal distribution N(μ,σ2) α and fixed ν in the contaminated distribution with heavy tails a total of 140 simulation scenarios were evaluated To evaluate the flexibility of the estimators in Table 1, we computed accuracy measures for a determinate choice of distributions from Table 2 We reparameterized the distributions in this table in terms of μ and σ of the normal distribution N(μ,σ2) so that they assumed the θ values close to the baseline distribution We generated samples for the Uniform distribution with the set parameters a=μ-σ3 and b=μ+σ3; thus the mean is μ and the variance is σ2 The values of μ and σ used the same values of the baseline distribution A total of 210 simulation scenarios were studied we impose the restriction 0 < σ2/μ < 1 to the probability of successes p = 1 − (σ2/μ) The number of trials m = ⌊μ/p⌋ make up a total of 3,950 simulation scenarios (combinations of p and m) the function ⌊·⌋ is the integer part of the desired rank The mean and standard deviation are mp and mp(1-p) for p = 0.5 (the value that maximizes the variance) it follows that θ=mp(1-p)/mp=1/m When working with the Poisson(λ) distribution we considered λ = μ2/σ2 The normal distribution can also be used to approximate the Poisson distribution for large values of λ Because values of λ > 20 produce suitable normal approximations We therefore evaluated 105 simulation scenarios obtained by combinations of μ and σ and imposed the condition 0 < λ−1/2 < 1 We set λ = μ/σ and used the same sample sizes mentioned above to compare the different estimators of the CV across a total of 175 simulation scenarios obtained by combinations of μ and σ with the condition 0 < σ/μ < 1 The Shifted Exponential (β, λ) distribution, where β ∈ ℝ is the threshold parameter such that β < x, and the scale parameter λ > 0, is widely used in applied statistics; principally in reliability (see [89–91]) When β = 0 we have the Exponential distribution and in that case we can evaluate “spread” effects The mean and standard deviations are μ = β + λ and σ = λ We used the same sample sizes mentioned above to compare the different estimators of the CV across a total of 915 simulation scenarios obtained using combinations of μ and σ with the condition 0 < σ/μ < 1 In order to evaluate the effect of the sample size and the parameter ν (degrees of freedom) on the estimators of the CV when the data came from a χν2 distribution n was varied as previously described and ν = ⌊2μ2/σ2⌋ By imposing the condition 0 < σ/μ < 1 we carried out a total of 105 simulation scenarios (combinations of μ and σ) we set the parameters α = μ((μ(1−μ)/σ2)−1) and β = (1−μ)((μ(1−μ)/σ2)−1) The mean and variance of the Beta(α β) distribution were given by α/(α + β) and αβ/{(α + β)2(α + β + 1)} respectively This showed that θ = β1/2{α(α + β + 1)}−1/2 the expression for the mode simplifies to 1/2 showing that for α = β > 1 the mode (anti-mode when α = β < 1) is at the center of the distribution and the θ=1/2β+1 is a decreasing function of β By reparameterization of this distribution in terms of μ and σ we have a mean μ and dispersion σ the variance measures the dispersion relative to how far the mean is from 0 or 1 (i.e. so the variance already contains the information in a CV measure: the CV measures the dispersion relative to the odds Combinations of μ and σ and imposing the restrictions α > 0 β > 0 and 0 < θ < 1 resulted in 655 scenarios combinations of μ and σ and imposing the restriction α > 0 β > 0 and 0 < θ < 1 resulted in 4,680 scenarios middle and highly exponent relaxation-time behaviors Under the restriction 0 < θ < 1 a total of 315 simulated scenarios were evaluated The Lognormal distribution has been widely employed in sciences [102]; in particular, it is used to fit empirical reaction times (RTs) and has the status of a baseline distribution in RT research [103] μ=exp(α+β22)∈ℝ and 0 < σ = [exp(β2)−1]exp(2α + β2) are the mean and standard deviation of the variable's natural logarithm combinations of μ and σ resulted in 3,045 scenarios In this section, we describe the simulation studies designed to compare the proposed estimators of the population CV. Figures 1–12 present scatter plots (with jitter) of the accuracy metric log10(γ) vs the true value of the population coefficient of variation θ ∈ (0 and orange points represent the sample sizes 10 The red horizontal line represents the benchmark of equal accuracy between the classic estimator and an alternative estimator Scatter plots comparing the log10(γ) accuracy metric as a function of θ ∈ (0 1) produced for each estimator and sample size under the Normal(μ These plots of the performance of the estimators can be interpreted as follows Higher positive values of log10(γ) indicate that the MSE of a particular CV estimator distribution is higher than that of the classic estimator implies that the estimator is not a good alternative to estimate the CV when the data come from those distributions Negative values of log10(γ) indicate that the proposed estimator is more efficient than the classic estimator and can be considered as an optimal estimator of the population CV Values of log10(γ) that closely approximate to zero indicate that the alternative estimator has a similar behavior to the classic estimator of the population CV Figure 1 shows the results of the log10(γ) accuracy metric for the Normal(μ, σ2) distribution. The plots in Figure 1 indicate an increase of the estimators' MSE as the sample size increases CVMnAD increases (values closer to zero) when θ tends to 1 the performance of the CVMnAD is less sensitive to changes of θ and the log10(γ)CVMAD slightly increases with n The values of log10(γ) for the CQV7 estimator increase as the sample size increases with a minimum value of log10(γ)min = −1.07 when n = 10 and a maximum value of log10(γ)max = 1.65 when n = 200 the CQV7 will produce higher MSE values than the classic CV estimator this finding plays against using the CQV7 estimator instead of the classic CV estimator when the data comes from a N(μ,σ2) distribution A similar result was obtained for the CQV8 estimator (log10(γ)min = 0.01 at n = 10 μ = 0.4 and σ = 0.3; log10(γ)max = 2.46 at n = 10 Our findings suggest that the CVMAD estimator performs better than the CVMnAD our results indicate a more consistent behavior of the former estimator over the latter Close inspection shows that log10(γ)CVMAD ∈ (0.28 2.49) and log10(γ)CVMnAD ∈ (0.13 from which it can be concluded that the MAD-based estimator is a better choice is sufficient to keep track of the robustness of the estimators considered in this study the CVMAD is the most robust and efficient estimator as the tail-weight of the underlying distribution increases The simulation results led us to suggest the use of the CVMAD estimator as a good alternative to the classic CV estimator Median values of log10(γ) accuracy metric across θ produced for each estimator and sample sizes under the scaled-contaminated normal distribution CN(μ,σ,α,λ) 1) produced for each estimator and sample size under the scaled-contaminated normal distribution CN(μ,σ,α,λ) distribution Each horizontal panel represents a level of contamination and λ are defined in section Identification of Simulation Scenarios Table 4 presents the median values of the log10(γ) accuracy metric of the contaminated normal distribution with heavy tails for all sample sizes Visual inspection of this table reveals that for α = 0% and α = 5% of contamination with heavy tails all estimators performed in a similar way to a scaled-contaminated normal distribution each estimator improved when θ tended to the value of 1 CVMAD estimators increased as sample size increased For large values of contamination (α = 10% CQV8 produced the smallest MSE values; however Median values of log10(γ) accuracy metric across θ produced for each estimator and sample sizes under the contaminated normal distribution HNt(μ,σ,α,ν) with heavy tails 1) produced for each estimator and sample size under the contaminated normal distribution HNt(μ,σ,α,ν) with heavy tails and ν are defined in section Identification of Simulation Scenarios Figure 4 displays our findings for the Uniform distribution Note that the values of the log10(γ) accuracy metric decreased when θ increased Compared with the classic estimator of the CV the CVMnAD estimator seems to be a plausible alternative with higher relative efficiency than the baseline the median values of the log10(γ) accuracy metric of the CVs were CQV7 = 1.122 Note there was a slight difference in performance between the CQV7 and CQV8 estimators of the CV which highlights the importance of carefully selecting the type of quantile estimator to be used A close inspection shows that log10(γ)CQV7 ∈ (0.14 and log10(γ)CVMnAD ∈ (0.34 From this it can be concluded that the CVMAD estimator is not necessarily a reasonable choice 1) produced for each estimator and sample size under the reparametrized Uniform distribution Table 5 presents the median values of the log10(γ) accuracy metric of the Binomial distribution This table reveals that the CQV7 and CQV8 estimators of the CV present the highest values among all evaluated estimators Median values of log10(γ) accuracy metric across θ produced for each estimator and sample sizes under the Binomial distribution Figure 5 shows that in the case of the Binomial distribution the CVMnAD estimator was more stable and showed the lowest values of log10(γ) regardless of m and p The plots also reveal that there were some combinations of m and p for which log10(γ) was negative; i.e. the alternative estimators were more efficient than the classic estimator of the CV for this distribution there is not a clear pattern for this behavior for n = 200 there were the following cases: m = 2 = 9 log10(γ)CV7 = −28.27; m = 9 log10(γ)CV8 = −28.27; and m = 90 the following estimation was observed: m = 2 1) produced for each estimator and sample size under the reparametrized Binomial distribution this implies that the median-based estimators evaluated herein perform better than the CQV7 and CQV8 estimators 1) produced for each estimator and sample size under the reparametrized Poisson distribution Median values of log10(γ) accuracy metric across θ produced for each estimator and sample sizes under Poisson distribution Figure 7 reveals our findings for the Exponential distribution Recall that if the number of arrivals in a time interval of length T follows a Poisson process with mean rate λ then the corresponding interarrival time follows an Exponential distribution.Values for T of the Poisson distribution were similar to those observed for the Exponential distribution While the CQV7 and CQV8 estimators performed poorly the CVMAD and CVMnAD estimators had a better performance That is log10(γ)CV7 ∈ (0.52 and log10(γ)CVMnAD ∈ (0.33 This result indicates that the CVMAD estimator is a feasible alternative to the classic estimator of the CV this implies that compared with that of the classic estimator of the CV the MSE of the CVMAD estimator is relatively low 1) produced for each estimator and sample size under the reparametrized Exponential distribution The behavior of the CV estimates for the Shifted-Exponential distribution is represented in Figure 8 The effect of the shift is to produce a large distortion on the MSE leading to a nonlinear form in relation to the values of θ for all estimators in the different sample sizes We observed an inflection point when θ = 0.1 For θ < 0.1 the MSE of the estimators' decreasing and smallest values were obtained when n = 200 When θ > 0.1 the MSE increased in θ with the larger value for the CVMnAD = −0.07 when n = 10 Note that the highest values were obtained when θ ≈ 0 or θ ≈ 1 we obtained a degenerate distribution at 0 when λ ≈ 0 and in the Exponential distribution when β = 0 1) produced for each estimator and sample size under the Shifted-Exponential distribution Figure 9 presents the behavior of the estimators when samples of size n are drawn from a χν2 distribution From the reparameterization of this distribution in terms of μ and σ we have an inverse relationship between θ and ν; that is we observe that the values of the log10(γ) accuracy metric decreases when θ increases (or similarly when ν decreases) and also that a large sample size is associated with a high MSE the CQV7 and CQV8 estimators behave similarly but perform poorly when compared with the CVMAD and CVMnAD estimators Detailed analysis of the results from estimators of the CV revealed that log10(γ)CV7 ∈ (0.54 and log10(γ)CVMnAD ∈ (0.26 our findings indicate that the CVMAD is a good alternative to the classic CV estimator for this particular distribution 1) produced for each estimator and sample size under the reparametrized Chi-square distribution It reveals that the MAD- and MnAD-based estimators are the better choices 1) produced for each estimator and sample size under the reparametrized Beta distribution Median values of log10(γ) accuracy metric across θ produced for each estimator and sample sizes under Beta distribution The results for the Gamma distribution are shown in Figure 11 the results are similar to what was found in the Chi-square distribution as that distribution is a special case of the Gamma distribution Reparameterization of the Gamma distribution in terms of μ and σ resulted in a positive relationship between θ and the log10(γ) accuracy metric; i.e. the accuracy increased when θ increased and a positive relationship existed between sample size and MSE The CQV7 and CQV8 estimators performed equally poorly This result implies that the CQV-based estimators do not perform as well as the classic estimator when n increases Detailed analysis of the results from estimators of the CV revealed that log10(γ)CV7 ∈ (0.52 and log10(γ)CVMnAD ∈ (0.24 our findings indicate that the behavior of estimators is close to the Chi-square case and that the CVMAD can be a good alternative to the classic estimator of the CV for this particular distribution 1) produced for each estimator and sample size under the reparametrized Gamma distribution Figure 12 shows the values of the log10(γ) accuracy metric for the Ex-Gaussian distribution Although in almost every case none of the evaluated estimators showed an equivalent or better performance than the classic estimator of the CV the values of the log10(γ) accuracy metric for the CQV7 and CQV8 estimators 1.64) and log10(γ)CQV8 ∈ (0 we found only one case where the CQV7 and CQV8 estimators are more efficient than the classic estimator when μ = 4.3 n = 10 such that log10(γ)CQV7 = −1.34 and log10(γ)CQV78 = −2.10 respectively the MSE of these two estimators is higher than that of the classic estimator of the CV less feasible alternatives to the classic CV these results indicate how similar the performances of the CQV-based estimators are and that they do not represent a suitable choice to replace the classic CV estimator for this particular distribution 1) produced for each estimator and sample size under the reparametrized Ex-Gaussian distribution Table 8 presents the median values of the log10(γ) accuracy metric of the Ex-Gaussian distribution This table reveals that the MAD estimator is the best choice There are cases where the CVMAD performs slightly better than the classic estimator of the CV we have log10(γ)CVMAD = −0.027 from which it can be concluded that the CVMAD estimator is a reasonable choice Median values of log10(γ) accuracy metric across θ produced for each estimator and sample sizes under the Ex-Gaussian distribution This table reveals that the MnAD estimator is the best choice 1) produced for each estimator and sample size under the Lognormal distribution Median values of log10(γ) accuracy metric across θ produced for each estimator and sample sizes under the Lognormal distribution we demonstrate that the proposed estimators can expand the existing methods for estimating the CV population and thus can enrich the literature some accepted estimators in the literature of the CV population have serious limitations and are less satisfactory in practice because they do not fully incorporate the distributional behavior of the data and thus we conclude that our estimator CVMAD is the most stable and robust in almost all the scenarios considered herein The simulation results in the previous section indicate that The applicability of this measure is examined by re-analysing two real data sets: one data set from the field of psychology and one from the field of genomics Both examples represent cases in which data do not resemble Gaussian shapes and thus preclude the use of the classic Pearson version of the CV Mean and standard error for the simple RT classic CV as a function of age when the (A) classic and (B) CVMAD estimators of the CV are used The plot (C) shows −log10(P) for the structural change test as a function of age the value P indicates the p-value obtained by performing the test Raw and FDR-corrected p-values are shown in red and green Values above the gray horizontal line are significant at 5% was used to further examine the data via the sctest() function: R> sctest(y gender + age, data = dataset type = ‘‘Chow'') they offer an extension of the original analyses by using robust CV estimators the presence of the APOE*E2 allele in patients with PSEN1 E280A AD increases AOO by approximately 8.2 years when no other genetic variants or demographic information are controlled for As shown in Table 10, the smallest median values of the log10(γ) accuracy metric were obtained by the CVMAD and CVMnAD estimators under different sample sizes and each distribution in Table 2 we recommend the MAD-based estimators as alternative estimators of the population CV Median values of log10(γ) accuracy metric across θ produced for each estimator and sample sizes by distribution Probability distributions with positive and negative support can arise in the case of electroencephalogram (EEG) data. Such data could be modeled via the Johnson distribution [129] This distribution is characterized by the parameter vector (μ Preliminary results show that the median-based estimators of the CV give negative values of the log10(γ) accuracy metric when random samples of size n = 50 from a negatively-skewed Johnson distribution with parameters (μ The estimation and suitable interpretations of the CV in asymmetric truncated (e.g. heavy-tailed and finite mixture distributions should be comprehensibly discussed and evaluated in upcoming in silico studies the type 6 quantile estimator seems to provide the most accurate results FM-R and RO proposed the overall idea and contributed equally to the discussion sections The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest RO acknowledges financial support from CNPq/Brazil and FACEPE/Brazil. The authors thank Susan Brunner for proofreading earlier versions of this manuscript and Joan Gladwyn (https://properwords.co.nz) for professionally proofreading the published version FM-R thanks Iryna Losyeva and Alexandra Marmolejo-Losyeva (ratoncito blanco) for their support 1. ^In this paper it is understood that the classic CV is a measure of relative variability (MRV) defined as σ/μ which are parameters of scale and location of normally distributed data We consider variations of this classic MRV—those cases in which the estimators in the numerator and the denominator are other than the classic sample dispersion and mean estimators we tend to refer to the MRVs considered here merely as CVs 2. ^Note the CV has been used in the context of pairwise comparisons and hypothesis testing [28, 47], see also [48 3. ^The breakdown value is the smallest fraction of contamination that can cause the estimator to take on values far from its value on the uncontaminated data [57, 58] 4. ^Suppose θ^ is an estimator for an unknown parameter θ Then the mean squared error (MSE) θ^ is defined as MSE(θ^)=Var(θ^)+[B(θ^)]2 where B(θ^)=E(θ^)-θ and Var(θ^)=E[(θ^-θ)2] are the bias and the variance term of the estimator If two estimators θ^1 and θ^2 of θ are given the estimator θ^2 is said to be superior to θ^1 with respect to the MSE criterion if and only if MSE(θ^1)-MSE(θ^2)≥0 Note that the MSE is a special case of a non-negative function called “loss function” that generally increases as the distance between θ^ and θ increases If θ is real-valued (as is the population CV) the most widely used loss function is defined as L(θ^,θ)=(θ^-θ)2 5. ^The R code used in this simulation study is available in the following repository: https://github.com/Raydonal/Performance-CV 6. ^The distribution of RTs rarely resembles a normal distribution. Instead, positively-skewed distributions, e.g., the Ex-Gaussian, fit RT data more appropriately [99] It is possible that estimating the mean and the SD as parameters of location and scale for RT data can lead to biased results as the mean and SD are optimal for normally-distributed data when dealing with non-normal distributions robust estimators of location and scale are preferred 7. ^Various robust estimators of location and scale are already implemented in the R packages robustbase The DescTools package offers some tools for winsorizing Measurement of relative variation: sociological examples CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Fundamentals of Probability and Statistics for Engineers Google Scholar Mathematical contributions to the theory of evolution CrossRef Full Text | Google Scholar Measurement scales and statistics: a clash of paradigms CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Estimation of the coefficient of variation with minimum risk: a sequential method for minimizing sampling error and study cost Regression analysis of variates observed on (0 CrossRef Full Text | Google Scholar Monitoring the coefficient of variation using a variable sampling interval control chart Comparison of splicing factor 3b inhibitors in human cells Evaluation of respiratory and cardiac motion correction schemes in dual gated PET/CT cardiac imaging Pervasive robustness in biological systems Applicability of the coefficient of variation method for analyzing synaptic plasticity mean interspike interval curves: what do they tell us about the brain CrossRef Full Text | Google Scholar Automatization in second language acquisition: what does the coefficient of variation tell us CrossRef Full Text | Google Scholar Age and sex differences in reaction time in adulthood: results from the United Kingdom health and lifestyle survey Handbook of Parametric and Nonparametric Statistical Procedures Google Scholar On teaching about the coefficient of variation in introductory statistics courses CrossRef Full Text | Google Scholar The implication of the coefficient of centrality for assessing the meaning of the mean On the use of coefficient of variation as a measure of diversity CrossRef Full Text | Google Scholar The use and misuse of the coefficient of variation in organizational demography research CrossRef Full Text | Google Scholar Google Scholar Distribution of the coefficient of variation and the extended t distribution CrossRef Full Text | Google Scholar Note on the application of Fisher's k-statistics Confidence intervals for a normal coefficient of variation CrossRef Full Text | Google Scholar Estimators of coefficient of variation using k samples CrossRef Full Text | Google Scholar Estimator and tests for common coefficients of variation in normal distributions CrossRef Full Text | Google Scholar The distribution of McKay's approximation for the coefficient of variation CrossRef Full Text | Google Scholar On the existence of a normal approximation to the distribution of the ratio of two independent normal random variables CrossRef Full Text | Google Scholar Google Scholar Two new confidence intervals for the coefficient of variation in a normal distribution CrossRef Full Text | Google Scholar Estimator and tests for coefficient of variation in uniform distribution CrossRef Full Text | Google Scholar Performance of coefficient of variation estimators in ranked set sampling Confidence interval estimation for the population coefficient of variation using ranked set sampling: a simulation study Empirical Bayes estimation of coefficient of variation in shifted exponential distributions CrossRef Full Text | Google Scholar Statistical Reasoning in Law and Public Policy Confidence interval for a coefficient of quartile variation CrossRef Full Text | Google Scholar Robustness of the studentized range statistic CrossRef Full Text | Google Scholar Comparisons of treatments after an analysis of variance in ecology CrossRef Full Text | Google Scholar Algorithm AS 190: probabilities and upper quantiles for the studentized range CrossRef Full Text | Google Scholar Understanding Robust and Exploratory Data Analysis Google Scholar CrossRef Full Text | Google Scholar Introduction to Robust Estimation and Hypothesis Testing CrossRef Full Text Google Scholar Maximum Entropy Econometrics: Robust Estimation With Limited Data Google Scholar A test of equality of two normal population means assuming homogeneous coefficients of variation CrossRef Full Text | Google Scholar Testing the population coefficient of variation CrossRef Full Text | Google Scholar Invalidation of parametric and nonparametric statistical tests by concurrent violation of two assumptions CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Confidence intervals for mean absolute deviations CrossRef Full Text | Google Scholar A highly efficient L-estimator for the location parameter of the Cauchy distribution CrossRef Full Text | Google Scholar Approximation Theorems of Mathematical Statistics Wiley series in probability and mathematical statistics: probability and mathematical statistics Google Scholar Quantile probability and statistical data modeling CrossRef Full Text | Google Scholar Robust Statistics: The Approach Based on Influence Functions Google Scholar R: A Language and Environment for Statistical Computing Available online at: http://www.r-project.org PubMed Abstract CrossRef Full Text | Google Scholar Alternatives to the median absolute deviation CrossRef Full Text | Google Scholar Detecting outliers: do not use standard deviation around the mean Statistical Data Analysis Based on the L1-Norm and Related Methods Google Scholar A note on Tyler's modification of the MAD for the Stahel-Donoho estimator CrossRef Full Text | Google Scholar Google Scholar Asymptotic representation of ratio statistics and their mean squared errors CrossRef Full Text | Google Scholar Exact sampling distribution of sample coefficient of variation Google Scholar A uniform approximation to the sampling distribution of the coefficient of variation CrossRef Full Text | Google Scholar Asymptotic Analysis of Measures of Variation Google Scholar Symmetric quantile averages and related estimators CrossRef Full Text | Google Scholar Confidence interval for a coefficient of dispersion in nonnormal distributions Monte Carlo experiments: design and implementation CrossRef Full Text | Google Scholar Design of computer experiments: space filling and beyond CrossRef Full Text | Google Scholar DiceDesign and DiceEval: two R packages for design and analysis of computer experiments CrossRef Full Text | Google Scholar “Measures of performance for evaluation of estimators and filters,” IN Proceedings of SPIE Conference on Signal and Data Processing of Small Targets Google Scholar A study of statistical techniques and performance measures for genetics-based machine learning: accuracy and interpretability Google Scholar A concordance correlation coefficient to evaluate reproducibility PubMed Abstract | CrossRef Full Text Should we think of a different Median estimator CrossRef Full Text Weight of evidence: a review of concept and methods CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Cambridge: Cambridge University Press (2003) Google Scholar Automatic outlier detection of discordant outliers using Ueda's method CrossRef Full Text A new approach to the Box-Cox transformation CrossRef Full Text | Google Scholar Heavy-tailed distributions: properties and tests CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Google Scholar Confidence intervals for the weighted coefficients of variation of two-parameter exponential distributions CrossRef Full Text | Google Scholar Inventory management systems with hazardous items of two-parameter exponential distribution CrossRef Full Text | Google Scholar Approximate fiducial bounds on reliability for the two parameter negative exponential distribution CrossRef Full Text | Google Scholar Significance tests and confidence intervals for coefficients of variation CrossRef Full Text | Google Scholar Understanding variation in disease risk: the elusive concept of frailty CrossRef Full Text | Google Scholar and skew in the responding of ADHD children: a response time distributional approach The basis for choice reaction time slowing in Alzheimer's disease What are the shapes of response time distributions in visual search On the efficacy of procedures to normalize Ex-Gaussian distributions A power comparison of various tests of univariate normality on ex-Gaussian distributions CrossRef Full Text | Google Scholar Evidence from auditory simple reaction times for both change and level detectors Fitting the ex-Gaussian equation to reaction time distributions CrossRef Full Text | Google Scholar Google Scholar Log-normal Distributions across the Sciences: Keys and Clues: On the charms of statistics and how mechanical models resembling gambling machines offer a link to a handy way to characterize log-normal distributions which can provide deeper insight into variability and probability—normal or log-normal: that is the question doi: 10.1641/0006-3568(2001)051[0341:LNDATS]2.0.CO;2 A survey of sampling from contaminated distributions Google Scholar CrossRef Full Text Google Scholar Preliminary Report of a Nationwide Survey of the Physical and Mental Health Attitudes and Lifestyle of a Random Sample of 9,003 British Adults London: Health Promotion Research Trust (1987) The Health and Lifestyle Survey: Seven Years on: A Longitudinal Study of a Nationwide Sample Measuring Changes in Physical and Mental Health Hanover: Dartmouth Publishing Group (1993) Google Scholar Bootstrap methods: another look at the jacknife CrossRef Full Text | Google Scholar Tests of equality between sets of coefficients in two linear regressions CrossRef Full Text | Google Scholar strucchange: an R package for testing for structural change in linear regression models CrossRef Full Text | Google Scholar Controlling the false discovery rate: a practical and powerful approach to multiple testing CrossRef Full Text | Google Scholar A new method for detecting significant p-values with applications to genetic data CrossRef Full Text 2014 Alzheimer's disease facts and figures CrossRef Full Text | Google Scholar About a peculiar disease of the cerebral cortex Epidemiology of Alzheimer's disease: occurrence Epidemiology of Alzheimer's disease CrossRef Full Text | Google Scholar The global prevalence of dementia: a systematic review and meta-analysis APOE*E2 allele delays age of onset in PSEN1 E280A Alzheimer's disease Clinical features of early-onset Alzheimer disease in a large kindred with an E280A presenilin-1 mutation Pre-dementia clinical stages in presenilin 1 E280A familial early-onset Alzheimer's disease: a retrospective cohort study Pooling/bootstrap-based GWAS (pbGWAS) identifies new loci modifying the age of onset in PSEN1 p A new distribution-free quantile estimator CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Robust and efficient estimation of the mode of continuous data: the mode as a viable measure of central tendency CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Systems of frequency curves derived from the first law of Laplace CrossRef Full Text | Google Scholar Citation: Ospina R and Marmolejo-Ramos F (2019) Performance of Some Estimators of Relative Variability Received: 28 August 2019; Accepted: 20 August 2019; Published: 21 August 2019 Copyright © 2019 Ospina and Marmolejo-Ramos. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) distribution or reproduction in other forums is permitted provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited in accordance with accepted academic practice distribution or reproduction is permitted which does not comply with these terms *Correspondence: Fernando Marmolejo-Ramos, ZmVybmFuZG8ubWFybW9sZWpvLXJhbW9zQHVuaXNhLmVkdS5hdQ== Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher. 94% of researchers rate our articles as excellent or goodLearn more about the work of our research integrity team to safeguard the quality of each article we publish. Volume 1 - 2015 | https://doi.org/10.3389/fams.2015.00012 We propose a new methodology to estimate λ the parameter of the Box–Cox transformation as well as an alternative method to determine plausible values for it The former is accomplished by defining a grid of values for λ and further perform a normality test on the λ-transformed data is such that the p-value from the normality test is the highest The set of plausible values is determined using the inverse probability method after plotting the p-values against the values of λ on the grid Our methodology is illustrated with two real-world data sets a simulation study suggests that our method improves the symmetry making it a feasible alternative to the traditional Box–Cox transformation Box and Cox [7] proposed a parametric power transformation technique defined by a single parameter λ, aimed at reducing anomalies in the data [7, 8] and ensuring that the usual assumptions for a linear model hold [9]. This transformation results from modifying the family of power transformations defined by Tukey [10] to account for the discontinuity at λ = 0 [8] Let y=(y1,y2,…,yn)′ be the data on which the Box–Cox transformation is to be applied. Box and Cox [7] defined their transformation as where y(λ) is the λ-transformed data X is the design matrix (possible covariates of interest) β is the set of parameters associated with the λ-transformed data then ϵ ~ N(0, σ2). Note that the transformation in Equation (1) is only valid for yi > 0, i = 1, 2, …, n, and modifications have to be made when negative observations are present [7–9, 30] λ has traditionally been estimated via the MLE method Our method is characterized by pairing the value of λ with the associated p-value of a desired normality test performed on the λ-transformed data A key characteristic of our method is that the selected λ is that paired with the largest p-value given by the desired normality test Let {λ}k be a sequence of k plausible values of λ (unknown) such that Here, λL and λU are, respectively, the lower and upper bounds of that sequence containing a (small) number of λs. This is justified by the “fix one, or possibly a small number of λs and go ahead with the detailed estimation” strategy presented in Box and Cox [7] Using the values of λ in Equation (4) our search strategy involves the following steps: Apply Equation (1) to y with λ = λj Since λ* is such that the p-value for the normality test is the highest for all possible values of λ ∈ (λL, λU) when a specific normality test is applied, Equation (3) follows. However, it could also be the case that, for the same value of λ*, other normality tests reject the null hypothesis (see Figure 2 In addition to the estimation of λ* our approach also allows the possibility of calculating a confidence interval (CI) for this parameter using the inverse probability method Consider Figure 1 and let α ∈ (0 The lower and upper limits 100 × (1−α) CI are given by These two expressions are derived as follows we estimate λ* by finding the jth index such that the p-value of the normality test is the highest and further partition (Equation 4) as A ∪ B with A={λ:λ∈(λL,λ*)} and B={λ:λ∈(λ*,λU)} we calculate the absolute difference between α and p-values of the normality test (associated with each λ value in the set) λL and λU are such that these differences are the minimum Construction of 100 × (1 − α) CI for λ* using the inverse probability method The x-axis corresponds to the plausible values of λ in Equation (4) the y-axis to the p-value of the normality test and the horizontal dotted line the type I error probability It is worth mentioning that the relationship between λ and the p-value of the normality shown here is for illustration purposes only Several normality tests could be used in the steps described above, but their power depends on the sample size and the type of distribution the data resembles [31] An automated approach to circumvent the job of selecting a normality test would be to fit several potential parent distributions to the λj-transformed data find the distribution that gives the best fit (e.g. the lowest Akaike's Information Criterion [AIC]) and then choose the normality test with the highest power against that distribution The drawback of this approach is there are also many candidate distributions that could be fitted to the data and not all of them have been studied in the context of normality tests the power of all normality tests against all probability distributions is unknown A more parsimonious approach would be to use the combined p-value of a selected number of normality tests. Approximately 40 different types of normality tests exist [32] all of which can be coarsely classified in three categories: (i) regression/correlation-, (ii) empirical distribution function-, and (iii) measure of moments-based tests [31] choosing an equal number of tests from each category would enable an educated assessment of the normality of a data set whose likely parent distribution is unknown we used the Shapiro–Wilk (SW) and Shapiro–Francia (SF) tests from category (i) Kolmogorov–Smirnov (KS) and Anderson–Darling (AD) from category (ii) and Doornik–Hansen (DH) and robust Jarque–Bera (rJB) from category (iii) the combined p-value of M independent statistical tests is Φ is the cumulative standard normal distribution and Zm is the quantile of the standard normal distribution associated with p-value of the m statistical test suppose M normality tests are chosen and used such that the pairs (λj combined being the maximum of the combined p-values we illustrate the usefulness of our method by estimating λ* for two real data sets and compare the power of our method with that of the traditional MLE approach via a statistical simulation we provide an example demonstrating how well our method works in the case of a linear regression model Figure 2. Published data sets in which the Box–Cox transformations was used. (A–C) correspond to the data (n = 13) analyzed in Salowitz et al. [37]; (D–F) to a set of observations (n = 917) analyzed in Marzouki et al. [38] (A,D) correspond to the real data; (B,E) to the p-values of six normality tests and the Stouffer method as a function of λ; and (C,F) to the λcombined*-transformed data sets The horizontal line represents α = 0.05 and the red vertical line the value of λ used by the researchers Measures of skewness and kurtosis are also shown along with the p-values of the Kolmogorov–Smirnov (PKS test) and Shapiro–Wilk (PSW test) normality tests this value of λ would not lead to a successful normalization A similar result is obtained only when the SW normality test and the combined p-value approach are used with our method. However, the value λDH*=-1.3703 (95%CI = [−1.469, −1.272]) transforms the data such that most normality tests, but SW, judge it as normal (Figure 2F shows the data transformed with this λ) methods of combining p-values are helpful in gauging a p-value that would represent the normality of a specific data set as evaluated by a set of selected normality tests Although our method can rely on one normality test alone is a sound and safe way to estimate λ Suppose an outcome y and a predictor x are measured and that it is of interest to apply a Box–Cox transformation when fitting the linear model we simulate n = 50 observations with x uniformly distributed on the interval (1 10) and ϵ from an exponential distribution with parameter μ = 1 it is clear that the error term does not meet the usual regression assumptions We estimate the parameter of the Box–Cox transformation using both MLE and our p-value-based approach using the Stouffer's method. In the first case, the boxcox function in Venables and Ripley [40] was used whereas for our method a function in R was programmed this function applies the Box–Cox transformation on y applies a normality test and finally extracts the p-value of such a test the value of λ* will be that for which the p-value of the normality test is the highest The combined p-value approach was also utilized to find the value of λ MLE and that provided by our p-value-based approach (see §3.3 λ^combinedMLE=0.282 and pcombinedMLE=0.0755 whereas λcombined*=0.4514 and pcombined = 0.7297 using our approach a comparison of the symmetry (γ) and kurtosis (β) coefficients for the residuals of model (Equation 11) after transforming y using the values found with each method revealed the residuals produced by our method are more symmetric (γMLE = −0.31 vs γcombined = 0.097) and have a kurtosis closer to that of the normal distribution (βMLE = 2.791 vs and pcombined → 1 suggest that our method is efficient at normalizing the residuals of this regression model Several probability distributions have been proposed in the literature to model positively-skewed data, such as reaction and response time data. Distributions such as the Log-Normal, Weibull, and Wald have been suggested as candidate parent distributions for these types of data. However, the most studied is the Ex-Gaussian (EG) distribution [39] We implemented the following algorithm in R: draw a random sample of size n from the ith distribution Let M be the total number of normality tests Apply Equation (1) on y and estimate λ^MLE apply the M normality tests and determine p1MLE,p2MLE,…,pMMLE combine these p-values using Equation (8) to obtain pcombinedMLE for each value of λj in the sequence λL combine the p-values of the M normality tests to obtain pj λ^MLE,p1MLE,p2MLE,…,pMMLE,pcombinedMLE the rejection rate (RR) of the M normality tests and the combined p-values were calculated for each method (MLE and p-value-based) where pb is the p-value of the test under evaluation in the bth iteration (b = 1 The proportion of normalization can be thus obtained as 1 − RR Figure 3. Proportion of normalization achieved by the (A) MLE and (B) p-value-based approaches for six normality tests in three Ex-Gaussian distributions, and (C) all p-values of the normality tests are combined using Stouffer's method. Abbreviations as in Figure 2 Figure 4. Comparison of probability densities between the MLE and Stouffer–combined p-value approaches to estimate λ. A permutation test of equality with nboot = 1000 samples was utilized (see [48] The x-axis represents the p-value obtained by combining six normality tests using the Stouffer's method; the support of the p-values is the interval (0 The gray band represents the 95% CI for the difference between densities In order to formally determine differences between the methods tested, we compared the probability densities of the combined p-values obtained in the simulation. The results corroborated that our method performed better than the MLE method at normalizing non-normal distributions with low skewness across the sample sizes tested (see the third column in Figure 4) this analysis showed that our method outperformed the MLE when distributions had mild (like EG2) to high (like EG1) levels of skewness and very small sample sizes (e.g. we believe that our methodology provides a new easy-to-use and robust tool to help the data analyst decide which value of λ needs to be used in order to transform any data set so that its normality is guaranteed The results of the simulation study suggest that our method seems to be an improvement in the search of a λ in order to achieve normality The value of λ* can be affected by (i) the number and type of normality tests employed and (ii) the way the combined p-value is computed It is important to recall that all available methods of combining p-values need the p-values given by independent tests Although six normality tests from three different categories were used it is tenable that if more tests are included in the estimation of the combined p-values results can be dramatically different if tests from only one or two categories are included in the computations the estimation of the combined p-values is another aspect that plays a key role We used the Stouffer's method based on its neat properties the statistical properties of λ* and the power of other methods which combine p-values remain to be elucidated Using the Ueda's method for the quantification of outliers affecting distributions transformed via the approach proposed herein is a topic for future investigation The application of our method in situations when the researcher has more than one vector of data to be transformed is important of reaction times measured on the same group of participants over x number of experimental conditions and it is of interest to use a One-Way repeated measures ANOVA By using our method and calculating the 95%CI around λ* we argue that the user has a range of λs to select from in order to achieve the normality of those vectors obtained in each experimental condition it is possible that these CIs overlap with each other and that a common λ* can successfully transform all vectors This same idea would apply to a hypothetical case in which different participants are allocated to different experimental conditions and a One-Way independent measures ANOVA is to be used to analyse the data we believe that the method proposed here is an improvement on the estimation of λ in the Box–Cox transformation and can be used by any researcher dealing with data that needs to meet the normality assumption The authors are particular grateful to Drs Yousri Marzouki and Nicole Salowitz for sharing their data sets and answering questions on specific analytical aspects we needed to clarify Rosie Gronthos and La Patulya for proofreading the document JV was supported by the Eccles Scholarship in Medical Sciences the Fenner Merit Scholarship and the Australian National University (ANU) High Degree Research Scholarship Mauricio Arcos-Burgos from ANU for his unconditional support 1 ^See [37] for details of the computation of the movement times A comparison of the randomization test with the F test when error is skewed Behav Res Methods (2005) 37:426–35 Log-transformation and its implications for data analysis PubMed Abstract | CrossRef Full Text On the efficacy of procedures to normalise Ex-Gaussian distributions Google Scholar Linear mixed model analysis of masked repetition priming CrossRef Full Text | Google Scholar Positively skewed data: revisiting the Box-Cox power transformation Google Scholar Google Scholar The Box-Cox transformation technique: a review CrossRef Full Text | Google Scholar 9. Li P. Box-Cox Transformations: An Overview. Department of Statistics, University of Connecticut (2005). Available online at: http://tinyurl.com/pli2005 The comparative anatomy of transformations CrossRef Full Text | Google Scholar Transformations and influential cases in regression CrossRef Full Text | Google Scholar Transformations in regression: a robust analysis CrossRef Full Text | Google Scholar Regression transformation diagnostics using local influence CrossRef Full Text | Google Scholar A note on Box-Cox transformation diagnostics Google Scholar CrossRef Full Text | Google Scholar On the choice of prior distribution for the Box-Cox transformed linear model CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Google Scholar CrossRef Full Text | Google Scholar On prediction and the power transformation family CrossRef Full Text | Google Scholar Prediction techniques for Box-Cox regression models Google Scholar Diagnostics in transformation and weighted regression CrossRef Full Text | Google Scholar Pitfalls of rescaling regression models with Box-Cox transformations CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Estimation for the Box-Cox transformation model without assuming parametric error distribution CrossRef Full Text | Google Scholar Google Scholar Google Scholar The large-sample behavior of transformations to normality CrossRef Full Text | Google Scholar 29. R Development Core Team. R: A Language and Environment for Statistical Computing. Vienna (2014). Available online at: http://www.R-project.org/ A new family of power transformations to improve normality or symmetry An empirical power comparison of univariate goodness-of-fit tests for normality CrossRef Full Text | Google Scholar Power comparison of some selected normality tests In: Proceedings of the Regional Conference on Statistical Sciences (2010) Combinación de pruebas de hipótesis independientes para proporciones: un estudio de simulación In: Proceedings of the IX International Colloquium in Statistics 34. Vovk V. Combining p-Values via Averaging (2012). Available online at: http://arxiv.org/pdf/1212.4966.pdf Combining probability from independent tests: the weighted Z-method is superior to Fisher's approach Optimally weighted Z-test is a powerful method for combining probabilities in meta-analysis Brief report: visuo-spatial guidance of movement during gesture imitation and mirror drawing in children with autism spectrum disorders PubMed Abstract | CrossRef Full Text Location invariance in masked repetition priming of letters and words A power comparison of various normality tests of univariate normality on Ex-Gaussian distributions CrossRef Full Text | Google Scholar 40. Venables WN, Ripley BD. Modern Applied Statistics with S, 4th Edn. New York, NY: Springer (2002). Available online at: http://www.stats.ox.ac.uk/pub/MASS4 41. Gross J, Ligges U. nortest: Tests for Normality. R package version 1.0-2 (2012). Available online at: http://CRAN.R-project.org/package=nortest 42. Gastwirth JL, Gel YR, Hui WLW, Lyubchich V, Miao W, Noguchi K. lawstat: An R Package for Biostatistics, Public Policy, and Law. R package version 2.4.1. (2013). Available online at: http://CRAN.R-project.org/package=lawstat 43. Wickham P. normwhn.test: Normality and White Noise Testing. R package version 1.0. (2012). Available online at: http://CRAN.R-project.org/package=normwhn.test 44. Fox J, Weisberg S. An R Companion to Applied Regression, 2nd Edn. Thousand Oaks, CA: Sage (2011). Available online at: http://socserv.socsci.mcmaster.ca/jfox/Books/Companion Type I error and the number of iterations in Monte Carlo studies of robustness CrossRef Full Text | Google Scholar CrossRef Full Text | Google Scholar Optimized relative step size random searches CrossRef Full Text | Google Scholar R package sm: Nonparametric Smoothing Methods (version 2.2-5.4) CrossRef Full Text | Google Scholar Powerful and Consistent Analysis of Likert-type Rating scales In: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems Why self-report “Likert” scale data should not be log-transformed PubMed Abstract | CrossRef Full Text An empirical study on the transformation of Likert-scale data to numerical scores Google Scholar A simple method for the detection of outliers CrossRef Full Text Automatic detection of discordant outliers via the Ueda's method CrossRef Full Text | Google Scholar Keywords: maximum likelihood estimation (MLE) Correa JC and Marmolejo-Ramos F (2015) A new approach to the Box–Cox transformation Received: 03 August 2015; Accepted: 12 October 2015; Published: 30 October 2015 Copyright © 2015 Vélez, Correa and Marmolejo-Ramos. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) provided the original author(s) or licensor are credited and that the original publication in this journal is cited *Correspondence: Fernando Marmolejo-Ramos, ZmVybmFuZG8ubWFybW9sZWpvLnJhbW9zQHBzeWNob2xvZ3kuc3Uuc2U= In honor of Latino Heritage Month, the latest installment of our Difference Makers series features Erika Marmolejo, Bears player engagement coordinator. Growing up in Merced, Calif., which is located about an hour northwest of Fresno, Erika Marmolejo played travel softball, basketball and volleyball. "That was my entire life," said the Bears player engagement coordinator. "I pretty much started playing sports as soon as I could walk. For me, it was the competition. I love competition. I hate to lose more than I love to win, so that fuels me." Marmolejo credits her father, Richard, for giving her her drive and determination. An accomplished softball player and team captain at Azusa Pacific University, Marmolejo earned her undergraduate degree in journalism. She aspired to become a sideline reporter, but having been part of sports teams her entire life, she was drawn back to that dynamic with the force of a high-powered magnet. "I had a big interest in sideline reporting," she said. "But then I always found myself back in a team setting, just being around different sports and working in different sports. I think I just really found my footing with a team and kind that player engagement, personal growth type mindset." In her role with the Bears, Marmolejo works closely with players and their families. She provides support and resources in four main areas: continuing education, financial literacy, professional development and personal growth. Marmolejo and others in player engagement throughout the NFL assist players and their entire ecosystems during and after their playing days. As part of the Bears' focus on total wellness, Marmolejo is also involved with providing mental health resources to players, working closely with director of player development and mental skills Mike Wiley Jr. and team clinician Carla Suber. "This role is just such a servant role," Marmolejo said. "I really enjoy that and I think it's because I didn't really have that when I was in college. I remember my coach in college actually took me to get my first checking account. We didn't have that, especially in women's sports. So I like being able to give back and educate." Marmolejo particularly enjoys helping rookies assimilate to life in the NFL. "It means a lot to me to impact these guys' lives and just really help them and help their families and help them transition in," she said. "It brings me joy to see them excel not only as athletes but young men." Marmolejo's first job in professional sports was as an intern for the NBA summer league in Las Vegas. She was hired by the Oakland Raiders in 2016 as an assistant to general manager Reggie McKenzie and worked as a player personnel assistant in the team's scouting department. She also began helping in player engagement in 2018. "She cares with a purpose in mind," said Lamonte Winston, who served as Raiders director of player engagement at the time. "When you're working with players and their families, there has to be care with a purpose and sometimes the purpose is not real comfortable for people because you've got to be able to tell them the truth. She was willing to have the hard conversation; she wasn't afraid to have it." Winston recalls that Marmolejo quickly gained relatability among Raiders players. "That locker room is a small community," Winston said. "Once players have a really good experience, they will communicate to one another like, 'Hey, Erika, you go to her, she's for real, she's going to tell you the truth and she's going to walk with you through the whole experience.'" Marmolejo has developed the same rapport with Bears players since being hired by the team in 2020. "She's someone you can go to and talk to and she'll be honest with you," said receiver Darnell Mooney, a 2020 fifth-round draft pick. "She's good to work with. Anything for you, she's always there." After a brief stint with the Chicago Bulls, Marmolejo was hired by the Bears during the COVID-19 pandemic. It was initially challenging for her to get to know players when all meetings were conducted on a virtual basis and masks were required to be worn at all times throughout Halas Hall. "That was a different experience," Marmolejo said. "It's crazy looking back now at how things were. It's totally different now. I remember back when I was with the Raiders and that's what this looks like now day-to-day. It's probably kind of a culture shock, especially for people who came in the past two years, even players, not having seen how it was pre-COVID." Marmolejo, who is of Mexican descent, is believed to be the only Latina woman working in player engagement for an NFL team. "I'm extremely proud of that," she said. "My culture is everything to me." Marmolejo's grandparents and mother were born in Mexico, and Spanish was her primary language growing up. Her mother, Martha, was previously a Spanish teacher and principal who is now dean of students at a high school in El Dorado, Calif. Her brother, Richard, coaches football at Atwater High School near their hometown of Merced. Marmolejo has spoken to students at different schools, hoping to serve as a role model and inspire them to follow in her footsteps. "A lot of people ask me, 'how do you get into player engagement?'" she said. "Even if you break down a résumé or job description, it's just such a relationship business. There's no blueprint for player engagement; 90 percent of the job is your relationships and working with people. I think that level of relatability and adaptability is a finite skill that I see with my colleagues throughout the league." Marmolejo is grateful for the support she receives from general manager Ryan Poles, coach Matt Eberflus and assistant GM Ian Cunningham. "I'm really thankful to coach Flus and to Ryan Poles and Ian," she said. "Having a dynamic player care support [system] here that they envision is huge for this organization and I'm really proud of what we're building here." As Bears director of team logistics and business processes liaison, Simon Gelan works with virtually every department in Halas Hall and sees himself as "a problem solver and a support system for the players, coaches and staff." No one embodies or understands the mission of the Bears' Diversity, Equity and Inclusion Council better than Ghada Yousef, a 30-year-old graphic designer with the team who was born and raised in the United Arab Emirates. Malcolm Neely made multiple stops on his path to becoming Bears marketing coordinator, gaining valuable experience with the Chicago Sky and Windy City Bulls. But he also made actual stops along the way as a ride-share driver. Drawing upon the lessons he learned from his father, Bears offensive quality control coach Henry Burris remains dedicated to assisting others, whether it's underprivileged kids or NFL players. Applying lessons she learned from her parents, Bears director of partnership activation Melissa McDermott views the clients she works with as more than just business associates and the team she leads at Halas Hall like a second family. Making a difference in the lives of others has always been a passion for Bears manager of youth football and community programs Gustavo Silva. But it didn't become a full-time endeavor until after he earned his college degree at the age of 35 in 2008. Aaron Clark isn't in the same tax bracket as the millionaire athletes he works with as Bears manager of football communications. But that hasn't stopped the 29-year-old from giving back and making a difference. Since becoming Bears Care director in 2006, Marge Hamm has helped the charitable arm of the Bears issue grants totaling nearly $23 million to over 100 qualifying agencies, benefitting countless Chicagoland residents. The NFL's first Indian-American coordinator, Sean Desai is an inspiration to those who share his background. But his ascension from volunteer high school assistant coach to Bears defensive coordinator can be appreciated by those of all colors and creeds. Bears director of events and entertainment Tanesha Wade feels fortunate to have her dream job and is driven to help mentor and teach others who possess similar aspirations. As the Bears' longtime head equipment manager, Tony Medlin is grateful for the opportunity to keep players safe, mentor them and establish lifelong friendships. Marina MarmolejoSocial and Behavioral Sciences (SBS) Connecticut's Emergency Rental Assistance Program hosted by the Department of Housing I currently manage a $400 million rental assistance program for Connecticut residents financially impacted by COVID-19 One of the most valuable parts of my job is ensuring that our fully online application is accessible to communities that do not have equitable access to technology By recognizing the impacts of the digital divide I work to ensure marginalized (yet incredibly resilient) communities have the tools to apply for rental assistance and hopefully regain their housing stability I find it deeply exciting to sit at the forefront of moving all statewide applications to an online platform and creating space to have critical conversations about technological advances and the need to ensure our neighbors have equal opportunities Before I started working for the Department of Housing I created an app-based non-profit called DreamKit that taught job skills to youth experiencing homelessness and paid them for their progress The idea for DreamKit started when I was at YSPH I was given the intellectual playground to explore the beginnings of DreamKit with the support of my mentor I also gained the practical skills I needed to work with communities directly From data management to qualitative research design and stakeholder engagement I learned both the management and data analysis skills needed to start an organization centered in public health and tech-equity but Thesis Day was one of my favorite memories at YSPH I spent over a year collecting and analyzing data around food insecurity and food acquisition strategies within the homeless youth community My thesis presentation was thus a celebration of the new knowledge I was adding to the field It was a beautiful moment to create more space at YSPH to have new conversations about the resiliency and tenacity of youth experiencing homelessness You have the power to redefine public health for yourself and truly society at large - be bold and live your truth I am actively carving out a place for myself in the industry that emphasizes the intersection of technology and social justice I did not see many examples of technology-based interventions that centered the feedback and leadership of folks with lived experience I created an organization and now work statewide to ensure this hyper-specific field of public health is appreciated Yusuf Ransome is the single reason why I had enough confidence to stay in New Haven and incorporate DreamKit as an official nonprofit after graduating from YSPH hosted brainstorming sessions and provided me with the emotional support I needed to continue building DreamKit Danya Keene was not only my advisor and mentor but she helped build my understanding of housing as a public health crisis We have the address for the funeral home & the family on file If you're not happy with your card we'll send a replacement or refund your money Patricia Marmolejo was born in San Antonio, Texas on December 17, 1958, to Gustavo and Aurora Casillas, and went to be with the Lord on June 25, 2023, at the age of 64. Patricia was a loving mother, grandmother, great grandmother, sister, and... View Obituary & Service Information Marmolejo created this Life Tributes page to make it easy to share your memories Patricia Marmolejo was born in San Antonio © 2025 Mission Park Funeral Chapels & Cemeteries Made with love by funeralOne Home » Gold & Blue Magazine » Alumna uses her experience to create judicial clerkships When the Hon. Marina Garcia Marmolejo (M.A./J.D ’96) described her first time in a federal courtroom it was to tell the story of taking her Oath of Allegiance to become a naturalized U.S now a Laredo-based federal district judge for the U.S District Court for the Southern District of Texas program when her paperwork finally came through Although the path to citizenship can take years reaching the end of her journey was one of the happiest moments of her life Now a Jurist in Residence at the School of Law Garcia Marmolejo helps students of her alma mater find opportunities Nominated by former President Barack Obama Garcia Marmolejo was judicially appointed by the U.S Garcia Marmolejo also presides over the swearing-in of new U.S “One of the things that I get to do now as a federal district judge is to preside over those same ceremonies,” Garcia Marmolejo said In 2022, Garcia Marmolejo approached St. Mary’s Law Dean Patricia Roberts, J.D. about growing the clerkship program from her court A clerkship is a postgraduate employment term during which a lawyer She stressed the importance of the program and how it benefits law students with employment and experience after graduation “I didn’t have any lawyers or judges or anybody in my own family who said ‘Why don’t you apply for the federal vacancy?’” Garcia Marmolejo said “Don’t wait for someone to invite you to submit an application or to run for office because that invitation may never come.” ’18) became a federal prosecutor in the U.S District Court for the Western District of Texas and said his writing and oral skills improved drastically after his clerkship with Garcia Marmolejo ‘Why don’t you apply for the federal vacancy?’” “My year as a law clerk in the judge’s chambers really prepared me for that experience,” Mendoza said Assistant Dean of Career Services Robin Thorner, J.D., assists Garcia Marmalejo’s efforts, which also compliment work done by South Texas Distinguished Professor of Law Vincent R. Johnson, J.D., LL.M, LL.D. in finding clerkship opportunities for law students Thorner said that with a federal judge backing the program students will likely get a boost in clerkship opportunities “Judge Garcia Marmolejo is putting her reputation on the line because she is affirmatively reaching out and recommending our students which is a tremendous asset,” Thorner said “The clerkship program is changing lives and putting the St. Mary’s School of Law on the map,” Chapa said. St. Mary's School of Law graduates secure judicial clerkship opportunities through a new clerkship program. Lawrence Scott, Ph.D. (B.A. ’01), found his mission at St. Mary's University to transform generations as a professor and nonprofit leader. Overcoming barriers. Closing deals. Rowing oceans. Whatever phrase you want to use for persisting through all manner of strenuous challenges, Andy Vargas-Hernandez (B.B.A. ’14) has done just that, though his favorite wording might be removing borders. Balancing academics as a Communication Studies major and athletics on the Women's Tennis team has been a challenge, but one Stela Sifuentes tackled with determination. a postdoctoral research in the David Parkes group (Eliza Grinnell/SEAS) Francisco Marmolejo-Cossío was pursuing a Ph.D. in computer science at the University of Oxford when he first discovered Mechanism Design for Social Good (MD4SG) MD4SG draws in researchers from all over the world to assess the needs of underserved communities with a strong emphasis placed on the insights of those communities’ members “The core mission of MD4SG is to improve access to opportunity for historically underserved communities,” said Marmolejo-Cossío “This mission is achieved through interdisciplinary techniques such as computer science but it’s achieved through the actual stakeholders from these communities which connects Spanish-speaking MD4SG members with student researchers from Indigenous communities in Mexico The teams work on projects that use design or computational perspectives to address challenges in the students’ communities then present their posters at MD4SG’s flagship event the ACM Conference on Equity and Access in Algorithms “What’s always resonated with me is the express value MD4SG and EAAMO place on the contributions that people from impacted communities can have on the research process,” Marmolejo-Cossío said “That gives validity to what’s actually happening on the ground and it led me to work on this summer project to highlight the contributions of these students.” a collaboration between MD4SG and the National Network of Councils and State Organizations of Science and Technology in Mexico attracted eight Indigenous female students from one Mexican state in its first year and 20 female students representing 11 Indigenous communities and seven Mexican states this year Most members of this year’s cohort are pursuing or in the process of admission to master’s degrees “What we wanted was a summer research program where our researchers in MD4SG can collaborate on projects with these wonderful scholars from these communities,” Marmolejo-Cossío said “It was a two-way exchange: we could teach them the technological limitations and possibilities of their proposals while also learning from them the constraints they face in their communities.” The SEAS Office of Diversity, Inclusion and Belonging, with the Committee on Ethnicity, Migration, Rights will host a lunch for the Summer of Science participants on Nov which will feature a conversation with David Alvarez-Melis This has been one of the most meaningful projects I’ve ever been on It’s been an eye-opening experience to get to know these students to talk about the realities of their day-to-day lives to learn about the innovations that take place within their communities Marmolejo-Cossío hopes to continue promoting Latinx representation in STEM as he applies for faculty positions in the coming months His research in Parkes’ group focuses on blockchains and cryptocurrency but his activities with MD4SG have also influenced his academic pursuits He recently co-authored a paper with Ariel Procaccia Gordon McKay Professor of Computer Science in which optimized strategies for COVID-19 testing with limited resources were implemented at a research institute in Mexico The paper was the "Exemplary Applied Modeling Track Paper" at the 24th ACM Conference on Economics and Computation “It’s wonderful working with David,” Marmolejo-Cossío said but at the same time he’s been enthusiastic and supportive of my other projects with MD4SG.” SEAS faculty among inductees across academia Patrick Slade uses machine learning techniques to better understand neuromotor control Will compete for share of $517,000 in prize money Be one of the first to try our new activity feed Each January we post a farewell tribute to those members of our community lost in the year just past All are part of our community and contributed to climbing You can read the full tribute to Climbers We Lost in 2023 here Ignacio “Nacho” Javier Lucero was an experienced and well-known mountain guide He died on November 30 just a few meters from the summit of Cerro Marmolejo with two clients Lucero began spending time in the mountains at age 13 He became certified by the Argentine Mountain Guide Association and has since completed more than 60 expeditions he also graduated as a professor of Literature at the National University of Cuyo In 2011, Lucero suffered a heart attack and then a stroke after reaching the summit of Manaslu. It would be years before he was able to regain his fine motor control. He recovered alongside his guide dog, Oro, and in 2019 he managed his highest peak, Gasherbrum II, without supplemental oxygen. Lucero gave a Tedx Talk about the event and his recovery Cerro Marmolejo is a rarely-climbed glaciated volcano in the Andes Mountains on the southern border of Chile it is the world’s southernmost 6,000-meter peak The surrounding landscape is known to be rugged and physically demanding Winds at the summit can reach over 50 miles per hour Lucero has never attempted Cerro Marmolejo but he had been planning the trip for months They proceeded via the normal route from the western saying that it had been a difficult ascent Buchbinder and his clients had decided to turn back due to weather after reaching 3,000 meters was “I’m on my way.” It was his goodbye Augustin Hernandez Marmolejo of Fair Oaks Ranch, TX was born on December 31, 1948 in San Antonio, TX and went to be with the Lord on January 16, 2018 at the age of 69. He is preceded in death by his parents, Carlos C. and Aurora H. Marmolejo... View Obituary & Service Information The family of Augustin Hernandez Marmolejo created this Life Tributes page to make it easy to share your memories Augustin Hernandez Marmolejo of Fair Oaks Ranch