Pub Date : 2019-01-01DOI: 10.1080/2330443X.2019.1619427
Wendy K. Tam Cho, Simon Rubinstein-Salzedo
We thank Chikina, Frieze, and Pegden for their reply to our article. We offer just a short clarification rejoinder. In particular, we would like to be clear that we are not challenging the CFP test as a partisan gerrymandering test. We also do not “cast doubt” on the CFP paper. We have clearly stated that “we take no issues with the mathematics behind the CFP theorem or its proof.” In addition, we do not “prefer” one partisan gerrymandering test over another or advocate a single test. We firmly believe that there is plenty of room for multiple partisan gerrymandering tests. In this space, one test need not be “worse” than another. At the same time, it is indisputable that whether the CFP test would constitute a legal test for partisan gerrymandering is a legal question for the courts to decide. Legal questions cannot be decided by mathematicians. Mathematicians may make proposals, but judges decide whether to accept those proposals. Our point is simply that judges must clearly understand the mathematical concepts (even if not the mathematical details) in order to make a reasoned judgment. However, when the science is unclear, we have only miscommunication, from which no one benefits.
{"title":"Rejoinder to “Understanding our Markov Chain Significance Test”","authors":"Wendy K. Tam Cho, Simon Rubinstein-Salzedo","doi":"10.1080/2330443X.2019.1619427","DOIUrl":"https://doi.org/10.1080/2330443X.2019.1619427","url":null,"abstract":"We thank Chikina, Frieze, and Pegden for their reply to our article. We offer just a short clarification rejoinder. In particular, we would like to be clear that we are not challenging the CFP test as a partisan gerrymandering test. We also do not “cast doubt” on the CFP paper. We have clearly stated that “we take no issues with the mathematics behind the CFP theorem or its proof.” In addition, we do not “prefer” one partisan gerrymandering test over another or advocate a single test. We firmly believe that there is plenty of room for multiple partisan gerrymandering tests. In this space, one test need not be “worse” than another. At the same time, it is indisputable that whether the CFP test would constitute a legal test for partisan gerrymandering is a legal question for the courts to decide. Legal questions cannot be decided by mathematicians. Mathematicians may make proposals, but judges decide whether to accept those proposals. Our point is simply that judges must clearly understand the mathematical concepts (even if not the mathematical details) in order to make a reasoned judgment. However, when the science is unclear, we have only miscommunication, from which no one benefits.","PeriodicalId":43397,"journal":{"name":"Statistics and Public Policy","volume":"6 1","pages":"54 - 54"},"PeriodicalIF":1.6,"publicationDate":"2019-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/2330443X.2019.1619427","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"48280021","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-01-01DOI: 10.1080/2330443X.2018.1564639
S. Millard
Abstract The United States Environmental Protection Agency (USEPA) is responsible for overseeing the cleanup of sites that fall within the jurisdiction of the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA; also known as “Superfund”). This process almost always involves a remedial investigation/feasibility (RI/FS) study, including deriving upper confidence, prediction, and/or tolerance limits based on concentrations from a designated “background” area which are subsequently used to determine whether a remediated site has achieved compliance. Past USEPA guidance states outlying observations in the background data should not be removed based solely on statistical tests, but rather on some scientific or quality assurance basis. However, recent USEPA guidance states “extreme” outliers, based on tests that assume a normal (Gaussian) distribution, should always be removed from background data, and because “extreme” is not defined, USEPA has interpreted this to mean all outliers identified by a test should be removed. This article discusses problems with current USEPA guidance and how it contradicts past guidance, and illustrates USEPA’s current policy via a case study of the Portland, Oregon Harbor Superfund site. Additional materials, including R code, data, and documentation of correspondence are available in the online supplement.
{"title":"EPA is Mandating the Normal Distribution","authors":"S. Millard","doi":"10.1080/2330443X.2018.1564639","DOIUrl":"https://doi.org/10.1080/2330443X.2018.1564639","url":null,"abstract":"Abstract The United States Environmental Protection Agency (USEPA) is responsible for overseeing the cleanup of sites that fall within the jurisdiction of the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA; also known as “Superfund”). This process almost always involves a remedial investigation/feasibility (RI/FS) study, including deriving upper confidence, prediction, and/or tolerance limits based on concentrations from a designated “background” area which are subsequently used to determine whether a remediated site has achieved compliance. Past USEPA guidance states outlying observations in the background data should not be removed based solely on statistical tests, but rather on some scientific or quality assurance basis. However, recent USEPA guidance states “extreme” outliers, based on tests that assume a normal (Gaussian) distribution, should always be removed from background data, and because “extreme” is not defined, USEPA has interpreted this to mean all outliers identified by a test should be removed. This article discusses problems with current USEPA guidance and how it contradicts past guidance, and illustrates USEPA’s current policy via a case study of the Portland, Oregon Harbor Superfund site. Additional materials, including R code, data, and documentation of correspondence are available in the online supplement.","PeriodicalId":43397,"journal":{"name":"Statistics and Public Policy","volume":"6 1","pages":"36 - 43"},"PeriodicalIF":1.6,"publicationDate":"2019-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/2330443X.2018.1564639","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"47615775","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-01-01DOI: 10.1080/2330443x.2019.1666761
P. Harmon, Sarah M McKnight, L. Hildreth, I. Godwin, M. Greenwood
Abstract The Carnegie Classification of Institutions of Higher Education is a commonly used framework for institutional classification that classifies doctoral-granting schools into three groups based on research productivity. Despite its wide use, the Carnegie methodology involves several shortcomings, including a lack of thorough documentation, subjectively placed thresholds between institutions, and a methodology that is not completely reproducible. We describe the methodology of the 2015 and 2018 updates to the classification and propose an alternative method of classification using the same data that relies on structural equation modeling (SEM) of latent factors rather than principal component-based indices of productivity. In contrast to the Carnegie methodology, we use SEM to obtain a single factor score for each school based on latent metrics of research productivity. Classifications are then made using a univariate model-based clustering algorithm as opposed to subjective thresholding, as is done in the Carnegie methodology. Finally, we present a Shiny web application that demonstrates sensitivity of both the Carnegie Classification and SEM-based classification of a selected university and generates a table of peer institutions in line with the stated goals of the Carnegie Classification.
{"title":"An Alternative to the Carnegie Classifications: Identifying Similar Doctoral Institutions With Structural Equation Models and Clustering","authors":"P. Harmon, Sarah M McKnight, L. Hildreth, I. Godwin, M. Greenwood","doi":"10.1080/2330443x.2019.1666761","DOIUrl":"https://doi.org/10.1080/2330443x.2019.1666761","url":null,"abstract":"Abstract The Carnegie Classification of Institutions of Higher Education is a commonly used framework for institutional classification that classifies doctoral-granting schools into three groups based on research productivity. Despite its wide use, the Carnegie methodology involves several shortcomings, including a lack of thorough documentation, subjectively placed thresholds between institutions, and a methodology that is not completely reproducible. We describe the methodology of the 2015 and 2018 updates to the classification and propose an alternative method of classification using the same data that relies on structural equation modeling (SEM) of latent factors rather than principal component-based indices of productivity. In contrast to the Carnegie methodology, we use SEM to obtain a single factor score for each school based on latent metrics of research productivity. Classifications are then made using a univariate model-based clustering algorithm as opposed to subjective thresholding, as is done in the Carnegie methodology. Finally, we present a Shiny web application that demonstrates sensitivity of both the Carnegie Classification and SEM-based classification of a selected university and generates a table of peer institutions in line with the stated goals of the Carnegie Classification.","PeriodicalId":43397,"journal":{"name":"Statistics and Public Policy","volume":"6 1","pages":"87 - 97"},"PeriodicalIF":1.6,"publicationDate":"2019-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/2330443x.2019.1666761","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46951949","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-01-01DOI: 10.1080/2330443X.2019.1637802
Michael D. Collins
Abstract Many sightings of the Ivory-billed Woodpecker (Campephilus principalis) have been reported during the past several decades, but nobody has managed to obtain the clear photo that is regarded as the standard form of evidence for documenting birds. Despite reports of sightings by teams of ornithologists working independently in Arkansas and Florida, doubts cast on the persistence of this iconic species have impeded the establishment of a meaningful conservation program. An analysis of the expected waiting time for obtaining a photo provides insights into why the policy of insisting upon ideal evidence has failed for this species. Concepts in statistics and probability are used to analyze video footage that was obtained during encounters with birds that were identified in the field as Ivory-billed Woodpeckers. One of the videos shows a series of events that are consistent with that species and are believed to be inconsistent with every other species of the region. Another video shows a large bird in flight with the distinctive wing motion of a large woodpecker. Only two large woodpeckers occur in the region, and the flap rate is about ten standard deviations greater than the mean flap rate of the Pileated Woodpecker (Dryocopus pileatus). Supplemental materials for this article are available online.
{"title":"Statistics, Probability, and a Failed Conservation Policy","authors":"Michael D. Collins","doi":"10.1080/2330443X.2019.1637802","DOIUrl":"https://doi.org/10.1080/2330443X.2019.1637802","url":null,"abstract":"Abstract Many sightings of the Ivory-billed Woodpecker (Campephilus principalis) have been reported during the past several decades, but nobody has managed to obtain the clear photo that is regarded as the standard form of evidence for documenting birds. Despite reports of sightings by teams of ornithologists working independently in Arkansas and Florida, doubts cast on the persistence of this iconic species have impeded the establishment of a meaningful conservation program. An analysis of the expected waiting time for obtaining a photo provides insights into why the policy of insisting upon ideal evidence has failed for this species. Concepts in statistics and probability are used to analyze video footage that was obtained during encounters with birds that were identified in the field as Ivory-billed Woodpeckers. One of the videos shows a series of events that are consistent with that species and are believed to be inconsistent with every other species of the region. Another video shows a large bird in flight with the distinctive wing motion of a large woodpecker. Only two large woodpeckers occur in the region, and the flap rate is about ten standard deviations greater than the mean flap rate of the Pileated Woodpecker (Dryocopus pileatus). Supplemental materials for this article are available online.","PeriodicalId":43397,"journal":{"name":"Statistics and Public Policy","volume":"6 1","pages":"67 - 79"},"PeriodicalIF":1.6,"publicationDate":"2019-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/2330443X.2019.1637802","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"60065970","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-10DOI: 10.1080/2330443x.2020.1796400
G. Herschlag, H. Kang, Justin Luo, Christy V. Graves, Sachet Bangia, Robert J. Ravier, Jonathan C. Mattingly
ABSTRACT By comparing a specific redistricting plan to an ensemble of plans, we evaluate whether the plan translates individual votes to election outcomes in an unbiased fashion. Explicitly, we evaluate if a given redistricting plan exhibits extreme statistical properties compared to an ensemble of nonpartisan plans satisfying all legal criteria. Thus, we capture how unbiased redistricting plans interpret individual votes via a state’s geo-political landscape. We generate the ensemble of plans through a Markov chain Monte Carlo algorithm coupled with simulated annealing based on a reference distribution that does not include partisan criteria. Using the ensemble and historical voting data, we create a null hypothesis for various election results, free from partisanship, accounting for the state’s geo-politics. We showcase our methods on two recent congressional districting plans of NC, along with a plan drawn by a bipartisan panel of retired judges. We find the enacted plans are extreme outliers whereas the bipartisan judges’ plan does not give rise to extreme partisan outcomes. Equally important, we illuminate anomalous structures in the plans of interest by developing graphical representations which help identify and understand instances of cracking and packing associated with gerrymandering. These methods were successfully used in recent court cases. Supplementary materials for this article are available online.
{"title":"Quantifying Gerrymandering in North Carolina","authors":"G. Herschlag, H. Kang, Justin Luo, Christy V. Graves, Sachet Bangia, Robert J. Ravier, Jonathan C. Mattingly","doi":"10.1080/2330443x.2020.1796400","DOIUrl":"https://doi.org/10.1080/2330443x.2020.1796400","url":null,"abstract":"ABSTRACT By comparing a specific redistricting plan to an ensemble of plans, we evaluate whether the plan translates individual votes to election outcomes in an unbiased fashion. Explicitly, we evaluate if a given redistricting plan exhibits extreme statistical properties compared to an ensemble of nonpartisan plans satisfying all legal criteria. Thus, we capture how unbiased redistricting plans interpret individual votes via a state’s geo-political landscape. We generate the ensemble of plans through a Markov chain Monte Carlo algorithm coupled with simulated annealing based on a reference distribution that does not include partisan criteria. Using the ensemble and historical voting data, we create a null hypothesis for various election results, free from partisanship, accounting for the state’s geo-politics. We showcase our methods on two recent congressional districting plans of NC, along with a plan drawn by a bipartisan panel of retired judges. We find the enacted plans are extreme outliers whereas the bipartisan judges’ plan does not give rise to extreme partisan outcomes. Equally important, we illuminate anomalous structures in the plans of interest by developing graphical representations which help identify and understand instances of cracking and packing associated with gerrymandering. These methods were successfully used in recent court cases. Supplementary materials for this article are available online.","PeriodicalId":43397,"journal":{"name":"Statistics and Public Policy","volume":"7 1","pages":"30 - 38"},"PeriodicalIF":1.6,"publicationDate":"2018-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/2330443x.2020.1796400","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42236973","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-01DOI: 10.1080/2330443X.2018.1438939
Dongfeng Wu, K. Kafadar, S. Rai
ABSTRACT We develop a probability model for evaluating long-term outcomes due to regular screening that incorporates the effects of prior screening examinations. Previous models assume that individuals have no prior screening examinations at their current ages. Due to current widespread medical emphasis on screening, the consideration of screening histories is essential, particularly in assessing the benefit of future screening examinations given a certain number of previous negative screens. Screening participants are categorized into four mutually exclusive groups: symptom-free-life, no-early-detection, true-early-detection, and overdiagnosis. For each case, we develop models that incorporate a person’s current age, screening history, expected future screening frequency, screening test sensitivity, and other factors, and derive the probabilities of occurrence for the four groups. The probability of overdiagnosis among screen-detected cases is derived and estimated. The model applies to screening for any disease or condition; for concreteness, we focus on female breast cancer and use data from the study conducted by the Health Insurance Plan of Greater New York (HIP) to estimate these probabilities and corresponding credible intervals. The model can provide policy makers with important information regarding ranges of expected lives saved and percentages of true-early-detection and overdiagnosis among the screen-detected cases.
{"title":"Inference of Long-Term Screening Outcomes for Individuals with Screening Histories","authors":"Dongfeng Wu, K. Kafadar, S. Rai","doi":"10.1080/2330443X.2018.1438939","DOIUrl":"https://doi.org/10.1080/2330443X.2018.1438939","url":null,"abstract":"ABSTRACT We develop a probability model for evaluating long-term outcomes due to regular screening that incorporates the effects of prior screening examinations. Previous models assume that individuals have no prior screening examinations at their current ages. Due to current widespread medical emphasis on screening, the consideration of screening histories is essential, particularly in assessing the benefit of future screening examinations given a certain number of previous negative screens. Screening participants are categorized into four mutually exclusive groups: symptom-free-life, no-early-detection, true-early-detection, and overdiagnosis. For each case, we develop models that incorporate a person’s current age, screening history, expected future screening frequency, screening test sensitivity, and other factors, and derive the probabilities of occurrence for the four groups. The probability of overdiagnosis among screen-detected cases is derived and estimated. The model applies to screening for any disease or condition; for concreteness, we focus on female breast cancer and use data from the study conducted by the Health Insurance Plan of Greater New York (HIP) to estimate these probabilities and corresponding credible intervals. The model can provide policy makers with important information regarding ranges of expected lives saved and percentages of true-early-detection and overdiagnosis among the screen-detected cases.","PeriodicalId":43397,"journal":{"name":"Statistics and Public Policy","volume":" ","pages":"1 - 10"},"PeriodicalIF":1.6,"publicationDate":"2018-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/2330443X.2018.1438939","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"48631957","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-01DOI: 10.1080/2330443X.2018.1438938
Michael Kurtz
ABSTRACT This study shows value-added models (VAM) and student growth percentile (SGP) models fundamentally disagree regarding estimated teacher effectiveness when the classroom distribution of test scores conditional on prior achievement is skewed (i.e., when a teacher serves a disproportionate number of high- or low-growth students). While conceptually similar, the two models differ in estimation method which can lead to sizable differences in estimated teacher effects. Moreover, the magnitude of conditional skewness needed to drive VAM and SGP models apart often by three and up to 6 deciles is within the ranges observed in actual data. The same teacher may appear weak using one model and strong with the other. Using a simulation, I evaluate the relationship under controllable conditions. I then verify that the results persist in observed student–teacher data from North Carolina.
{"title":"Value-Added and Student Growth Percentile Models: What Drives Differences in Estimated Classroom Effects?","authors":"Michael Kurtz","doi":"10.1080/2330443X.2018.1438938","DOIUrl":"https://doi.org/10.1080/2330443X.2018.1438938","url":null,"abstract":"ABSTRACT This study shows value-added models (VAM) and student growth percentile (SGP) models fundamentally disagree regarding estimated teacher effectiveness when the classroom distribution of test scores conditional on prior achievement is skewed (i.e., when a teacher serves a disproportionate number of high- or low-growth students). While conceptually similar, the two models differ in estimation method which can lead to sizable differences in estimated teacher effects. Moreover, the magnitude of conditional skewness needed to drive VAM and SGP models apart often by three and up to 6 deciles is within the ranges observed in actual data. The same teacher may appear weak using one model and strong with the other. Using a simulation, I evaluate the relationship under controllable conditions. I then verify that the results persist in observed student–teacher data from North Carolina.","PeriodicalId":43397,"journal":{"name":"Statistics and Public Policy","volume":" ","pages":"1 - 8"},"PeriodicalIF":1.6,"publicationDate":"2018-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/2330443X.2018.1438938","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"47419044","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-01DOI: 10.1080/2330443X.2018.1438940
P. Brantingham, Matthew A. Valasik, G. Mohler
ABSTRACT Racial bias in predictive policing algorithms has been the focus of a number of recent news articles, statements of concern by several national organizations (e.g., the ACLU and NAACP), and simulation-based research. There is reasonable concern that predictive algorithms encourage directed police patrols to target minority communities with discriminatory consequences for minority individuals. However, to date there have been no empirical studies on the bias of predictive algorithms used for police patrol. Here, we test for such biases using arrest data from the Los Angeles predictive policing experiments. We find that there were no significant differences in the proportion of arrests by racial-ethnic group between control and treatment conditions. We find that the total numbers of arrests at the division level declined or remained unchanged during predictive policing deployments. Arrests were numerically higher at the algorithmically predicted locations. When adjusted for the higher overall crime rate at algorithmically predicted locations, however, arrests were lower or unchanged.
{"title":"Does Predictive Policing Lead to Biased Arrests? Results From a Randomized Controlled Trial","authors":"P. Brantingham, Matthew A. Valasik, G. Mohler","doi":"10.1080/2330443X.2018.1438940","DOIUrl":"https://doi.org/10.1080/2330443X.2018.1438940","url":null,"abstract":"ABSTRACT Racial bias in predictive policing algorithms has been the focus of a number of recent news articles, statements of concern by several national organizations (e.g., the ACLU and NAACP), and simulation-based research. There is reasonable concern that predictive algorithms encourage directed police patrols to target minority communities with discriminatory consequences for minority individuals. However, to date there have been no empirical studies on the bias of predictive algorithms used for police patrol. Here, we test for such biases using arrest data from the Los Angeles predictive policing experiments. We find that there were no significant differences in the proportion of arrests by racial-ethnic group between control and treatment conditions. We find that the total numbers of arrests at the division level declined or remained unchanged during predictive policing deployments. Arrests were numerically higher at the algorithmically predicted locations. When adjusted for the higher overall crime rate at algorithmically predicted locations, however, arrests were lower or unchanged.","PeriodicalId":43397,"journal":{"name":"Statistics and Public Policy","volume":"5 1","pages":"1 - 6"},"PeriodicalIF":1.6,"publicationDate":"2018-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/2330443X.2018.1438940","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46587771","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-01DOI: 10.1080/2330443X.2018.1442271
R. Kosar, D. W. Scott
ABSTRACT University ranking is a popular yet controversial endeavor. Most rankings are based on both public data, such as student test scores and retention rates, and proprietary data, such as school reputation as perceived by high school counselors and academic peers. The weights applied to these characteristics to compute the rankings are often determined in a subjective fashion. Of significant importance in the academic field, the Carnegie Classification was developed by the Carnegie Foundation for the Advancement of Teaching. It has been updated approximately every 5 years since 1973, most recently in February 2016. Based on bivariate scores, Carnegie assigns one of three classes (R1/R2/R3) to doctorate-granting universities according to their level of research activity. The Carnegie methodology uses only publicly available data and determines weights via principal component analysis. In this article, we review Carnegie’s stated goals and the extent to which their methodology achieves those goals. In particular, we examine Carnegie’s separation of aggregate and per capita (per tenured/tenure-track faculty member) variables and its use of two separate principal component analyses on each; the resulting bivariate scores are very highly correlated. We propose and evaluate two alternatives and provide a graphical tool for evaluating and comparing the three scenarios.
{"title":"Examining the Carnegie Classification Methodology for Research Universities","authors":"R. Kosar, D. W. Scott","doi":"10.1080/2330443X.2018.1442271","DOIUrl":"https://doi.org/10.1080/2330443X.2018.1442271","url":null,"abstract":"ABSTRACT University ranking is a popular yet controversial endeavor. Most rankings are based on both public data, such as student test scores and retention rates, and proprietary data, such as school reputation as perceived by high school counselors and academic peers. The weights applied to these characteristics to compute the rankings are often determined in a subjective fashion. Of significant importance in the academic field, the Carnegie Classification was developed by the Carnegie Foundation for the Advancement of Teaching. It has been updated approximately every 5 years since 1973, most recently in February 2016. Based on bivariate scores, Carnegie assigns one of three classes (R1/R2/R3) to doctorate-granting universities according to their level of research activity. The Carnegie methodology uses only publicly available data and determines weights via principal component analysis. In this article, we review Carnegie’s stated goals and the extent to which their methodology achieves those goals. In particular, we examine Carnegie’s separation of aggregate and per capita (per tenured/tenure-track faculty member) variables and its use of two separate principal component analyses on each; the resulting bivariate scores are very highly correlated. We propose and evaluate two alternatives and provide a graphical tool for evaluating and comparing the three scenarios.","PeriodicalId":43397,"journal":{"name":"Statistics and Public Policy","volume":"5 1","pages":"1 - 12"},"PeriodicalIF":1.6,"publicationDate":"2018-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/2330443X.2018.1442271","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42101780","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-01DOI: 10.1080/2330443X.2018.1448733
Nathan Sanders, Victor Lei
ABSTRACT While public debate over gun control in the United States has often hinged on individual public mass shooting incidents, legislative action should be informed by knowledge of the long-term evolution of these events. We present a new Bayesian model for the annualized rate of public mass shootings in the United States based on a Gaussian process with a time-varying mean function. While we present specific findings on long- and short-term trends of these shootings in the U.S., our focus is on understanding the role of model design and prior information in policy analysis. Using a Markov chain Monte Carlo inference technique, we explore the posterior consequences of different prior choices and explore correlations between hyperparameters. We demonstrate that the findings about the long-term evolution of the annualized rate of public mass shootings are robust to choices about prior information, while inferences about the timescale and amplitude of short-term variation depend sensitively on the prior. This work addresses the policy implications of implicit and explicit choices of prior information in model design and the utility of full Bayesian inference in evaluating the consequences of those choices.
{"title":"The Role of Prior Information in Inference on the Annualized Rates of Mass Shootings in the United States","authors":"Nathan Sanders, Victor Lei","doi":"10.1080/2330443X.2018.1448733","DOIUrl":"https://doi.org/10.1080/2330443X.2018.1448733","url":null,"abstract":"ABSTRACT While public debate over gun control in the United States has often hinged on individual public mass shooting incidents, legislative action should be informed by knowledge of the long-term evolution of these events. We present a new Bayesian model for the annualized rate of public mass shootings in the United States based on a Gaussian process with a time-varying mean function. While we present specific findings on long- and short-term trends of these shootings in the U.S., our focus is on understanding the role of model design and prior information in policy analysis. Using a Markov chain Monte Carlo inference technique, we explore the posterior consequences of different prior choices and explore correlations between hyperparameters. We demonstrate that the findings about the long-term evolution of the annualized rate of public mass shootings are robust to choices about prior information, while inferences about the timescale and amplitude of short-term variation depend sensitively on the prior. This work addresses the policy implications of implicit and explicit choices of prior information in model design and the utility of full Bayesian inference in evaluating the consequences of those choices.","PeriodicalId":43397,"journal":{"name":"Statistics and Public Policy","volume":"5 1","pages":"1 - 8"},"PeriodicalIF":1.6,"publicationDate":"2018-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/2330443X.2018.1448733","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"45772292","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：