Generalized Linear Models for Aggregated Data

Total Page:16

File Type:pdf, Size:1020Kb

Generalized Linear Models for Aggregated Data Generalized Linear Models for Aggregated Data Avradeep Bhowmik Joydeep Ghosh Oluwasanmi Koyejo University of Texas at Austin University of Texas at Austin Stanford University [email protected] [email protected] [email protected] Abstract 1 Introduction Modern life is highly data driven. Datasets with Databases in domains such as healthcare are records at the individual level are generated every routinely released to the public in aggregated day in large volumes. This creates an opportunity form. Unfortunately, na¨ıve modeling with for researchers and policy-makers to analyze the data aggregated data may significantly diminish and examine individual level inferences. However, in the accuracy of inferences at the individ- many domains, individual records are difficult to ob- ual level. This paper addresses the scenario tain. This particularly true in the healthcare industry where features are provided at the individual where protecting the privacy of patients restricts pub- level, but the target variables are only avail- lic access to much of the sensitive data. Therefore, able as histogram aggregates or order statis- in many cases, multiple Statistical Disclosure Limita- tics. We consider a limiting case of gener- tion (SDL) techniques are applied [9]. Of these, data alized linear modeling when the target vari- aggregation is the most widely used technique [5]. ables are only known up to permutation, and explore how this relates to permutation test- It is common for agencies to report both individual ing; a standard technique for assessing statis- level information for non-sensitive attributes together tical dependency. Based on this relationship, with the aggregated information in the form of sample we propose a simple algorithm to estimate statistics. Care must be taken in the analysis of such the model parameters and individual level in- data, as na¨ıve modeling with aggregated data may sig- ferences via alternating imputation and stan- nificantly diminish the accuracy of inferences at the dard generalized linear model fitting. Our re- individual level. In particular, inferences drawn from sults suggest the effectiveness of the proposed aggregated data may lead to the problem of ecologi- approach when, in the original data, permu- cal fallacy [23], hence the resulting conclusions at the tation testing accurately ascertains the verac- group level may be misleading to researchers and pol- ity of the linear relationship. The framework icy makers interested in individual level inferences. An is extended to general histogram data with example that has been cited [21] is the high correla- larger bins - with order statistics such as the tion between per capita consumption of dietary fat and median as a limiting case. Our experimen- breast cancer in different countries, which may lead to tal results on simulated data and aggregated the incorrect conclusion that dietary fat causes breast healthcare data suggest a diminishing returns cancer [7]. property with respect to the granularity of Aggregated data in the form of histograms and other the histogram - when a linear relationship sample statistics are becoming more and more com- holds in the original data, the targets can be mon. Further, most of the data that is collected relates predicted accurately given relatively coarse to questions for which the respondents have only a few histograms. discrete options from which to select their answer. For example, data available from the Generalized Social Survey (GSS) [1] are often in this form. This paper Appearing in Proceedings of the 18th International Con- addresses the scenario where features are provided at ference on Artificial Intelligence and Statistics (AISTATS) the individual level, but the target variables are only 2015, San Diego, CA, USA. JMLR: W&CP volume 38. available as histogram aggregates or order statistics. Copyright 2015 by the authors. Despite the prevalence of order-statistic and histogram 93 Generalized Linear Models for Aggregated Data aggregated data, to the best of our knowledge, this ments of the vector by the same lowercase letter with problem has not been addressed in the literature. the boldface removed and the index added as a super- script. v> refers to the transpose of the column vector We consider a limiting case of generalized linear mod- v. We denote column partitions using semicolons, that eling when the target variables are only known up to is, M = [X; Y] implies that the columns of the subma- permutation, and explore how this relates to permu- trices X and Y are, in order, the columns of the full tation testing [12]; a standard technique for assessing matrix M. We use k · k to denote the L norm for vec- statistical dependency. Based on this relationship, we 2 tors and Frobenius norm for matrices. The vector v is propose a simple algorithm to estimate the model pa- said to be in increasing order if v(i) ≤ v(j) whenever rameters and individual level inferences via alternat- i ≤ j, and the set of all such vectors in n is denoted ing imputation and standard generalized linear model R with a subscripted downward pointing arrow as n. fitting. Our results suggest the effectiveness of the R# Two vectors v and w are said to be isotonic, v ∼ w, proposed approach when, in the original data, permu- # if v(i) ≥ v(j) if and only if w(i) ≥ w(j) for all i; j. tation testing accurately ascertains the veracity of the linear relationship. The framework is extended to gen- eral histogram data with larger bins - with order statis- 1.1 Preliminaries and Related Work tics such as the median as a limiting case. Our exper- imental results suggest a diminishing returns property Aggregated data is often summarized using a sample - when a linear relationship holds in the original data, statistic, which provides a succinct descriptive sum- the targets can be predicted accurately given relatively mary [26]. Examples of sample statistics include the coarse histograms. Our results also suggest caution in average, median and various other quantiles. While in the widespread use of aggregation for ensuring the the mean is still the most common choice, the best privacy of sensitive data. choice for summarizing a sample generally depends on the distribution the sample has been generated from. In summary, the main contributions of this manuscript In many cases, the use of histograms [24] or order are as follows: statistic summaries is much more \natural" e.g. for categorical data, binary data, count valued data, etc. (i) we propose a framework for estimating the re- sponse variables of a generalized linear model The problem of imputing individual level records from given only a histogram aggregate summary by the sample mean has been studied in [20] and [21] formulating it as an optimization problem that among others. In particular, the paper [21] attempts alternates between imputation and generalized to reconstruct the individual level matrix by assuming linear model fitting. a low rank structure and compares their framework with other approaches which include an extension of (ii) we examine a limiting case of the framework the neighborhood model [10] and a variation of ecolog- when all the data is known up to permutation. ical regression [13] for the task of imputing individual Our examination suggests the effectiveness of the level records of the response variable. However, these proposed approach when, in the original data, approaches exploit the fact that sample mean is a lin- permutation testing accurately ascertains the ve- ear function of the sample values. Hence, none of these racity of the linear relationship. approaches are extendable to non-linear functions such (iii) we examine a second limiting case where only a as order statistics and histogram aggregates. To the few order statistics are provided. Our experimen- best of our knowledge, the special case of individual in- tal results suggest a diminishing returns property ferences based on sample statistic aggregate data has - when a linear relationship holds in the origi- not been addressed before. Our work proposes a first nal data, the targets can be predicted accurately solution to address this open problem. given relatively coarse histograms. Order Statistics: Given a sample of n real valued datapoints, the τ th order statistic of the sample is the The proposed approach is applied to the analysis of τ th smallest value in the sample. For example, the simulated datasets. In addition, we examine the Texas first order statistic is the minimum value of the sam- Inpatient Discharge dataset from the Texas Depart- n th th ple, the 2 order statistic is the median and the n ment of State Health Services [3] and a subset of the order statistic is the maximum value of the sample. 2008-2010 SynPUF dataset [2]. We specifically design a framework which makes it rel- atively straightforward to work with order statistics. Notation Histograms : Given a finite sample of n items from Matrices are denoted by boldface capital letters, vec- a set C, a histogram is a partition of the set C into tors by boldface lower case letters and individual ele- disjoint bins Ci : [iCi = C and the respective count or 94 Avradeep Bhowmik, Joydeep Ghosh, Oluwasanmi Koyejo percentage of elements from the sample in each bin. A generalized linear model (GLM) [18] is a generaliza- Seen this way, for any sample from C ⊆ R, a his- tion of linear regression that subsumes various models togram is essentially a set of order statistics for that like Poisson regression, logistic regression, etc. as spe- sample. Histograms can sometimes be specified with- cial cases1. A generalized linear model assumes that out their boundary values (eg. "x < 30" as opposed the response variables, y are generated from a distribu- to "0 < x < 30")- this is equivalent to leaving out tion in the exponential family with the mean param- the first and the nth order statistic.
Recommended publications
  • Design and Implementation of N-Of-1 Trials: a User's Guide
    Design and Implementation of N-of-1 Trials: A User’s Guide N of 1 The Agency for Healthcare Research and Quality’s (AHRQ) Effective Health Care Program conducts and supports research focused on the outcomes, effectiveness, comparative clinical effectiveness, and appropriateness of pharmaceuticals, devices, and health care services. More information on the Effective Health Care Program and electronic copies of this report can be found at www.effectivehealthcare.ahrq. gov. This report was produced under contract to AHRQ by the Brigham and Women’s Hospital DEcIDE (Developing Evidence to Inform Decisions about Effectiveness) Methods Center under Contract No. 290-2005-0016-I. The AHRQ Task Order Officer for this project was Parivash Nourjah, Ph.D. The findings and conclusions in this document are those of the authors, who are responsible for its contents; the findings and conclusions do not necessarily represent the views ofAHRQ or the U.S. Department of Health and Human Services. Therefore, no statement in this report should be construed as an official position of AHRQ or the U.S. Department of Health and Human Services. Persons using assistive technology may not be able to fully access information in this report. For assistance contact [email protected]. The following investigators acknowledge financial support: Dr. Naihua Duan was supported in part by NIH grants 1 R01 NR013938 and 7 P30 MH090322, Dr. Heather Kaplan is part of the investigator team that designed and developed the MyIBD platform, and this work was supported in part by Cincinnati Children’s Hospital Medical Center. Dr. Richard Kravitz was supported in part by National Institute of Nursing Research Grant R01 NR01393801.
    [Show full text]
  • Meta-Analysis Using Individual Participant Data: One-Stage and Two-Stage Approaches, and Why They May Differ Danielle L
    CORE Metadata, citation and similar papers at core.ac.uk Provided by Keele Research Repository Tutorial in Biostatistics Received: 11 March 2016, Accepted: 13 September 2016 Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/sim.7141 Meta-analysis using individual participant data: one-stage and two-stage approaches, and why they may differ Danielle L. Burke*† Joie Ensor and Richard D. Riley Meta-analysis using individual participant data (IPD) obtains and synthesises the raw, participant-level data from a set of relevant studies. The IPD approach is becoming an increasingly popular tool as an alternative to traditional aggregate data meta-analysis, especially as it avoids reliance on published results and provides an opportunity to investigate individual-level interactions, such as treatment-effect modifiers. There are two statistical approaches for conducting an IPD meta-analysis: one-stage and two-stage. The one-stage approach analyses the IPD from all studies simultaneously, for example, in a hierarchical regression model with random effects. The two-stage approach derives aggregate data (such as effect estimates) in each study separately and then combines these in a traditional meta-analysis model. There have been numerous comparisons of the one-stage and two-stage approaches via theoretical consideration, simulation and empirical examples, yet there remains confusion regarding when each approach should be adopted, and indeed why they may differ. In this tutorial paper, we outline the key statistical methods for one-stage and two-stage IPD meta-analyses, and provide 10 key reasons why they may produce different summary results. We explain that most differences arise because of different modelling assumptions, rather than the choice of one-stage or two-stage itself.
    [Show full text]
  • Learning to Personalize Medicine from Aggregate Data
    medRxiv preprint doi: https://doi.org/10.1101/2020.07.07.20148205; this version posted July 8, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-ND 4.0 International license . Learning to Personalize Medicine from Aggregate Data Rich Colbaugh Kristin Glass Volv Global Lausanne, Switzerland Abstract There is great interest in personalized medicine, in which treatment is tailored to the individual characteristics of pa- tients. Achieving the objectives of precision healthcare will require clinically-grounded, evidence-based approaches, which in turn demands rigorous, scalable predictive analytics. Standard strategies for deriving prediction models for medicine involve acquiring ‘training’ data for large numbers of patients, labeling each patient according to the out- come of interest, and then using the labeled examples to learn to predict the outcome for new patients. Unfortunate- ly, labeling individuals is time-consuming and expertise-intensive in medical applications and thus represents a ma- jor impediment to practical personalized medicine. We overcome this obstacle with a novel machine learning algo- rithm that enables individual-level prediction models to be induced from aggregate-level labeled data, which is read- ily-available in many health domains. The utility of the proposed learning methodology is demonstrated by: i.) lev- eraging US county-level mental health statistics to create a screening tool which detects individuals suffering from depression based upon their Twitter activity; ii.) designing a decision-support system that exploits aggregate clinical trials data on multiple sclerosis (MS) treatment to predict which therapy would work best for the presenting patient; iii.) employing group-level clinical trials data to induce a model able to find those MS patients likely to be helped by an experimental therapy.
    [Show full text]
  • Good Statistical Practices for Contemporary Meta-Analysis: Examples Based on a Systematic Review on COVID-19 in Pregnancy
    Review Good Statistical Practices for Contemporary Meta-Analysis: Examples Based on a Systematic Review on COVID-19 in Pregnancy Yuxi Zhao and Lifeng Lin * Department of Statistics, Florida State University, Tallahassee, FL 32306, USA; [email protected] * Correspondence: [email protected] Abstract: Systematic reviews and meta-analyses have been increasingly used to pool research find- ings from multiple studies in medical sciences. The reliability of the synthesized evidence depends highly on the methodological quality of a systematic review and meta-analysis. In recent years, several tools have been developed to guide the reporting and evidence appraisal of systematic reviews and meta-analyses, and much statistical effort has been paid to improve their methodological quality. Nevertheless, many contemporary meta-analyses continue to employ conventional statis- tical methods, which may be suboptimal compared with several alternative methods available in the evidence synthesis literature. Based on a recent systematic review on COVID-19 in pregnancy, this article provides an overview of select good practices for performing meta-analyses from sta- tistical perspectives. Specifically, we suggest meta-analysts (1) providing sufficient information of included studies, (2) providing information for reproducibility of meta-analyses, (3) using appro- priate terminologies, (4) double-checking presented results, (5) considering alternative estimators of between-study variance, (6) considering alternative confidence intervals, (7) reporting predic- Citation: Zhao, Y.; Lin, L. Good tion intervals, (8) assessing small-study effects whenever possible, and (9) considering one-stage Statistical Practices for Contemporary methods. We use worked examples to illustrate these good practices. Relevant statistical code Meta-Analysis: Examples Based on a is also provided.
    [Show full text]
  • Reconciling Household Surveys and National Accounts Data Using a Cross Entropy Estimation Method
    TMD DISCUSSION PAPER NO. 50 RECONCILING HOUSEHOLD SURVEYS AND NATIONAL ACCOUNTS DATA USING A CROSS ENTROPY ESTIMATION METHOD Anne-Sophie Robilliard Sherman Robinson International Food Policy Research Institute Trade and Macroeconomics Division International Food Policy Research Institute 2033 K Street, N.W. Washington, D.C. 20006, U.S.A. November 1999 (Revised January 2001) TMD Discussion Papers contain preliminary material and research results, and are circulated prior to a full peer review in order to stimulate discussion and critical comment. It is expected that most Discussion Papers will eventually be published in some other form, and that their content may also be revised. This paper is available at http://www.cgiar.org/ifpri/divs/tmd/dp/papers/dp50.pdf Abstract This paper presents an approach to reconciling household surveys and national accounts data that starts from the assumption that the macro data represent control totals to which the household data must be reconciled, but the macro aggregates may be measured with error. The economic data gathered in the household survey are assumed to be accurate, or have been adjusted to be accurate. Given these assumptions, the problem is how to use the additional information provided by the national accounts data to re-estimate the household weights used in the survey so that the survey results are consistent with the aggregate data, while simultaneously estimating the errors in the aggregates. The estimation approach represents an efficient “information processing rule” using an estimation criterion based on an entropy measure of information. The survey household weights are treated as a prior. New weights are estimated that are close to the prior using a cross-entropy metric and that are also consistent with the additional information.
    [Show full text]
  • Supervised Learning by Training on Aggregate Outputs
    Supervised Learning by Training on Aggregate Outputs David R. Musicant, Janara M. Christensen, Jamie F. Olson Department of Computer Science Carleton College Northfield, MN [email protected], [email protected], [email protected] Abstract black carbon (BC) associated with an individual aerosol particle. While SPMS data provides detailed mass spectra Supervised learning is a classic data mining problem for individual aerosol particles, these spectra are not quanti- where one wishes to be be able to predict an output value tative enough to allow one to precisely identify the amount associated with a particular input vector. We present a new of BC present in a particle. On the other hand, filter-based twist on this classic problem where, instead of having the machines can collect quantitative BC data, but in aggregate training set contain an individual output value for each in- form. These devices allow one to determine how much put vector, the output values in the training set are only BC has been observed in aggregate over a particular time given in aggregate over a number of input vectors. This span, typically on the order of an hour. SPMS data, on the new problem arose from a particular need in learning on other hand, may be collected for individual particles (non- mass spectrometry data, but could easily apply to situations aggregate data) many times per minute. Our goal is to be when data has been aggregated in order to maintain pri- able to predict the quantity of BC present in an individual vacy. We provide a formal description of this new prob- aerosol particle.
    [Show full text]
  • Census Aggregate Data Workshop – 17 February 2015 • Ukdataservice.Ac.Uk/News-And-Events/Events
    2011 Census – aggregate data Richard Wiseman Justin Hayes Webinar January 2015 2011 Census – aggregate data The webinar will begin at 2pm • You now have a menu in the top right corner of your screen. • The red button with a white arrow allows you to expand and contract the webinar menu, in which you can write questions/comments. • We won’t have time to answer questions while we are presenting, but will answer them at the end • You will be on mute throughout – we can’t hear you. Welcome Some introductions… Richard Wiseman UK Data Service Mimas, Jisc Justin Hayes UK Data Service Mimas, Jisc Can you hear us? If you can’t hear us • Check your speakers/headset • Check your volume • Use the phone number in your invitation to call in • The webinar is being recorded - we’ll send a link Webinar structure • 30 minute presentation • 20 minutes for questions and answers • Finish by 2.50pm • Please use chat facility to ask questions or let us know of any problems • Some questions for the audience during presentation Presentation content • The UK Data Service • UK Data Service Census Support • Census background • Aggregate data • Geographies • InFuse • Demo • Data / geography model • Next release • Future plans The UK Data Service • Funded by the ESRC, integrating several previous resources • A single, comprehensive and integrated point of access to a wide range of social science data • Support, training and guidance • ukdataservice.ac.uk UK Data Service Census Support • A specialist unit of the UK Data Service • Access to, and support for use of data
    [Show full text]
  • Aggregation Effects in Generalized Linear Models: a Biochemical Engineering Application
    Iowa State University Capstones, Theses and Creative Components Dissertations Summer 2019 Aggregation Effects in Generalized Linear Models: A Biochemical Engineering Application Xiaojing Zhong Iowa State University Follow this and additional works at: https://lib.dr.iastate.edu/creativecomponents Part of the Engineering Commons Recommended Citation Zhong, Xiaojing, "Aggregation Effects in Generalized Linear Models: A Biochemical Engineering Application" (2019). Creative Components. 463. https://lib.dr.iastate.edu/creativecomponents/463 This Creative Component is brought to you for free and open access by the Iowa State University Capstones, Theses and Dissertations at Iowa State University Digital Repository. It has been accepted for inclusion in Creative Components by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected]. Aggregation Effects in Generalized Linear Models: A Biochemical Engineering Application by Xiaojing Zhong Major: Statistics Program of Study Committee: Mark Kaiser, Major Professor Emily Berg Yumou Qiu Iowa State University Ames, Iowa 2019 ii TABLE OF CONTENTS Page TABLE OF CONTENTS .................................................................................................... ii LIST OF FIGURES ........................................................................................................... iii LIST OF TABLES .............................................................................................................. v CHAPTER
    [Show full text]
  • Effects of Data Aggregation on Time Series Analysis of Seasonal Infections
    International Journal of Environmental Research and Public Health Article Effects of Data Aggregation on Time Series Analysis of Seasonal Infections Tania M. Alarcon Falconi 1,*, Bertha Estrella 2, Fernando Sempértegui 2 and Elena N. Naumova 1 1 Division of Nutrition Epidemiology and Data Science, Friedman School of Nutrition Science and Policy, Tufts University, Boston, MA 02111, USA; [email protected] 2 Department of Immunology, Faculty of Medical Sciences, Central University, Quito 170136, Ecuador; [email protected] (B.E.); [email protected] (F.S.) * Correspondence: [email protected]; Tel.: +1-617-320-4306 Received: 30 June 2020; Accepted: 5 August 2020; Published: 13 August 2020 Abstract: Time series analysis in epidemiological studies is typically conducted on aggregated counts, although data tend to be collected at finer temporal resolutions. The decision to aggregate data is rarely discussed in epidemiological literature although it has been shown to impact model results. We present a critical thinking process for making decisions about data aggregation in time series analysis of seasonal infections. We systematically build a harmonic regression model to characterize peak timing and amplitude of three respiratory and enteric infections that have different seasonal patterns and incidence. We show that irregularities introduced when aggregating data must be controlled during modeling to prevent erroneous results. Aggregation irregularities had a minimal impact on the estimates of trend, amplitude, and peak timing for daily and weekly data regardless of the disease. However, estimates of peak timing of the more common infections changed by as much as 2.5 months when controlling for monthly data irregularities.
    [Show full text]
  • A Critique of Structural Vars Using Real Business Cycle Theory∗
    Federal Reserve Bank of Minneapolis Research Department A Critique of Structural VARs Using Real Business Cycle Theory∗ V. V. Chari, Patrick J. Kehoe, Ellen R. McGrattan Working Paper 631 Revised February 2005 ABSTRACT The main substantive finding of the recent structural vector autoregression literature with a differ- enced specification of hours (DSVAR) is that technology shocks lead to a fall in hours. Researchers have used these results to argue that standard business cycle models in which technology shocks leads to a rise in hours should be discarded. We evaluate the DSVAR approach by asking the following: Is the specification derived from this approach misspecified when the data is generated by the very model the literature is trying to discard, namely the standard business cycle model? We find that it is misspecified. Moreover, this misspecification is so great that it leads to mistaken inferences that are quantitatively large. We show that the other popular specification which uses the level of hours (LSVAR) is also misspecified with respect to the standard business cycle model. We argue that an alternative approach, the business cycle accounting approach, is a more fruitful technique for guiding the development of business cycle theory. ∗Chari, University of Minnesota and Federal Reserve Bank of Minneapolis; Kehoe, Federal Reserve Bank of Minneapolis and University of Minnesota; McGrattan, Federal Reserve Bank of Minneapolis and University of Minnesota. The authors thank the National Science Foundation for support. The views expressed herein are those of the authors and not necessarily those of the Federal Reserve Bank of Minneapolis or the Federal Reserve System.
    [Show full text]
  • Comparing the Forecasting Performance of Var, Bvar and U-Midas
    COMPARING THE FORECASTING PERFORMANCE OF VAR, BVAR AND U-MIDAS Submitted by Alessio Belloni A thesis submitted to the Department of Statistics in partial fulfillment of the requirements for a one-year Master degree in Statistics in the Faculty of Social Sciences Supervisors: Sebastian Ankargren & Mattias Nordin Spring, 2017 ABSTRACT This paper aims to compare the forecasting performance of the widely used VAR and Bayesian VAR model to the unrestricted MIDAS regression. The models are tested on a real-time macroeconomic data set ranging from 2000 to 2015. The variables are mixed frequency data, specifically, predictions are made for GDP, using economic tendency indicator, unemployment and inflation as predicting variables. The baseline model of this analysis is a simple VAR, while it has great flexibility, this model risks to overfit the data and as a consequence makes unreliable predictions. The Villani Bayesian VAR is meant to solve this problem by introducing long run beliefs about the data structure and the steady state unconditional means of each se- ries. When facing mixed frequency data, both these approached aggregate at the lower level by discarding useful information. In this scenario, the unrestricted MIDAS model addresses this problem without losing high frequency information. The results show how both BVAR and U-MIDAS outperform VAR at every horizon, while there is no absolute winner among BVAR and U-MIDAS. Evidence suggests that U-MIDAS is superior for short horizons, specifically up to the 5th step ahead, which corresponds to one year and a quarter. Contents 1 Introduction 3 2 Theory 4 2.1 VAR........................................4 2.2 Bayesian VAR .
    [Show full text]
  • Improving Ecological Inference by Predicting Individual Ethnicity from Voter Registration Records
    Advance Access publication March 17, 2016 Political Analysis (2016) 24:263–272 doi:10.1093/pan/mpw001 Improving Ecological Inference by Predicting Individual Ethnicity from Voter Registration Records Kosuke Imai Department of Politics and Center for Statistics and Machine Learning, Princeton University, Princeton, NJ 08544 e-mail: [email protected]; URL: http://imai.princeton.edu (corresponding author) Kabir Khanna Department of Politics, Princeton University, Princeton, NJ 08544 Downloaded from Edited by Justin Grimmer In both political behavior research and voting rights litigation, turnout and vote choice for different racial groups are often inferred using aggregate election results and racial composition. Over the past several http://pan.oxfordjournals.org/ decades, many statistical methods have been proposed to address this ecological inference problem. We propose an alternative method to reduce aggregation bias by predicting individual-level ethnicity from voter registration records. Building on the existing methodological literature, we use Bayes’s rule to combine the Census Bureau’s Surname List with various information from geocoded voter registration records. We evaluate the performance of the proposed methodology using approximately nine million voter registration records from Florida, where self-reported ethnicity is available. We find that it is possible to reduce the false positive rate among Black and Latino voters to 6% and 3%, respectively, while maintaining the true positive rate above 80%. Moreover, we use our predictions to estimate turnout by race and find that our estimates yields substantially less amounts of bias and root mean squared error than standard ecological inference at Princeton University on April 22, 2016 estimates. We provide open-source software to implement the proposed methodology.
    [Show full text]