Working Papers
2000
46 records found
Government Formation in Parliamentary Democracies
Martin, Lanny W., Stevenson, Randolph T.
Submitted: 2000-01-27
Keywords: government formation, coalition politics, conditional logit
Abstract: (click to show/hide) The literature on cabinet formation in parliamentary democracies is replete with theoretical explanations of why some cabinets form and others do not. This theoretical richness, however, has not led to the development of a healthy empirical literature designed to choose between competing theories. In this paper, we try to rectify this problem by developing an empirical model that can adequately capture the kind of choice situation that is inherent in cabinet selection and then using it to evaluate the leading theories of cabinet formation that have been advanced in the literature. For example, this analysis allows us to make conclusions about the relative importance in cabinet formation of traditional variables like size and ideology, as well as to evaluate the impact that recent new-institutionalist theories (such as Laver and Shepsle 1996) have on our ability to predict and explain cabinet formation over and above the more traditional explanations.
Martin, Lanny W., Stevenson, Randolph T.
Submitted: 2000-01-27
Keywords: government formation, coalition politics, conditional logit
Abstract: (click to show/hide) The literature on cabinet formation in parliamentary democracies is replete with theoretical explanations of why some cabinets form and others do not. This theoretical richness, however, has not led to the development of a healthy empirical literature designed to choose between competing theories. In this paper, we try to rectify this problem by developing an empirical model that can adequately capture the kind of choice situation that is inherent in cabinet selection and then using it to evaluate the leading theories of cabinet formation that have been advanced in the literature. For example, this analysis allows us to make conclusions about the relative importance in cabinet formation of traditional variables like size and ideology, as well as to evaluate the impact that recent new-institutionalist theories (such as Laver and Shepsle 1996) have on our ability to predict and explain cabinet formation over and above the more traditional explanations.
Post-stratification without population level information on the post-stratifying variable, with application to political polling
Gelman, Andrew, Katz, Jonathan, Riley, Cavan
Submitted: 2000-02-10
Keywords: Bayesian Inference, Post-stratification, Sample surveys, State-space models
Abstract: (click to show/hide) We investigate the construction of more precise estimates of a collection of population means using information about a related variable in the context of repeated sample surveys. The method is illustrated using poll results concerning presidential approval rating (our related variable is political party identification). We use post-stratification to construct these improved estimates, but since we don't have population level information on the post-stratifying variable, we construct a model for the manner in which the post-stratifier develops over time. In this manner, we obtain more precise estimates without making possibly untenable assumptions about the dynamics of our variable of interest, the presidential approval rating.
Gelman, Andrew, Katz, Jonathan, Riley, Cavan
Submitted: 2000-02-10
Keywords: Bayesian Inference, Post-stratification, Sample surveys, State-space models
Abstract: (click to show/hide) We investigate the construction of more precise estimates of a collection of population means using information about a related variable in the context of repeated sample surveys. The method is illustrated using poll results concerning presidential approval rating (our related variable is political party identification). We use post-stratification to construct these improved estimates, but since we don't have population level information on the post-stratifying variable, we construct a model for the manner in which the post-stratifier develops over time. In this manner, we obtain more precise estimates without making possibly untenable assumptions about the dynamics of our variable of interest, the presidential approval rating.
Mixed Logit Models for Multiparty Elections
Glasgow, Garrett
Submitted: 2000-02-24
Keywords: mixed logit, random parameters logit, multinomial probit, IIA
Abstract: (click to show/hide) This is a significantly updated version of my February 24 submission, with several mathematical errors corrected and new information on multinomial probit models and IIA violations. In this paper I introduce the mixed logit (MXL), a flexible discrete choice model based on random utility maximization. Mixed logit is the most flexible discrete choice model available for the study of multiparty and multicandidate elections --- even more flexible than multinomial probit (MNP), the discrete choice model currently favored for the study of elections of this type. Like MNP, MXL does not assume IIA, and can thus estimate realistic substitution patterns between alternatives. In fact, MXL can be specified to estimate the same substitution patterns as any specification of MNP. Further, since the unobserved components of MXL are not constrained to follow a normal distribution, and are not estimated as elements in a covariance matrix, MXL can include any number of random coefficients or error components that can follow any distribution. MXL is no more difficult to estimate than MNP. An empirical example using data from the 1987 British general election demonstrates the utility of MXL in the study of multicandidate and multiparty elections.
Glasgow, Garrett
Submitted: 2000-02-24
Keywords: mixed logit, random parameters logit, multinomial probit, IIA
Abstract: (click to show/hide) This is a significantly updated version of my February 24 submission, with several mathematical errors corrected and new information on multinomial probit models and IIA violations. In this paper I introduce the mixed logit (MXL), a flexible discrete choice model based on random utility maximization. Mixed logit is the most flexible discrete choice model available for the study of multiparty and multicandidate elections --- even more flexible than multinomial probit (MNP), the discrete choice model currently favored for the study of elections of this type. Like MNP, MXL does not assume IIA, and can thus estimate realistic substitution patterns between alternatives. In fact, MXL can be specified to estimate the same substitution patterns as any specification of MNP. Further, since the unobserved components of MXL are not constrained to follow a normal distribution, and are not estimated as elements in a covariance matrix, MXL can include any number of random coefficients or error components that can follow any distribution. MXL is no more difficult to estimate than MNP. An empirical example using data from the 1987 British general election demonstrates the utility of MXL in the study of multicandidate and multiparty elections.
A Positive Theory of Bureaucratic Discretion as Agency Choice
Krause, George
Submitted: 2000-02-24
Keywords: bureaucratic discretion, administrative decision-making, uncertainty, policy implementation, formal theory
Abstract: (click to show/hide) Existing research on the positive theory of bureaucratic discretion views this phenomenon as a "supply-side" concept that elected officials determine without considering bureaucratic preferences altogether, or by merely treating it as being exogenous to the optimization problem confronting politicians. It has been well established by bureaucracy scholars that agencies have preferences concerning bureaucratic discretion and are proactive in trying to get these preferences met (e.g., Rourke 1984; Wilson 1989). In this essay, I set forth a "demand-side" theory of bureaucratic discretion where an administrative agency's preferences for this commodity under conditions of uncertainty is determined through the relationship between its utility and (a) bureaucratic discretion, and (b) policy (implementation) outcome uncertainty, separately. Moreover, I argue that the discretionary context confronting the agency will matter, and thus incorporate this into the theoretical model. Hypotheses concerning the discretionary context by which administrative agencies will view bureaucratic discretion are generated from this model. Finally, I propose a statistical test that could be employed to empirically test the theoretical predictions of the "demand-side" model of bureaucratic discretion set forth in this paper.
Krause, George
Submitted: 2000-02-24
Keywords: bureaucratic discretion, administrative decision-making, uncertainty, policy implementation, formal theory
Abstract: (click to show/hide) Existing research on the positive theory of bureaucratic discretion views this phenomenon as a "supply-side" concept that elected officials determine without considering bureaucratic preferences altogether, or by merely treating it as being exogenous to the optimization problem confronting politicians. It has been well established by bureaucracy scholars that agencies have preferences concerning bureaucratic discretion and are proactive in trying to get these preferences met (e.g., Rourke 1984; Wilson 1989). In this essay, I set forth a "demand-side" theory of bureaucratic discretion where an administrative agency's preferences for this commodity under conditions of uncertainty is determined through the relationship between its utility and (a) bureaucratic discretion, and (b) policy (implementation) outcome uncertainty, separately. Moreover, I argue that the discretionary context confronting the agency will matter, and thus incorporate this into the theoretical model. Hypotheses concerning the discretionary context by which administrative agencies will view bureaucratic discretion are generated from this model. Finally, I propose a statistical test that could be employed to empirically test the theoretical predictions of the "demand-side" model of bureaucratic discretion set forth in this paper.
The Vote-Stealing and Turnout Effects of Third-Party Candidates in U.S. Presidential Elections, 1968-1996
Lacy, Dean, Burden, Barry C.
Submitted: 2000-03-03
Keywords: vote choice, turnout, third parties, multinomial probit
Abstract: (click to show/hide) A multinomial probit model of electoral choice in the 1968, 1980, 1992, and 1996 U.S. presidential elections, estimated using data from the American National Election Studies, reveals similarities and differences in electoral support for George Wallace, John Anderson, and Ross Perot. Estimates from the models are used to simulate the outcomes of the elections in the absence of the third-party candidate and under full turnout. In three of the four elections, the third-party candidates stole more votes from the challengers than from the incumbents. Only in 1996 did the third-party candidate take more votes away from the incumbent than the challenger. None of the four third-party candidacies increased turnout by more than 2.3 percentage points, and Perot's 1996 candidacy had the smallest impact on turnout of all of the third-party candidacies. Under full turnout, the outcome of only one election - 1968 - may have changed. All four third-party candidates increase their vote share under full turnout, and Democratic candidates gain vote share under full turnout in all elections except 1980. The paper also describes a new method for estimating the error variances and covariances in an MNP model.
Lacy, Dean, Burden, Barry C.
Submitted: 2000-03-03
Keywords: vote choice, turnout, third parties, multinomial probit
Abstract: (click to show/hide) A multinomial probit model of electoral choice in the 1968, 1980, 1992, and 1996 U.S. presidential elections, estimated using data from the American National Election Studies, reveals similarities and differences in electoral support for George Wallace, John Anderson, and Ross Perot. Estimates from the models are used to simulate the outcomes of the elections in the absence of the third-party candidate and under full turnout. In three of the four elections, the third-party candidates stole more votes from the challengers than from the incumbents. Only in 1996 did the third-party candidate take more votes away from the incumbent than the challenger. None of the four third-party candidacies increased turnout by more than 2.3 percentage points, and Perot's 1996 candidacy had the smallest impact on turnout of all of the third-party candidacies. Under full turnout, the outcome of only one election - 1968 - may have changed. All four third-party candidates increase their vote share under full turnout, and Democratic candidates gain vote share under full turnout in all elections except 1980. The paper also describes a new method for estimating the error variances and covariances in an MNP model.
The-Stage Estimation of Stochastic Truncation Models with Limited Dependent Variables
Boehmke, Frederick
Submitted: 2000-04-13
Keywords: selection bias, stochastic truncation, maximum likelihood, simulation, monte carlo, initiative, interest groups
Abstract: (click to show/hide) Recent work has made progress in estimating models involving selection bias of a par ticularly strong nature: all nonrespondents are unit nonresponders, meaning that no data is available for them. These models are reasonable successful in circumstances where the dependent variable of interest is continuous, but they are less practical empirically when it is latent and only discrete outcomes or choices are observed. I develop a method in this paper to estimate these models that is much more practical in terms of estimation. The model uses a small amount of auxiliary information to estimate the selection equation and these parameters are then used to estimate the equation of interest in a maximum likelihood setting. After presenting monte carlo analysis to support the model, I apply the technique to a substantive problem: which interest groups are likely to turn to the initiative process to achieve their policy goals.
Boehmke, Frederick
Submitted: 2000-04-13
Keywords: selection bias, stochastic truncation, maximum likelihood, simulation, monte carlo, initiative, interest groups
Abstract: (click to show/hide) Recent work has made progress in estimating models involving selection bias of a par ticularly strong nature: all nonrespondents are unit nonresponders, meaning that no data is available for them. These models are reasonable successful in circumstances where the dependent variable of interest is continuous, but they are less practical empirically when it is latent and only discrete outcomes or choices are observed. I develop a method in this paper to estimate these models that is much more practical in terms of estimation. The model uses a small amount of auxiliary information to estimate the selection equation and these parameters are then used to estimate the equation of interest in a maximum likelihood setting. After presenting monte carlo analysis to support the model, I apply the technique to a substantive problem: which interest groups are likely to turn to the initiative process to achieve their policy goals.
Ideology and U.S. Senate Candidates
Burden, Barry C., Kenny, Christopher B.
Submitted: 2000-04-19
Keywords: ideology, measurement, elite surveys
Abstract: (click to show/hide) This paper reports on a pilot study for what will become the Candidate Ideology Survey (CIS). Beginning in 2000, the CIS will survey all major-party House and Senate candidates, asking them to locate themselves on the left-right ideological spectrum. Such an approach improves on existing ideology measures such those based on roll call votes because it puts both incumbents and challengers on a common scale. Existing studies of congressional elections that include only the ideology of the incumbent in vote models are likely underestimating the importance of ideology generally, the positions of challengers are useful if not necessary. The paper presents findings from a preliminary survey of senators and Senate challengers in 1998. It explains the ususual elite mail survey methodology used in terms of response rate and representativeness of the sample. It also examines the validity of the data in terms of partisan and regional differences and relationships with existing ideological measures. Among other substance results, we find that the ideological "fit" of incumbents with constituents is much better than the "fit" of challengers with constituents. By improving on this design and adding the House in the 2000 CIS wave, we hope to generate data that will be of great use to researchers who study congressional elections.
Burden, Barry C., Kenny, Christopher B.
Submitted: 2000-04-19
Keywords: ideology, measurement, elite surveys
Abstract: (click to show/hide) This paper reports on a pilot study for what will become the Candidate Ideology Survey (CIS). Beginning in 2000, the CIS will survey all major-party House and Senate candidates, asking them to locate themselves on the left-right ideological spectrum. Such an approach improves on existing ideology measures such those based on roll call votes because it puts both incumbents and challengers on a common scale. Existing studies of congressional elections that include only the ideology of the incumbent in vote models are likely underestimating the importance of ideology generally, the positions of challengers are useful if not necessary. The paper presents findings from a preliminary survey of senators and Senate challengers in 1998. It explains the ususual elite mail survey methodology used in terms of response rate and representativeness of the sample. It also examines the validity of the data in terms of partisan and regional differences and relationships with existing ideological measures. Among other substance results, we find that the ideological "fit" of incumbents with constituents is much better than the "fit" of challengers with constituents. By improving on this design and adding the House in the 2000 CIS wave, we hope to generate data that will be of great use to researchers who study congressional elections.
Forecasting State Failure
King, Gary, Zeng, Langche
Submitted: 2000-04-19
Keywords: forecast, neural network, committee methods, case-control
Abstract: (click to show/hide) We offer the first independent scholarly evaluation of the claims, forecasts, and causal inferences of the State Failure Task Force and their efforts to forecast when states will fail. This Task Force, set up at the behest of Vice President Gore in 1994, has been led by a group of distinguished academics working as consultants to the U.S.\ Government. State failure is a grave condition that includes civil wars, revolutionary wars, genocides, politicides, and adverse or disruptive regime transitions. State Failure Task Force reports and publications have received widespread attention in the media, in academia, and from public policy decision-makers. In this paper, we identify several methodological errors in the Task Force work that cause their reported forecast probabilities of conflict to be much too large, their causal inferences to be biased in unpredictable directions, and their claims of forecasting performance to be exaggerated. However, we also find that the Task Force has amassed the best and most carefully collected data on state failure in existence, and the required corrections, although very large in effect, are easy to implement. We also reanalyze their data with better statistical and other procedures and demonstrate how to improve forecasting performance to levels significantly greater than even corrected versions of their models. We hope that this work leads to better use of political science and statistical analyses in public policy, but most of the claims analyzed are also of direct relevance to ongoing scholarly debates in political science, public health, and other disciplines.
King, Gary, Zeng, Langche
Submitted: 2000-04-19
Keywords: forecast, neural network, committee methods, case-control
Abstract: (click to show/hide) We offer the first independent scholarly evaluation of the claims, forecasts, and causal inferences of the State Failure Task Force and their efforts to forecast when states will fail. This Task Force, set up at the behest of Vice President Gore in 1994, has been led by a group of distinguished academics working as consultants to the U.S.\ Government. State failure is a grave condition that includes civil wars, revolutionary wars, genocides, politicides, and adverse or disruptive regime transitions. State Failure Task Force reports and publications have received widespread attention in the media, in academia, and from public policy decision-makers. In this paper, we identify several methodological errors in the Task Force work that cause their reported forecast probabilities of conflict to be much too large, their causal inferences to be biased in unpredictable directions, and their claims of forecasting performance to be exaggerated. However, we also find that the Task Force has amassed the best and most carefully collected data on state failure in existence, and the required corrections, although very large in effect, are easy to implement. We also reanalyze their data with better statistical and other procedures and demonstrate how to improve forecasting performance to levels significantly greater than even corrected versions of their models. We hope that this work leads to better use of political science and statistical analyses in public policy, but most of the claims analyzed are also of direct relevance to ongoing scholarly debates in political science, public health, and other disciplines.
Flexible Prior Specifications for Factor Analytic Models with an Application to the Measurement of American Political Ideology
Quinn, Kevin M.
Submitted: 2000-04-20
Keywords: factor analysis, intrinsic autoregression, hierarchical modeling, Bayesian inference, political ideology
Abstract: (click to show/hide) Factor analytic measurement models are widely used in the social sciences to measure latent variables and functions thereof. Examples include the measurement of: political preferences, liberal democracy, latent determinants of exchange rates, and latent factors in arbitrage pricing theory models and the corresponding pricing deviations. Oftentimes, the results of these measurement models are sensitive to distributional assumptions that are made regarding the latent factors. In this paper I demonstrate how prior distributions commonly used in image processing and spatial statistics provide a flexible means to model dependencies among the latent factor scores that cannot be easily captured with standard prior distributions that treat the factor scores as (conditionally) exchangeable. Markov chain Monte Carlo techniques are used to fit the resulting models. These modeling techniques are illustrated with a simulated data example and an analysis of American political attitudes drawn from the 1996 American National Election Study.
Quinn, Kevin M.
Submitted: 2000-04-20
Keywords: factor analysis, intrinsic autoregression, hierarchical modeling, Bayesian inference, political ideology
Abstract: (click to show/hide) Factor analytic measurement models are widely used in the social sciences to measure latent variables and functions thereof. Examples include the measurement of: political preferences, liberal democracy, latent determinants of exchange rates, and latent factors in arbitrage pricing theory models and the corresponding pricing deviations. Oftentimes, the results of these measurement models are sensitive to distributional assumptions that are made regarding the latent factors. In this paper I demonstrate how prior distributions commonly used in image processing and spatial statistics provide a flexible means to model dependencies among the latent factor scores that cannot be easily captured with standard prior distributions that treat the factor scores as (conditionally) exchangeable. Markov chain Monte Carlo techniques are used to fit the resulting models. These modeling techniques are illustrated with a simulated data example and an analysis of American political attitudes drawn from the 1996 American National Election Study.
Estimating King's ecological inference normal model via the EM algorithm
Mattos, Rogerio, Veiga, Alvaro
Submitted: 2000-04-20
Keywords: ecological inference, disaggregate data, exponential families, truncated normal, EM Algorithm
Abstract: (click to show/hide) Recently, Gary King introduced a new model for ecological inference, based on a truncated bivariate normal, which he estimates by maximum probability and uses to simulate the predictive densities of the disaggregate data. This paper reviews King's model and its assumption of truncated normality, with the aim to implement maximum probability estimation of his model and disaggregate data prediction in an alternative fashion via the EM Algorithm. In addition, we highlight and discuss important modeling issues related to the chance of non-existence of maximum likelihood estimates, and to the degree that corrections for this non-existence by means of suitably chosen priors are effective. At the end, a Monte Carlo simulation study is run in order to compare the two approaches.
Mattos, Rogerio, Veiga, Alvaro
Submitted: 2000-04-20
Keywords: ecological inference, disaggregate data, exponential families, truncated normal, EM Algorithm
Abstract: (click to show/hide) Recently, Gary King introduced a new model for ecological inference, based on a truncated bivariate normal, which he estimates by maximum probability and uses to simulate the predictive densities of the disaggregate data. This paper reviews King's model and its assumption of truncated normality, with the aim to implement maximum probability estimation of his model and disaggregate data prediction in an alternative fashion via the EM Algorithm. In addition, we highlight and discuss important modeling issues related to the chance of non-existence of maximum likelihood estimates, and to the degree that corrections for this non-existence by means of suitably chosen priors are effective. At the end, a Monte Carlo simulation study is run in order to compare the two approaches.
Is Ticket Splitting Strategic? Evidence from the 1998 Election in Germany
Gschwend, Thomas
Submitted: 2000-04-20
Keywords: ticket splitting, strategic voting, Germany, election, MNP
Abstract: (click to show/hide) Germany provides an especially interesting case for the study of strategic voting because they use a two-ballot system on Election Day. Voters are encouraged to split their votes using different strategies. The paper is an example of how much more can be learned if we reconsider and refine our theories. I provide a first step towards a theory of strategic voting and add it to the typical ticket splitting discussion. In order to test more refined hypotheses about ticket splitting and strategic voting I use cross-sectional data from the German National Post Election Study of 1998. Empirically, the results indicate that strategic voters are different from ordinary ticket splitters. Evidence from separate MNP estimation for East and West Germany shows that identifier of the FDP or the Greens are more likely strategic voters as opposed to non-strategic ticket splitters. Non-strategic ticket splitters in East Germany do not feel close to any political party. In West Germany non-strategic ticket splitters have conflicting party preferences. Thus, it proves useful to separate out strategic voters from ordinary ticket splitters in future work.
Gschwend, Thomas
Submitted: 2000-04-20
Keywords: ticket splitting, strategic voting, Germany, election, MNP
Abstract: (click to show/hide) Germany provides an especially interesting case for the study of strategic voting because they use a two-ballot system on Election Day. Voters are encouraged to split their votes using different strategies. The paper is an example of how much more can be learned if we reconsider and refine our theories. I provide a first step towards a theory of strategic voting and add it to the typical ticket splitting discussion. In order to test more refined hypotheses about ticket splitting and strategic voting I use cross-sectional data from the German National Post Election Study of 1998. Empirically, the results indicate that strategic voters are different from ordinary ticket splitters. Evidence from separate MNP estimation for East and West Germany shows that identifier of the FDP or the Greens are more likely strategic voters as opposed to non-strategic ticket splitters. Non-strategic ticket splitters in East Germany do not feel close to any political party. In West Germany non-strategic ticket splitters have conflicting party preferences. Thus, it proves useful to separate out strategic voters from ordinary ticket splitters in future work.
Bayesian Inference for Heterogeneous Event Counts
Martin, Andrew D.
Submitted: 2000-04-20
Keywords: hierarchical models, Poisson, event count, heterogeneity
Abstract: (click to show/hide) This paper presents a handful of Bayesian tools one can use to model heterogeneous event counts. In many political science applications we are interested in modeling the number of times a particular event takes place. While models for event count cross-sections are now widely used in political science (King, 1988, 1989b), little has been written about how to model counts when contextual factors introduce heterogeneity. I begin with a discussion of Bayesian cross-sectional count models and introduce an alternative model for counts with overdispersion. To illustrate the Bayesian framework, I model event counts of the number of discharge petitions from the 61st to the 105th House, and the number of women's rights bills cosponsored by each member in the 92nd House. I then generalize the model to allow for contextual heterogeneity and posit a hierarchical Poisson regression model, fitting this model to the number of women rights cosponsorships for each member of the 83rd to 102nd House. I demonstrate the advantages of this approach over pooled and independent Poisson regressions. The hierarchical model allows one to explicitly model contextual factors and test alternative contextual explanations. Additionally, I discuss software one can use to easily implement these models with little start-up cost.
Martin, Andrew D.
Submitted: 2000-04-20
Keywords: hierarchical models, Poisson, event count, heterogeneity
Abstract: (click to show/hide) This paper presents a handful of Bayesian tools one can use to model heterogeneous event counts. In many political science applications we are interested in modeling the number of times a particular event takes place. While models for event count cross-sections are now widely used in political science (King, 1988, 1989b), little has been written about how to model counts when contextual factors introduce heterogeneity. I begin with a discussion of Bayesian cross-sectional count models and introduce an alternative model for counts with overdispersion. To illustrate the Bayesian framework, I model event counts of the number of discharge petitions from the 61st to the 105th House, and the number of women's rights bills cosponsored by each member in the 92nd House. I then generalize the model to allow for contextual heterogeneity and posit a hierarchical Poisson regression model, fitting this model to the number of women rights cosponsorships for each member of the 83rd to 102nd House. I demonstrate the advantages of this approach over pooled and independent Poisson regressions. The hierarchical model allows one to explicitly model contextual factors and test alternative contextual explanations. Additionally, I discuss software one can use to easily implement these models with little start-up cost.
Anatomy of a Third-Party Victory: Electoral Support for Jesse Ventura in the 1998 Minnesota Gubernatorial Election
Lacy, Dean, Monson, Quin
Submitted: 2000-04-24
Keywords: vote-stealing, turnout, third-party, Condorcet winner, multinomial probit
Abstract: (click to show/hide) [not transcribed]
Lacy, Dean, Monson, Quin
Submitted: 2000-04-24
Keywords: vote-stealing, turnout, third-party, Condorcet winner, multinomial probit
Abstract: (click to show/hide) [not transcribed]
The Dynamics of the Gender Gap
Box-Steffensmeier, Janet M., De Boef, Suzanna, Lin, Tse-Min
Submitted: 2000-04-24
Keywords:
Abstract: (click to show/hide) Ever since the late 1970s, women have been significantly more likely to identify with the Democratic Party than men. Most efforts to explain this gender gap are based on static explanations and are tested using individual level analyses of cross-sectional data, typically election data. We look at the gender gap in partisanship over time. We use data from 1979 to 1998 and sophisticated time series methods to examine the dynamics of the gender gap. Specifically, we present a Vector Autoregression (VAR) framework to estimate relationships after appropriately characterizing each series with an AutoRegressive Fractionally Integrated Moving Average (ARFIMA) model. We demonstrate that the gender gap is a continuous and evolving feature of the political, economic, and structural environment. As the mood of the country becomes more conservative, the gender gap increases. Economic measures of unemployment, inflation, and income growth all have a statistically significant effect on the gap. Finally, the gender gap increases as the percentage of economically vulnerable, single women increases.
Box-Steffensmeier, Janet M., De Boef, Suzanna, Lin, Tse-Min
Submitted: 2000-04-24
Keywords:
Abstract: (click to show/hide) Ever since the late 1970s, women have been significantly more likely to identify with the Democratic Party than men. Most efforts to explain this gender gap are based on static explanations and are tested using individual level analyses of cross-sectional data, typically election data. We look at the gender gap in partisanship over time. We use data from 1979 to 1998 and sophisticated time series methods to examine the dynamics of the gender gap. Specifically, we present a Vector Autoregression (VAR) framework to estimate relationships after appropriately characterizing each series with an AutoRegressive Fractionally Integrated Moving Average (ARFIMA) model. We demonstrate that the gender gap is a continuous and evolving feature of the political, economic, and structural environment. As the mood of the country becomes more conservative, the gender gap increases. Economic measures of unemployment, inflation, and income growth all have a statistically significant effect on the gap. Finally, the gender gap increases as the percentage of economically vulnerable, single women increases.
Interest Group Ratings, Measurement Error, and Regression Inconsistency
Herron, Michael C.
Submitted: 2000-04-25
Keywords: interest group ratings, regressions, measurement error, party effects
Abstract: (click to show/hide) This paper uses spatial voting theory to analyze the properties of errors in interest group ratings insofar as ratings are used to measure legislator policy preferences. I show that, in general, rating errors are not mean zero; that the errors in a set of ratings are correlated with the underlying legislator preferences measured by the ratings; that ordinary least squares estimation of a bivariate regression which uses ratings as independent variables produces inconsistent coefficient estimates; that instrumenting for the aforementioned interest group ratings with a second set of ratings, as proposed by Brunell et al., will not necessarily fix this problem and can actually make matters worse; that, paradoxically, biased interest ratings are sometimes better for regression estimates than are unbiased ratings; and, finally, that estimating a trivariate regression with both interest group ratings and a party indicator on its right hand side produces inconsistent estimates and, moreover, a party coefficient estimate which has an unreliable sign.
Herron, Michael C.
Submitted: 2000-04-25
Keywords: interest group ratings, regressions, measurement error, party effects
Abstract: (click to show/hide) This paper uses spatial voting theory to analyze the properties of errors in interest group ratings insofar as ratings are used to measure legislator policy preferences. I show that, in general, rating errors are not mean zero; that the errors in a set of ratings are correlated with the underlying legislator preferences measured by the ratings; that ordinary least squares estimation of a bivariate regression which uses ratings as independent variables produces inconsistent coefficient estimates; that instrumenting for the aforementioned interest group ratings with a second set of ratings, as proposed by Brunell et al., will not necessarily fix this problem and can actually make matters worse; that, paradoxically, biased interest ratings are sometimes better for regression estimates than are unbiased ratings; and, finally, that estimating a trivariate regression with both interest group ratings and a party indicator on its right hand side produces inconsistent estimates and, moreover, a party coefficient estimate which has an unreliable sign.
Stability and Change in State Electorates, Carter through Clinton
Erikson, Robert S., Wright, Gerald C., McIver, John P., Holian, David B.
Submitted: 2000-04-25
Keywords: partisanship, political ideology, public opinion, state electorates
Abstract: (click to show/hide) This paper extends the time series and advances the argument presented in _Statehouse Democracy_, which provided a public opinion basis for the study of state politics. The analysis covers the dynamics of partisanship and ideology in state electorates from 1977 through 1999. Incorporating the Bush and Clinton years allows for a number of conclusions. In the aggregate, state partisanship changed over the course of the last two presidential administrations, but state ideology did not. However, this change was not uniform across the country, but differed by region and resulted in higher levels of polarization between party and ideological identifications. Finally, consistent with the findings in _Statehouse Democracy_, state partisanship and
Erikson, Robert S., Wright, Gerald C., McIver, John P., Holian, David B.
Submitted: 2000-04-25
Keywords: partisanship, political ideology, public opinion, state electorates
Abstract: (click to show/hide) This paper extends the time series and advances the argument presented in _Statehouse Democracy_, which provided a public opinion basis for the study of state politics. The analysis covers the dynamics of partisanship and ideology in state electorates from 1977 through 1999. Incorporating the Bush and Clinton years allows for a number of conclusions. In the aggregate, state partisanship changed over the course of the last two presidential administrations, but state ideology did not. However, this change was not uniform across the country, but differed by region and resulted in higher levels of polarization between party and ideological identifications. Finally, consistent with the findings in _Statehouse Democracy_, state partisanship and
Time Remembered: A Dynamic Model of Interstate Interaction
Crescenzi, Mark J. C., Enterline, Andrew J.
Submitted: 2000-05-23
Keywords: dynamic model, growth, decay, cooperation, conflict
Abstract: (click to show/hide) Over time, states form relationships. These relationships, mosaics of past interactions, provide political leaders with information about how states are likely to behave in the future. Although simple, this claim holds important implications for the manner in which we construct and test empirically our expectations about interstate behavior. Empirical analyses of interstate relations implicitly assume that the units of analysis, principally dyad-years, are independent. Formal models of interstate interaction are often cast in the absence of historical context. In the following paper, we construct a dynamic model of interstate interaction that we believe will assist scholars employing empirical and formal methods by incorporating history into their models of interstate relations. Our conceptual model includes both conflictual and cooperative components, and exhibits the basic properties of growth and decay indicative of a dyadic behavioral history. In an empirical exposition, we derive a continuous measure of interstate conflict from the conflictual component of the model. Turning to Oneal and Russett's (1997) analysis of dyadic conflict for the period 1950-85 as a benchmark, we examine whether the inclusion of our measure of interstate conflict significantly improves our ability to predict militarized conflict. We find empirical support for this hypothesis, indicating that our continuous measure of interstate conflict significantly augments a fully specified statistical model of dyadic militarized conflict. We conclude that our research underscores the considerable purchase gained by incorporating historical context into models of interstate relations.
Crescenzi, Mark J. C., Enterline, Andrew J.
Submitted: 2000-05-23
Keywords: dynamic model, growth, decay, cooperation, conflict
Abstract: (click to show/hide) Over time, states form relationships. These relationships, mosaics of past interactions, provide political leaders with information about how states are likely to behave in the future. Although simple, this claim holds important implications for the manner in which we construct and test empirically our expectations about interstate behavior. Empirical analyses of interstate relations implicitly assume that the units of analysis, principally dyad-years, are independent. Formal models of interstate interaction are often cast in the absence of historical context. In the following paper, we construct a dynamic model of interstate interaction that we believe will assist scholars employing empirical and formal methods by incorporating history into their models of interstate relations. Our conceptual model includes both conflictual and cooperative components, and exhibits the basic properties of growth and decay indicative of a dyadic behavioral history. In an empirical exposition, we derive a continuous measure of interstate conflict from the conflictual component of the model. Turning to Oneal and Russett's (1997) analysis of dyadic conflict for the period 1950-85 as a benchmark, we examine whether the inclusion of our measure of interstate conflict significantly improves our ability to predict militarized conflict. We find empirical support for this hypothesis, indicating that our continuous measure of interstate conflict significantly augments a fully specified statistical model of dyadic militarized conflict. We conclude that our research underscores the considerable purchase gained by incorporating historical context into models of interstate relations.
Supplement to 'Democracy and Markets: The Case of Exchange Rates'
Freeman, John R., Hays, Jude, Stix, Helmut
Submitted: 2000-06-30
Keywords: Markov switching model, specification testing
Abstract: (click to show/hide) This is a methods supplement to "Democracy and Markets: The Case of Exchange Rates," American Journal of Political Science July 2000.
Freeman, John R., Hays, Jude, Stix, Helmut
Submitted: 2000-06-30
Keywords: Markov switching model, specification testing
Abstract: (click to show/hide) This is a methods supplement to "Democracy and Markets: The Case of Exchange Rates," American Journal of Political Science July 2000.
A Frailty Model of Negatively Dependent Competing Risks
Gordon, Sanford C.
Submitted: 2000-07-02
Keywords: duration, competing risks, simulation, multivariate, cabinet survival
Abstract: (click to show/hide) "Competing Risks" is a term used to describe duration models in which an individual spell may terminate via more than one outcome. Numerous applications in political science exist: For example, the term of a cabinet may end either with or without an election; criminal investigations may terminate either with prosecution or abandonment of a case; wars persist until the loss or victory of the aggressor state. Analysts typically assume stochastic independence among risks. However, many political examples are characterized by negative risk dependence: A high hazard rate for termination via one risk implies a low rate for termination via another. Ignoring this dependence can potentially bias inference. This paper suggests a class of bivariate (i.e. two risk), negatively dependent competing risks models. Negative risk dependence enters through an individual-specific random effect that simultaneously increases the hazard rate for one risk while decreasing the hazard for the second. Monte Carlo simulation reveals this specification to be superior to a naive model in which risks are assumed independent. Finally, I examine an application of the negative dependence model using Strom's (1985) and King et. al.'s (1990) data on cabinet survival.
Gordon, Sanford C.
Submitted: 2000-07-02
Keywords: duration, competing risks, simulation, multivariate, cabinet survival
Abstract: (click to show/hide) "Competing Risks" is a term used to describe duration models in which an individual spell may terminate via more than one outcome. Numerous applications in political science exist: For example, the term of a cabinet may end either with or without an election; criminal investigations may terminate either with prosecution or abandonment of a case; wars persist until the loss or victory of the aggressor state. Analysts typically assume stochastic independence among risks. However, many political examples are characterized by negative risk dependence: A high hazard rate for termination via one risk implies a low rate for termination via another. Ignoring this dependence can potentially bias inference. This paper suggests a class of bivariate (i.e. two risk), negatively dependent competing risks models. Negative risk dependence enters through an individual-specific random effect that simultaneously increases the hazard rate for one risk while decreasing the hazard for the second. Monte Carlo simulation reveals this specification to be superior to a naive model in which risks are assumed independent. Finally, I examine an application of the negative dependence model using Strom's (1985) and King et. al.'s (1990) data on cabinet survival.
Representative Bureaucracy and Harder Questions: A Response to Meier, Wrinkle, and Polinard
Nielsen, Laura B., Wolf, Patrick J.
Submitted: 2000-07-10
Keywords: representative bureaucracy, minorities, education policy, model specification, diagnostics, organizaional outputs
Abstract: (click to show/hide) In a recently published article, Meier, Wrinkle, and Polinard (1999) reach the tantalizing conclusion that increases in the representation of minority teachers in the public school bureaucracy actually enhance the academic achievement of both minority and Anglo groups of students. However, diagnostic and statistical tests on their data suggest that their analysis may suffer from specification, selection, and categorization limitations. When corrections for these problems are introduced into the analysis, the results that are the basis for the Meier, Wrinkle and Polinard conclusions change significantly, thereby undermining our confidence in the validity of
Nielsen, Laura B., Wolf, Patrick J.
Submitted: 2000-07-10
Keywords: representative bureaucracy, minorities, education policy, model specification, diagnostics, organizaional outputs
Abstract: (click to show/hide) In a recently published article, Meier, Wrinkle, and Polinard (1999) reach the tantalizing conclusion that increases in the representation of minority teachers in the public school bureaucracy actually enhance the academic achievement of both minority and Anglo groups of students. However, diagnostic and statistical tests on their data suggest that their analysis may suffer from specification, selection, and categorization limitations. When corrections for these problems are introduced into the analysis, the results that are the basis for the Meier, Wrinkle and Polinard conclusions change significantly, thereby undermining our confidence in the validity of
Two-stage approaches to regression models in which the dependent variable is based on estimates
Lewis, Jeffrey B.
Submitted: 2000-07-12
Keywords: heteroscedasticity, weighted least squares, root-n weighting, politics of race
Abstract: (click to show/hide) Researchers often use as dependent variables quantities estimated from auxiliary data sets. Estimated dependent variables (EDV) models arise, for example, in studies where counties or states are the units of analysis and the dependent variable is an estimated mean or fraction. A new source of such EDV regressions has been created by King's ecological inference estimator \cite{King:1997}. Researchers have fit regression models to quantities such as percent minority turnout that were estimated using King's EI \cite{Gay:1998}. Scholars fitting EDV models have generally recognized that variation in the sampling variance of the observations on the dependent variable will induce heteroscedasticity. In this paper, I show that the most common approach to this problem, weighting the regression by the inverses of the sampling standard errors of the dependent variable, will usually lead to inefficient estimates and underestimated standard errors. I show that the degree of this inefficiency and overconfidence can be very large. I also suggest two alternative approaches that are simple to implement and more efficient and yield consistent standard error estimates.
Lewis, Jeffrey B.
Submitted: 2000-07-12
Keywords: heteroscedasticity, weighted least squares, root-n weighting, politics of race
Abstract: (click to show/hide) Researchers often use as dependent variables quantities estimated from auxiliary data sets. Estimated dependent variables (EDV) models arise, for example, in studies where counties or states are the units of analysis and the dependent variable is an estimated mean or fraction. A new source of such EDV regressions has been created by King's ecological inference estimator \cite{King:1997}. Researchers have fit regression models to quantities such as percent minority turnout that were estimated using King's EI \cite{Gay:1998}. Scholars fitting EDV models have generally recognized that variation in the sampling variance of the observations on the dependent variable will induce heteroscedasticity. In this paper, I show that the most common approach to this problem, weighting the regression by the inverses of the sampling standard errors of the dependent variable, will usually lead to inefficient estimates and underestimated standard errors. I show that the degree of this inefficiency and overconfidence can be very large. I also suggest two alternative approaches that are simple to implement and more efficient and yield consistent standard error estimates.
A Specification Test for Linear Regressions that use King-Based Ecological Inference Point Estimates as Dependent Variables
Herron, Michael C., Shotts, Kenneth W.
Submitted: 2000-07-14
Keywords: ecological inference, second stage regressions, ordinary least squares, logical consistency
Abstract: (click to show/hide) Many researchers use point estimates produced by the King (1997) ecological inference technique as dependent variables in second stage linear regressions. We show, however, that this two stage procedure is at risk of logical inconsistency. Namely, the assumptions necessary to support the procedure's first stage (ecological inference via King's method) can be incompatible with the assumptions supporting the second (linear regression). We derive a specification test for logical consistency of the two stage procedure and describe options available to a researcher whose ecological dataset fails the test.
Herron, Michael C., Shotts, Kenneth W.
Submitted: 2000-07-14
Keywords: ecological inference, second stage regressions, ordinary least squares, logical consistency
Abstract: (click to show/hide) Many researchers use point estimates produced by the King (1997) ecological inference technique as dependent variables in second stage linear regressions. We show, however, that this two stage procedure is at risk of logical inconsistency. Namely, the assumptions necessary to support the procedure's first stage (ecological inference via King's method) can be incompatible with the assumptions supporting the second (linear regression). We derive a specification test for logical consistency of the two stage procedure and describe options available to a researcher whose ecological dataset fails the test.
Using Ecological Inference Point Estimates in Second Stage Linear Regressions
Herron, Michael C., Shotts, Kenneth W.
Submitted: 2000-07-14
Keywords: ecological inference, second stage regressions, ordinary least squares, consistency
Abstract: (click to show/hide) The practice of using point estimates produced by the King (1997) ecological inference technique in second stage linear regressions leads to second stage results that, in general, are inconsistent. This conclusion holds, notably, even when all the assumptions behind King's ecological technique are satisfied. Second stage inconsistency is a consequence of the fact that King--based point estimates of disaggregated quantities are themselves inconsistent, and, moreover, these point estimates are contaminated by errors correlated with the true quantities the estimates measure. Our findings on second stage inconsistency follow from econometric theory in conjunction with an analysis of simulated and real ecological datasets, and based on the findings we propose a bootstrap that researchers can use to produce consistent second stage estimates and valid confidence intervals.
Herron, Michael C., Shotts, Kenneth W.
Submitted: 2000-07-14
Keywords: ecological inference, second stage regressions, ordinary least squares, consistency
Abstract: (click to show/hide) The practice of using point estimates produced by the King (1997) ecological inference technique in second stage linear regressions leads to second stage results that, in general, are inconsistent. This conclusion holds, notably, even when all the assumptions behind King's ecological technique are satisfied. Second stage inconsistency is a consequence of the fact that King--based point estimates of disaggregated quantities are themselves inconsistent, and, moreover, these point estimates are contaminated by errors correlated with the true quantities the estimates measure. Our findings on second stage inconsistency follow from econometric theory in conjunction with an analysis of simulated and real ecological datasets, and based on the findings we propose a bootstrap that researchers can use to produce consistent second stage estimates and valid confidence intervals.
Why Lagged Dependent Variables Can Suppress the Explanatory Power of Other Independent Variables
Achen, Christopher H.
Submitted: 2000-07-14
Keywords: time series, autoregressive, lags, serial correlation, budgets, arms races
Abstract: (click to show/hide) In many time series applications in the social sciences, lagged dependent variables have no obvious causal interpretation, and researchers omit them. When they are left out, the other coefficients take on sensible values. However, when an autoregressive term is put in ``as a control,'' it often acquires a large, statistically significant coefficient and improves the fit dramatically, while many or all of the remaining substantive coefficients collapse to implausibly small and insignificant values. Occasionally, the substantive variables even take on the wrong sign. This paper explains why this phenomenon occurs and how the resulting confusions have often misled researchers into inaccurate inferences. The standard findings that government budgets are caused primarily by past budgets and that arms races are driven mainly by domestic forces are shown to be likely statistical artifacts. Applications are made to vector autoregressions, error-correction models, and panel studies.
Achen, Christopher H.
Submitted: 2000-07-14
Keywords: time series, autoregressive, lags, serial correlation, budgets, arms races
Abstract: (click to show/hide) In many time series applications in the social sciences, lagged dependent variables have no obvious causal interpretation, and researchers omit them. When they are left out, the other coefficients take on sensible values. However, when an autoregressive term is put in ``as a control,'' it often acquires a large, statistically significant coefficient and improves the fit dramatically, while many or all of the remaining substantive coefficients collapse to implausibly small and insignificant values. Occasionally, the substantive variables even take on the wrong sign. This paper explains why this phenomenon occurs and how the resulting confusions have often misled researchers into inaccurate inferences. The standard findings that government budgets are caused primarily by past budgets and that arms races are driven mainly by domestic forces are shown to be likely statistical artifacts. Applications are made to vector autoregressions, error-correction models, and panel studies.
What to Do (and Not Do) With Dynamic Panel Data in Political Science (with apologies to Beck and Katz)
Wawro, Gregory
Submitted: 2000-07-16
Keywords: dynamic panel data models, lagged endogenous variables, GMM estimators, party identification, campaign finance
Abstract: (click to show/hide) Panel data is a very valuable resource for finding empirical solutions to political science puzzles. Yet numerous published studies in political science that use panel data to estimate models with dynamics have failed to take into account important estimation issues which call into question the inferences we can make from these analyses. Simply put, the failure to account explicitly for unobserved individual effects in panel data leads to inconsistent estimates of parameters of interest. The fundamental requirement for consistency of parameter estimates---that the explanatory variables in a regression equation must be uncorrleated with the disturbance term---is not met unless individual specific effects are adequately accounted for. Dynamic panel data estimators that eliminate this problem have become fairly standard in the economics literature. The purpose of this paper is to introduce these methods to political scientists. First, I show how the problem of inconsistency arises in dynamic panel data. I then show how to correct for this problem using generalized method of moments (GMM) estimators. I then demonstrate the usefulness of these methods with replications of published analyses.
Wawro, Gregory
Submitted: 2000-07-16
Keywords: dynamic panel data models, lagged endogenous variables, GMM estimators, party identification, campaign finance
Abstract: (click to show/hide) Panel data is a very valuable resource for finding empirical solutions to political science puzzles. Yet numerous published studies in political science that use panel data to estimate models with dynamics have failed to take into account important estimation issues which call into question the inferences we can make from these analyses. Simply put, the failure to account explicitly for unobserved individual effects in panel data leads to inconsistent estimates of parameters of interest. The fundamental requirement for consistency of parameter estimates---that the explanatory variables in a regression equation must be uncorrleated with the disturbance term---is not met unless individual specific effects are adequately accounted for. Dynamic panel data estimators that eliminate this problem have become fairly standard in the economics literature. The purpose of this paper is to introduce these methods to political scientists. First, I show how the problem of inconsistency arises in dynamic panel data. I then show how to correct for this problem using generalized method of moments (GMM) estimators. I then demonstrate the usefulness of these methods with replications of published analyses.
Application of Panel Data Analysis to Kramer's Economic Voting Problem
Yoon, David
Submitted: 2000-07-16
Keywords: economic voting, panel data
Abstract: (click to show/hide) Although the health of a nation's economy has come to be seen as a reliable predictor of election outcome at the national level (e.g., Fair 1978, 1988), the corollary link between economic conditions and electoral behavior at the individual level remains less clear. Kinder and Kiewiet (1979) concluded that while the ups and downs of personal finances had negligible effect on an individual's voting behavior in national elections, the trajectory of the national economy had a significant effect. The hypothesis of the ``sociotropic'' voter was to be preferred over the ``pocketbook'' voter in thinking about whose economy mattered in elections. In an influential critique, Kramer (1983) argued that such a conclusion could not be drawn from purely cross-sectional survey data (data type used by Kinder and Kiewiet). According to Kramer, only the analysis of aggregate-level time-series data provide unbiased estimates of the effects of economic conditions on votes. Unfortunately, the two main competing hypotheses cannot be tested since individual-level economic factors cannot be studied with aggregate-level time series data alone. In contrast to previous analyses, I employ panel data (also known as longitudinal data) and analytical methods sensitive to the individual-level time-series structure of the data to estimate the relative magnitudes of the sociotropic and pocketbook effects, and test the merits of the respective hypotheses. Others have attempted to solve the Kramer problem by pooling cross-sectional data (e.g., Markus (1988, 1992)). Although pooled cross-sectional data allow investigators to compare sociotropic and pocketbook effects, they suffer from many of the same shortcomings of purely cross-sectional data. I use the 1993-1996 NES panel study to demonstrate the robustness of the sociotropic model and the strengths of panel analysis. I explain the battery of tests, estimators, and statistical assumptions used and relate these in detail to prevalent substantive political assumptions. And finally an uncommonly long panel from an Italian Nielsen survey is analyzed to demonstrate the utility of such
Yoon, David
Submitted: 2000-07-16
Keywords: economic voting, panel data
Abstract: (click to show/hide) Although the health of a nation's economy has come to be seen as a reliable predictor of election outcome at the national level (e.g., Fair 1978, 1988), the corollary link between economic conditions and electoral behavior at the individual level remains less clear. Kinder and Kiewiet (1979) concluded that while the ups and downs of personal finances had negligible effect on an individual's voting behavior in national elections, the trajectory of the national economy had a significant effect. The hypothesis of the ``sociotropic'' voter was to be preferred over the ``pocketbook'' voter in thinking about whose economy mattered in elections. In an influential critique, Kramer (1983) argued that such a conclusion could not be drawn from purely cross-sectional survey data (data type used by Kinder and Kiewiet). According to Kramer, only the analysis of aggregate-level time-series data provide unbiased estimates of the effects of economic conditions on votes. Unfortunately, the two main competing hypotheses cannot be tested since individual-level economic factors cannot be studied with aggregate-level time series data alone. In contrast to previous analyses, I employ panel data (also known as longitudinal data) and analytical methods sensitive to the individual-level time-series structure of the data to estimate the relative magnitudes of the sociotropic and pocketbook effects, and test the merits of the respective hypotheses. Others have attempted to solve the Kramer problem by pooling cross-sectional data (e.g., Markus (1988, 1992)). Although pooled cross-sectional data allow investigators to compare sociotropic and pocketbook effects, they suffer from many of the same shortcomings of purely cross-sectional data. I use the 1993-1996 NES panel study to demonstrate the robustness of the sociotropic model and the strengths of panel analysis. I explain the battery of tests, estimators, and statistical assumptions used and relate these in detail to prevalent substantive political assumptions. And finally an uncommonly long panel from an Italian Nielsen survey is analyzed to demonstrate the utility of such
Legislator Quality and Campaign Contributions
Mebane, Walter R., Ratkovic, Marc T., Tofias, Michael W.
Submitted: 2000-07-18
Keywords: campaign contributions, U.S. House of Representatives, constrained nonlinear optimization, two-limit tobit, 1992 election, telecommunications PACs, political action committees
Abstract: (click to show/hide) We introduce a simple theoretical model of the relationship between the campaign contributions a legislator receives from a PAC and the amount of ``service'' the legislator provides to the PAC, a key assumption being that the marginal cost of service decreases as the quality of the legislator increases. Optimal solution of the constrained optimization problem that each PAC faces in allocating its campaign contributions among legislators implies a conditional two-limit tobit model for the relationship between contributions and aspects of the quality of each legislator. The constraints arise because PAC contributions must be positive but no greater than a legally limiting value and because each PAC's budget for contributions is finite. We extend the tobit model to support pooling data >from several similar PACs. We estimate the empirical model using data from the U.S.\ House of Representatives. The fact that optimal PAC behavior implies censoring suggests that it is usually inappropriate to aggregate contributions from different PACs; but pooling can work well.
Mebane, Walter R., Ratkovic, Marc T., Tofias, Michael W.
Submitted: 2000-07-18
Keywords: campaign contributions, U.S. House of Representatives, constrained nonlinear optimization, two-limit tobit, 1992 election, telecommunications PACs, political action committees
Abstract: (click to show/hide) We introduce a simple theoretical model of the relationship between the campaign contributions a legislator receives from a PAC and the amount of ``service'' the legislator provides to the PAC, a key assumption being that the marginal cost of service decreases as the quality of the legislator increases. Optimal solution of the constrained optimization problem that each PAC faces in allocating its campaign contributions among legislators implies a conditional two-limit tobit model for the relationship between contributions and aspects of the quality of each legislator. The constraints arise because PAC contributions must be positive but no greater than a legally limiting value and because each PAC's budget for contributions is finite. We extend the tobit model to support pooling data >from several similar PACs. We estimate the empirical model using data from the U.S.\ House of Representatives. The fact that optimal PAC behavior implies censoring suggests that it is usually inappropriate to aggregate contributions from different PACs; but pooling can work well.
A Note Relating Ideal Point Estimates to the Spatial Model
Clinton, Joshua, Meirowitz, Adam
Submitted: 2000-07-18
Keywords: ideal points, preference estimation, NOMINATE, spatial model
Abstract: (click to show/hide) Existing preference estimators do not incorporate the full structure of the spatial model. Specifically, they fail to use the sequential nature of the agenda by not constraining the nay location of a bill to be the yea location of the last successful policy. The consequences of this omission may be far-reaching. Not only is information useful for the identification of the model neglected, but more seriously, the dimensionality of the policy space may be incorrectly estimated. Preference and bill location estimates are uninterpretable in terms of the spatial model. We show that under very general assumptions, ML estimates of ideal points that do not constrain the nay locations will differ from ML estimates that constrain the nay locatios -- a difference that does not vanish as the numbers of votes goes to infinity. Additionally, unconstrained models underestimate the true dimensionality of the policy space. We derive a Maximum Likelihood estimator of legislative preferences and bill locations that shares basic assumptions with the spatial model of voting.
Clinton, Joshua, Meirowitz, Adam
Submitted: 2000-07-18
Keywords: ideal points, preference estimation, NOMINATE, spatial model
Abstract: (click to show/hide) Existing preference estimators do not incorporate the full structure of the spatial model. Specifically, they fail to use the sequential nature of the agenda by not constraining the nay location of a bill to be the yea location of the last successful policy. The consequences of this omission may be far-reaching. Not only is information useful for the identification of the model neglected, but more seriously, the dimensionality of the policy space may be incorrectly estimated. Preference and bill location estimates are uninterpretable in terms of the spatial model. We show that under very general assumptions, ML estimates of ideal points that do not constrain the nay locations will differ from ML estimates that constrain the nay locatios -- a difference that does not vanish as the numbers of votes goes to infinity. Additionally, unconstrained models underestimate the true dimensionality of the policy space. We derive a Maximum Likelihood estimator of legislative preferences and bill locations that shares basic assumptions with the spatial model of voting.
A System Model of American Macro Politics
Erikson, Robert S., MacKuen, Michael, Stimson, James
Submitted: 2000-07-18
Keywords: macro, system, simulation, American
Abstract: (click to show/hide) The paper develops a single macro-level model of American politics based upon the research program of The Macro Polity, (Cambridge, in press). The multi-equation simulation model is constructed from individual regressions on core elements such as Presidential Approval, Macropartisanship, Public Policy Mood, Elections (House, Senate, and Presidency), Policy Activity (House, Senate, and Presidency), Policy (laws) and the lagged feedbacks involved in the full structure of relationships. (Economic performance indicators are also both caused by and causes of the political variables.) In trial analyses, not implementing the full power of simulation methods, the model is tweaked to examine the impact of changing historical outcomes to observe system behavior. Two of them are a shock to unemployment and reversing the outcome of the 1980 presidential election. These are seen to produce unexpected outcomes when the full range of interconnections is allowed to work.
Erikson, Robert S., MacKuen, Michael, Stimson, James
Submitted: 2000-07-18
Keywords: macro, system, simulation, American
Abstract: (click to show/hide) The paper develops a single macro-level model of American politics based upon the research program of The Macro Polity, (Cambridge, in press). The multi-equation simulation model is constructed from individual regressions on core elements such as Presidential Approval, Macropartisanship, Public Policy Mood, Elections (House, Senate, and Presidency), Policy Activity (House, Senate, and Presidency), Policy (laws) and the lagged feedbacks involved in the full structure of relationships. (Economic performance indicators are also both caused by and causes of the political variables.) In trial analyses, not implementing the full power of simulation methods, the model is tweaked to examine the impact of changing historical outcomes to observe system behavior. Two of them are a shock to unemployment and reversing the outcome of the 1980 presidential election. These are seen to produce unexpected outcomes when the full range of interconnections is allowed to work.
Bayesian and Frequentist Inference for Ecological Inference: The RxC Case
Rosen, Ori, King, Gary, Jiang, Wenxin, Tanner, Martin A.
Submitted: 2000-07-25
Keywords: EM, MCMC, least squares, ecological inference, Bayes, Nazi
Abstract: (click to show/hide) In this paper we propose Bayesian and frequentist approaches to ecological inference, based on RxC contingency tables, including a covariate. The proposed Bayesian model extends the binomial-beta hierarchical model developed by King, Rosen and Tanner (1999) from the 2x2 case to the RxC case. As in the 2x2 case, the inferential procedure employs Markov chain Monte Carlo (MCMC) methods. As such, the resulting MCMC analysis is rich but computationally intensive. The frequentist approach, based on first moments rather than on the entire likelihood, provides quick inference via nonlinear least-squares, while retaining good frequentist properties. The two approaches are illustrated with simulated data, as well as with real data on voting patterns in Weimar Germany. In the final section of the paper we provide an overview of a range of alternative inferential approaches which trade-off computational intensity for statistical efficiency.
Rosen, Ori, King, Gary, Jiang, Wenxin, Tanner, Martin A.
Submitted: 2000-07-25
Keywords: EM, MCMC, least squares, ecological inference, Bayes, Nazi
Abstract: (click to show/hide) In this paper we propose Bayesian and frequentist approaches to ecological inference, based on RxC contingency tables, including a covariate. The proposed Bayesian model extends the binomial-beta hierarchical model developed by King, Rosen and Tanner (1999) from the 2x2 case to the RxC case. As in the 2x2 case, the inferential procedure employs Markov chain Monte Carlo (MCMC) methods. As such, the resulting MCMC analysis is rich but computationally intensive. The frequentist approach, based on first moments rather than on the entire likelihood, provides quick inference via nonlinear least-squares, while retaining good frequentist properties. The two approaches are illustrated with simulated data, as well as with real data on voting patterns in Weimar Germany. In the final section of the paper we provide an overview of a range of alternative inferential approaches which trade-off computational intensity for statistical efficiency.
Is There a Gender Gap in Fiscal Political Preferences
Alvarez, R. Michael, McCaffery, Edward J.
Submitted: 2000-08-12
Keywords: Gender gap, fiscal politics, taxation, budget surplus, multinomial logit, missing data, imputation, framing, survey experiments
Abstract: (click to show/hide) This paper examines the relationship between attitudes on potential uses of the budget surplus and gender. Survey results show relatively weak support overall for using a projected surplus to reduce taxes, with respondents much likelier to prefer increased social spending on education or social security. There is a significant gender gap with men being far more likely than women to support tax cuts or paying down the national debt. Given a menu of particular types of tax cuts, women are marginally more likely to favor child-care relief or working poor tax credits whereas men are marginally more likely to favor capital gains reduction or tax rate cuts. When primed that the tax laws are biased against two-worker families, men significantly change their preferences, moving from support for general tax rate cuts to support for working poor tax relief, but not to child-care relief. One of the strongest results to emerge is that women are far more likely than men not to express an opinion or to confess ignorance about fiscal matters. Both genders increase their ``no opinion'' answer in the face of priming, but men more so than women. Further research will explore this no opinion/uncertainty aspect.
Alvarez, R. Michael, McCaffery, Edward J.
Submitted: 2000-08-12
Keywords: Gender gap, fiscal politics, taxation, budget surplus, multinomial logit, missing data, imputation, framing, survey experiments
Abstract: (click to show/hide) This paper examines the relationship between attitudes on potential uses of the budget surplus and gender. Survey results show relatively weak support overall for using a projected surplus to reduce taxes, with respondents much likelier to prefer increased social spending on education or social security. There is a significant gender gap with men being far more likely than women to support tax cuts or paying down the national debt. Given a menu of particular types of tax cuts, women are marginally more likely to favor child-care relief or working poor tax credits whereas men are marginally more likely to favor capital gains reduction or tax rate cuts. When primed that the tax laws are biased against two-worker families, men significantly change their preferences, moving from support for general tax rate cuts to support for working poor tax relief, but not to child-care relief. One of the strongest results to emerge is that women are far more likely than men not to express an opinion or to confess ignorance about fiscal matters. Both genders increase their ``no opinion'' answer in the face of priming, but men more so than women. Further research will explore this no opinion/uncertainty aspect.
A Specification Test for Linear Regressions that use King-Based Ecological Inference Point Estimates as Dependent Variables
Herron, Michael C., Shotts, Kenneth W.
Submitted: 2000-08-16
Keywords: ecological inference, second stage regressions, ordinary least squares, logical consistency
Abstract: (click to show/hide) Many researchers use point estimates produced by the King (1997) ecological inference technique as dependent variables in second stage linear regressions. We show, however, that this two stage procedure is at risk of logical inconsistency. Namely, the assumptions necessary to support the procedure's first stage (ecological inference via King's method) can be incompatible with the assumptions supporting the second (linear regression). We derive a specification test for logical consistency of the two stage procedure and describe options available to a researcher whose ecological dataset fails the test.
Herron, Michael C., Shotts, Kenneth W.
Submitted: 2000-08-16
Keywords: ecological inference, second stage regressions, ordinary least squares, logical consistency
Abstract: (click to show/hide) Many researchers use point estimates produced by the King (1997) ecological inference technique as dependent variables in second stage linear regressions. We show, however, that this two stage procedure is at risk of logical inconsistency. Namely, the assumptions necessary to support the procedure's first stage (ecological inference via King's method) can be incompatible with the assumptions supporting the second (linear regression). We derive a specification test for logical consistency of the two stage procedure and describe options available to a researcher whose ecological dataset fails the test.
Sink or Swim:" What Happened to California's Bilingual Students after Proposition 227?
Bali, Valentina A.
Submitted: 2000-08-21
Keywords: initiative, bilingual education, California, Proposition 227, Heckman selection
Abstract: (click to show/hide) Proposition 227, passed in California in 1998, aimed to dismantle bilingual programs in the state's public schools. Using individual level data from a southern California school district, I find that in 1998, before Proposition 227, limited-English-proficient (LEP) students enrolled in bilingual classes had lower scores in reading than LEP students not enrolled in bilingual classes: 2.4 points less on a scale from 1 to 99. In math these bilingual students scored 0.5 points higher than non-bilingual LEPs. But in 1999, after Proposition 227 the same set of students had scores no worse than non-bilingual LEP students in reading and were still 0.5 points higher in math. Proposition 227, which interrupted bilingual programs early and emphasized English instruction, then, did not set bilingual students back relative to non-bilingual LEP and may have even benefitted them.
Bali, Valentina A.
Submitted: 2000-08-21
Keywords: initiative, bilingual education, California, Proposition 227, Heckman selection
Abstract: (click to show/hide) Proposition 227, passed in California in 1998, aimed to dismantle bilingual programs in the state's public schools. Using individual level data from a southern California school district, I find that in 1998, before Proposition 227, limited-English-proficient (LEP) students enrolled in bilingual classes had lower scores in reading than LEP students not enrolled in bilingual classes: 2.4 points less on a scale from 1 to 99. In math these bilingual students scored 0.5 points higher than non-bilingual LEPs. But in 1999, after Proposition 227 the same set of students had scores no worse than non-bilingual LEP students in reading and were still 0.5 points higher in math. Proposition 227, which interrupted bilingual programs early and emphasized English instruction, then, did not set bilingual students back relative to non-bilingual LEP and may have even benefitted them.
Ticket-splitting and Strategic Voting
Gschwend, Thomas
Submitted: 2000-08-22
Keywords: Ticket Splitting, Strategic voting, Germany EI, Multiple imputation
Abstract: (click to show/hide) Germany provides an especially interesting case for the study of strategic voting because a two-ballot system is used. Voters are encouraged to split their votes using different strategies. I disentangle different types of strategic voting that have been mixed in the literature so far: On the first vote there is \emph{tactical} voting, and on the second vote there is \emph{loan} voting. Therefore, I focus particularly on ticket splitting patterns. The data set I use contains official election results of first and second votes for all German districts from the federal election of 1998. To obtain estimates that determines quantity of straight and split ticket voting between political parties I employ King's EI for a first-stage analysis and use these estimates as independent variables in second-stage models. In order to account for the uncertainty in first-stage EI-point estimates I use a multiple imputation approach. I show that tactical and loan voters secured the representation of FDP and the Greens in the German Parliament. Several validation attempts of the second-stage prediction results prove that not every second-stage analysis based on first stage EI-point estimates is doomed to fail.
Gschwend, Thomas
Submitted: 2000-08-22
Keywords: Ticket Splitting, Strategic voting, Germany EI, Multiple imputation
Abstract: (click to show/hide) Germany provides an especially interesting case for the study of strategic voting because a two-ballot system is used. Voters are encouraged to split their votes using different strategies. I disentangle different types of strategic voting that have been mixed in the literature so far: On the first vote there is \emph{tactical} voting, and on the second vote there is \emph{loan} voting. Therefore, I focus particularly on ticket splitting patterns. The data set I use contains official election results of first and second votes for all German districts from the federal election of 1998. To obtain estimates that determines quantity of straight and split ticket voting between political parties I employ King's EI for a first-stage analysis and use these estimates as independent variables in second-stage models. In order to account for the uncertainty in first-stage EI-point estimates I use a multiple imputation approach. I show that tactical and loan voters secured the representation of FDP and the Greens in the German Parliament. Several validation attempts of the second-stage prediction results prove that not every second-stage analysis based on first stage EI-point estimates is doomed to fail.
A Practical Statistical Model for Multiparty Electoral Data
Honaker, James, Katz, Jonathan, King, Gary
Submitted: 2000-08-23
Keywords: compositional data, multiparty electoral data, EM algorithms
Abstract: (click to show/hide) Katz and King (1999) develop a model for predicting or explaining aggregate electoral results in multiparty democracies. This model is, in principle, analogous to what least squares regression provides American politics researchers in that two-party system. Katz and King applied this model to three-party elections in England and revealed a variety of new features of incumbency advantage and where each party pulls support from. Although the mathematics of their statistical model covers any number of political parties, it is computationally very demanding, and hence slow and numerically imprecise, with more than three. The original goal of our work was to produce an approximate method that works quicker in practice with many parties without making too many theoretical compromises. As it turns out, the method we offer here improves on Katz and King's (in bias, variance, numerical stability, and computational speed) even when the latter is computationally feasible. We also offer easy-to-use software that implements our suggestions.
Honaker, James, Katz, Jonathan, King, Gary
Submitted: 2000-08-23
Keywords: compositional data, multiparty electoral data, EM algorithms
Abstract: (click to show/hide) Katz and King (1999) develop a model for predicting or explaining aggregate electoral results in multiparty democracies. This model is, in principle, analogous to what least squares regression provides American politics researchers in that two-party system. Katz and King applied this model to three-party elections in England and revealed a variety of new features of incumbency advantage and where each party pulls support from. Although the mathematics of their statistical model covers any number of political parties, it is computationally very demanding, and hence slow and numerically imprecise, with more than three. The original goal of our work was to produce an approximate method that works quicker in practice with many parties without making too many theoretical compromises. As it turns out, the method we offer here improves on Katz and King's (in bias, variance, numerical stability, and computational speed) even when the latter is computationally feasible. We also offer easy-to-use software that implements our suggestions.
Forecasting Conflict in the Balkans using Hidden Markov Models
Schrodt, Philip A.
Submitted: 2000-08-24
Keywords: forecasting, event data, hidden Markov models, conflict, Balkans, Yugoslavia
Abstract: (click to show/hide) This study uses hidden Markov models (HMM) to forecast conflict in the former Yugoslavia for the period January 1991 through January 1999. The political and military events reported in the lead sentences of Reuters news service stories were coded into the World Events Interaction Survey (WEIS) event data scheme. The forecasting scheme involved randomly selecting eight 100-event "templates" taken at a 1-, 3- or 6-month forecasting lag for high-conflict and low-conflict weeks. A separate HMM is developed for the high-conflict-week sequences and the low-conflict-week sequences. Forecasting is done by determining whether a sequence of observed events fit the high-conflict or low-conflict model with higher probability. Models were selected to maximize the difference between correct and incorrect predictions, evaluated by week. Three weighting schemes were used: unweighted (U), penalize false positives (P) and penalize false negatives (N). There is a relatively high level of convergence in the estimates‹the best and worst models of a given type vary in accuracy by only about 15% to 20%. In full-sample tests, the U and P models produce at overall accuracy of around 80%. However, these models correctly forecast only about 25% of the high-conflict weeks, although about 60% of the cases where a high-conflict week has been forecast turn out to have high conflict. In contrast, the N model has an overall accuracy of only about 50% in full-sample tests, but it correctly forecasts high-conflict weeks with 85% accuracy in the 3- and 6-month horizon and 92% accuracy in the 1-month horizon. However, this is achieved by excessive predictions of high-conflict weeks: only about 30% of the cases where a high-conflict week has been forecast are high-conflict. Models that use templates from only the previous year usually do about as well as models based on the entire sample. The models are remarkably insensitive to the length of the forecasting horizon‹the drop-off in accuracy at longer forecasting horizons is very small, typically around 2%-4%. There is also no clear difference in the estimated coefficients for the 1-month and 6-month models. An extensive analysis was done of the coefficient estimates in the full-sample model to determine what the model was "looking at" in order to make predictions. While a number of statistically significant differences exist between the high and low conflict models, these do not fall into any neat patterns. This is probably due to a combination of the large number of parameters being estimated, the multiple local maxima in the estimation surface, and the complications introduced by the presence of a number of very low probability event categories. Some experiments with simplified models indicate that it is possible to use models with substantially fewer parameters without markedly decreasing the accuracy of the predictions; in fact predictions of the high conflict periods actually increase in accuracy quite substantially.
Schrodt, Philip A.
Submitted: 2000-08-24
Keywords: forecasting, event data, hidden Markov models, conflict, Balkans, Yugoslavia
Abstract: (click to show/hide) This study uses hidden Markov models (HMM) to forecast conflict in the former Yugoslavia for the period January 1991 through January 1999. The political and military events reported in the lead sentences of Reuters news service stories were coded into the World Events Interaction Survey (WEIS) event data scheme. The forecasting scheme involved randomly selecting eight 100-event "templates" taken at a 1-, 3- or 6-month forecasting lag for high-conflict and low-conflict weeks. A separate HMM is developed for the high-conflict-week sequences and the low-conflict-week sequences. Forecasting is done by determining whether a sequence of observed events fit the high-conflict or low-conflict model with higher probability. Models were selected to maximize the difference between correct and incorrect predictions, evaluated by week. Three weighting schemes were used: unweighted (U), penalize false positives (P) and penalize false negatives (N). There is a relatively high level of convergence in the estimates‹the best and worst models of a given type vary in accuracy by only about 15% to 20%. In full-sample tests, the U and P models produce at overall accuracy of around 80%. However, these models correctly forecast only about 25% of the high-conflict weeks, although about 60% of the cases where a high-conflict week has been forecast turn out to have high conflict. In contrast, the N model has an overall accuracy of only about 50% in full-sample tests, but it correctly forecasts high-conflict weeks with 85% accuracy in the 3- and 6-month horizon and 92% accuracy in the 1-month horizon. However, this is achieved by excessive predictions of high-conflict weeks: only about 30% of the cases where a high-conflict week has been forecast are high-conflict. Models that use templates from only the previous year usually do about as well as models based on the entire sample. The models are remarkably insensitive to the length of the forecasting horizon‹the drop-off in accuracy at longer forecasting horizons is very small, typically around 2%-4%. There is also no clear difference in the estimated coefficients for the 1-month and 6-month models. An extensive analysis was done of the coefficient estimates in the full-sample model to determine what the model was "looking at" in order to make predictions. While a number of statistically significant differences exist between the high and low conflict models, these do not fall into any neat patterns. This is probably due to a combination of the large number of parameters being estimated, the multiple local maxima in the estimation surface, and the complications introduced by the presence of a number of very low probability event categories. Some experiments with simplified models indicate that it is possible to use models with substantially fewer parameters without markedly decreasing the accuracy of the predictions; in fact predictions of the high conflict periods actually increase in accuracy quite substantially.
Strategic voting in mixed-member electoral systems: The Italian case
Benoit, Kenneth, Laver, Michael, Giannetti, Daniela
Submitted: 2000-08-26
Keywords: elections, italy, strategic voting, ecological inference
Abstract: (click to show/hide) The new Italian electoral system has two elements, a plurality element in single member districts and a PR element in larger multimember constituencies. The plurality element provides strong incentives for groups of parties to form pre-electoral coalitions. The PR element offers incentives for parties to contest the elections individually. We can think of two types of voter. The first type, whom we characterize as "strategic," votes for his or her first choice party in the PR election since there is no strategy that can improve on this. In the plurality election, a strategic voter supports the candidate sponsored by the coalition with which his or her first choice party is affiliated, even if this is not from the first choice party. The second type of voter, whom we characterize as "non-strategic," also votes for his or her first choice party in the PR election. In the plurality election, the non-strategic voter will vote for a first choice party if a candidate of this party is on the ballot but, if not, votes unpredictably. In this paper, we model the "strategic" and "non- strategic" elements of the vote flowing to candidates in the plurality element of the election. Using data from the 1996 and 1994 elections on both PR and plurality voting patterns in each single member district, and confining ourselves to districts where there is a run-off between two coalitions, we are able to estimate the relative numbers of strategic and non- strategic voters in each district, and characterize this in terms of a range of strategic variables.
Benoit, Kenneth, Laver, Michael, Giannetti, Daniela
Submitted: 2000-08-26
Keywords: elections, italy, strategic voting, ecological inference
Abstract: (click to show/hide) The new Italian electoral system has two elements, a plurality element in single member districts and a PR element in larger multimember constituencies. The plurality element provides strong incentives for groups of parties to form pre-electoral coalitions. The PR element offers incentives for parties to contest the elections individually. We can think of two types of voter. The first type, whom we characterize as "strategic," votes for his or her first choice party in the PR election since there is no strategy that can improve on this. In the plurality election, a strategic voter supports the candidate sponsored by the coalition with which his or her first choice party is affiliated, even if this is not from the first choice party. The second type of voter, whom we characterize as "non-strategic," also votes for his or her first choice party in the PR election. In the plurality election, the non-strategic voter will vote for a first choice party if a candidate of this party is on the ballot but, if not, votes unpredictably. In this paper, we model the "strategic" and "non- strategic" elements of the vote flowing to candidates in the plurality element of the election. Using data from the 1996 and 1994 elections on both PR and plurality voting patterns in each single member district, and confining ourselves to districts where there is a run-off between two coalitions, we are able to estimate the relative numbers of strategic and non- strategic voters in each district, and characterize this in terms of a range of strategic variables.
Binding the Frame: How Important are Frames for Survey Response?
Alvarez, R. Michael, Brehm, John
Submitted: 2000-08-26
Keywords: framing, survey experiments, heteroskedastic choice
Abstract: (click to show/hide) [not transcribed]
Alvarez, R. Michael, Brehm, John
Submitted: 2000-08-26
Keywords: framing, survey experiments, heteroskedastic choice
Abstract: (click to show/hide) [not transcribed]
Aggregation Among Binary, Count, and Duration Models
King, Gary, Signorino, Curtis S., Alt, James E.
Submitted: 2000-08-28
Keywords: Duration, event count, binary, renewal process, aggregation
Abstract: (click to show/hide) Binary, count, and duration data all code discrete events occurring at points in time. Although a single data generation process can produce all of these three data types, the statistical literature is not very helpful in providing methods to estimate parameters of the same process from each. In fact, only a single theoretical process exists for which known statistical methods can estimate the same parameters --- and it is generally used only for count and duration data. The result is that seemingly trivial decisions about which level of data to use can have important consequences for substantive interpretations. We describe the theoretical event process for which results exist, based on time-independence. We also derive a set of models for a time-dependent process and compare their predictions to those of a commonly used model. Any hope of understanding and avoiding the more serious problems of aggregation bias in events data is contingent on first deriving a much wider arsenal of statistical models and theoretical processes that are not constrained by the particular forms of data that happen to be available. We discuss these issues and suggest an agenda for political methodologists interested in this very large class of aggregation problems.
King, Gary, Signorino, Curtis S., Alt, James E.
Submitted: 2000-08-28
Keywords: Duration, event count, binary, renewal process, aggregation
Abstract: (click to show/hide) Binary, count, and duration data all code discrete events occurring at points in time. Although a single data generation process can produce all of these three data types, the statistical literature is not very helpful in providing methods to estimate parameters of the same process from each. In fact, only a single theoretical process exists for which known statistical methods can estimate the same parameters --- and it is generally used only for count and duration data. The result is that seemingly trivial decisions about which level of data to use can have important consequences for substantive interpretations. We describe the theoretical event process for which results exist, based on time-independence. We also derive a set of models for a time-dependent process and compare their predictions to those of a commonly used model. Any hope of understanding and avoiding the more serious problems of aggregation bias in events data is contingent on first deriving a much wider arsenal of statistical models and theoretical processes that are not constrained by the particular forms of data that happen to be available. We discuss these issues and suggest an agenda for political methodologists interested in this very large class of aggregation problems.
Strategic Misspecification in Discrete Choice Models
Signorino, Curtis S., Yilmaz, Kuzey
Submitted: 2000-08-28
Keywords:
Abstract: (click to show/hide) [not transcribed]
Signorino, Curtis S., Yilmaz, Kuzey
Submitted: 2000-08-28
Keywords:
Abstract: (click to show/hide) [not transcribed]
A Unified Theory and Test of Extended Immediate Deterrence
Signorino, Curtis S., Tarar, Ahmer
Submitted: 2000-09-05
Keywords:
Abstract: (click to show/hide) We present a unified theory and test of extended immediate deterrence --- unified in the sense that we employ our theoretical deterrence model as our statistical model in the empirical analysis. The theoretical model is a straightforward formalization of the extended immediate deterrence logic in Huth (1988), coupled with private information concerning utilities. Contrary to Huth (1988), our empirical analysis suggests that nuclear weapons, military alliances, military arms transfers, and foreign trade all affect deterrence success. Our model correctly predicts almost 97\% of the potential Attacker's actions and over 91\% of the crisis outcomes. Finally, we find strong evidence that the likelihood of deterrence success and of war are not monotonically related to the variables involved in the deterrence calculus. This contradicts a fundamental assumption of most previous studies.
Signorino, Curtis S., Tarar, Ahmer
Submitted: 2000-09-05
Keywords:
Abstract: (click to show/hide) We present a unified theory and test of extended immediate deterrence --- unified in the sense that we employ our theoretical deterrence model as our statistical model in the empirical analysis. The theoretical model is a straightforward formalization of the extended immediate deterrence logic in Huth (1988), coupled with private information concerning utilities. Contrary to Huth (1988), our empirical analysis suggests that nuclear weapons, military alliances, military arms transfers, and foreign trade all affect deterrence success. Our model correctly predicts almost 97\% of the potential Attacker's actions and over 91\% of the crisis outcomes. Finally, we find strong evidence that the likelihood of deterrence success and of war are not monotonically related to the variables involved in the deterrence calculus. This contradicts a fundamental assumption of most previous studies.
A New Look at Cold War Presidents' Use of Force: Aggregation Bias, Truncation, and Temporal Dynamic Issues
Mitchell, Sara McLaughlin, Moore, Will H.
Submitted: 2000-09-07
Keywords: aggregation bias, truncation bias, use of force, PAR
Abstract: (click to show/hide) This study re-examines the findings reported in a seminal study of US presidents' use of force during the Cold War (Ostrom and Job 1986). We identify three methodological issues that affect inferences drawn in studies of presidential decisions to use force: aggregation, truncation, and dynamics. We suggest that a dichotomous measure of uses of force introduces aggregation bias, while the decision to examine only major uses of force introduces truncation bias. To address these issues, we compare two types of use of force measures (dichotomous and event count), in addition to comparing results for major, minor, and all uses of force. In addition, we argue that Ostrom and Job's focus on rivalry leads one to anticipate the presence of temporal dependence or dynamics in the use of force series. We estimate a Poisson Autoregressive (PAR) model proposed by Brandt and Williams (2000), which is able to account for temporal dynamics in an event count model. Our findings demonstrate the importance of these three methodological issues. Results of the PAR model show dynamics in the use of force series. We also find that international variables have a larger substantive effect on the president's decision to use force than do the domestic variables. Our study thus overturns the most dramatic finding reported in the Ostrom and Job study, a finding that we contend was driven by bias and model specification problems.
Mitchell, Sara McLaughlin, Moore, Will H.
Submitted: 2000-09-07
Keywords: aggregation bias, truncation bias, use of force, PAR
Abstract: (click to show/hide) This study re-examines the findings reported in a seminal study of US presidents' use of force during the Cold War (Ostrom and Job 1986). We identify three methodological issues that affect inferences drawn in studies of presidential decisions to use force: aggregation, truncation, and dynamics. We suggest that a dichotomous measure of uses of force introduces aggregation bias, while the decision to examine only major uses of force introduces truncation bias. To address these issues, we compare two types of use of force measures (dichotomous and event count), in addition to comparing results for major, minor, and all uses of force. In addition, we argue that Ostrom and Job's focus on rivalry leads one to anticipate the presence of temporal dependence or dynamics in the use of force series. We estimate a Poisson Autoregressive (PAR) model proposed by Brandt and Williams (2000), which is able to account for temporal dynamics in an event count model. Our findings demonstrate the importance of these three methodological issues. Results of the PAR model show dynamics in the use of force series. We also find that international variables have a larger substantive effect on the president's decision to use force than do the domestic variables. Our study thus overturns the most dramatic finding reported in the Ostrom and Job study, a finding that we contend was driven by bias and model specification problems.
Issue Voting and Ecological Inference
Thomsen, Soren R.
Submitted: 2000-09-14
Keywords: issue voting, ecological inference, electoral geography, multinomial logit
Abstract: (click to show/hide) This article proposes a unifying framework for individual and aggregate voting behavior. The proposed individual level model is a version of the multinomial logit model that applies to both issue voting, ideological voting and normative voting providing a close fit to survey data. The aggregate model is derived by using the binary logit model as an approximation to the multinomial logit model. The aggregate model is useful for modeling electoral change and for identification of homogenous political regions. Further, the unifying framework derives a method for ecological inference that applies to large tables and gives estimates of voter transitions close to survery results.
Thomsen, Soren R.
Submitted: 2000-09-14
Keywords: issue voting, ecological inference, electoral geography, multinomial logit
Abstract: (click to show/hide) This article proposes a unifying framework for individual and aggregate voting behavior. The proposed individual level model is a version of the multinomial logit model that applies to both issue voting, ideological voting and normative voting providing a close fit to survey data. The aggregate model is derived by using the binary logit model as an approximation to the multinomial logit model. The aggregate model is useful for modeling electoral change and for identification of homogenous political regions. Further, the unifying framework derives a method for ecological inference that applies to large tables and gives estimates of voter transitions close to survery results.
The Likely Consequences of Internet Voting for Political Representation
Alvarez, R. Michael, Nagler, Jonathan
Submitted: 2000-11-03
Keywords: Internet voting, digital divide, civil rights act, minority
Abstract: (click to show/hide) In this paper we examine how internet voting might impact political representation. We begin by reviewing the existing academic literature on NVRA and vote-by-mail elections, and then we turn more directly to the internet and electronic elections. First we look carefully at the ``digital divide'' in the United States, using recent survey data. Then we examine the sole existing electoral experiment with internet voting: the 2000 Arizona Democratic presidential primary. We provide evidence indicating that the internet voting experiment in Arizona might have had a negative impact on minority voter rights and political representation. After that, we consider the possible constituencies for internet voting, using polling data from California. We conclude with a summary of our results and our inferences the representational consequences of internet voting.
Alvarez, R. Michael, Nagler, Jonathan
Submitted: 2000-11-03
Keywords: Internet voting, digital divide, civil rights act, minority
Abstract: (click to show/hide) In this paper we examine how internet voting might impact political representation. We begin by reviewing the existing academic literature on NVRA and vote-by-mail elections, and then we turn more directly to the internet and electronic elections. First we look carefully at the ``digital divide'' in the United States, using recent survey data. Then we examine the sole existing electoral experiment with internet voting: the 2000 Arizona Democratic presidential primary. We provide evidence indicating that the internet voting experiment in Arizona might have had a negative impact on minority voter rights and political representation. After that, we consider the possible constituencies for internet voting, using polling data from California. We conclude with a summary of our results and our inferences the representational consequences of internet voting.
An Individual-Level Approach to Health Inequality: Child Survival in 50 Countries
King, Gary, Gakidou, Emmanuela
Submitted: 2000-11-27
Keywords: beta-binomial, health inequality, survey research
Abstract: (click to show/hide) BACKGROUND: Reducing health inequalities is an important part of the agenda of health policymakers globally. Studies of health inequalities have revealed large variations in average health status across social, economic, and other _groups_. However, no studies have been conducted on the distribution of the risk of ill-health across _individuals_. METHODS: We use an extended beta-binomial model to estimate the distribution the risk of death in children under the age of two in the 50 developing countries where data from a Demographic and Health Survey are available. Inequality in these distributions is measured by the WHO health inequality index. FINDINGS: At the same level of average child mortality, inequality in the risk of death across children can vary considerably across countries. Representing the entire distribution of risk with a single measure of inequality involves normative choices that we delineate and then formalise with quantitative measures. The results are not very sensitive to the choice of measure. Liberia, Mozambique and the Central African Republic have the largest inequalities in child survival, while Colombia, the Philippines and Kazakhstan have the lowest among the 50 countries measured. Exploratory analyses indicate that health inequality is predicted by low GDP, low health expenditures, and poverty, but not by income inequality or democratization. INTERPRETATION: Inequality estimates should routinely be reported alongside average levels of health, as they reveal important information about the distribution of health across individuals within populations. Measuring inequality with individual level data, rather than quantifying differences in average levels of health across social groups, enables meaningful comparisons of inequality across countries and analyses of the determinants of inequality. This approach should be extended to the measurement of inequalities in health expectancy (i.e., life expectancy discounted by expected disabilities).
King, Gary, Gakidou, Emmanuela
Submitted: 2000-11-27
Keywords: beta-binomial, health inequality, survey research
Abstract: (click to show/hide) BACKGROUND: Reducing health inequalities is an important part of the agenda of health policymakers globally. Studies of health inequalities have revealed large variations in average health status across social, economic, and other _groups_. However, no studies have been conducted on the distribution of the risk of ill-health across _individuals_. METHODS: We use an extended beta-binomial model to estimate the distribution the risk of death in children under the age of two in the 50 developing countries where data from a Demographic and Health Survey are available. Inequality in these distributions is measured by the WHO health inequality index. FINDINGS: At the same level of average child mortality, inequality in the risk of death across children can vary considerably across countries. Representing the entire distribution of risk with a single measure of inequality involves normative choices that we delineate and then formalise with quantitative measures. The results are not very sensitive to the choice of measure. Liberia, Mozambique and the Central African Republic have the largest inequalities in child survival, while Colombia, the Philippines and Kazakhstan have the lowest among the 50 countries measured. Exploratory analyses indicate that health inequality is predicted by low GDP, low health expenditures, and poverty, but not by income inequality or democratization. INTERPRETATION: Inequality estimates should routinely be reported alongside average levels of health, as they reveal important information about the distribution of health across individuals within populations. Measuring inequality with individual level data, rather than quantifying differences in average levels of health across social groups, enables meaningful comparisons of inequality across countries and analyses of the determinants of inequality. This approach should be extended to the measurement of inequalities in health expectancy (i.e., life expectancy discounted by expected disabilities).
Comparing GEE and Robust Standard Errors, with an Application to Judicial Voting
Zorn, Christopher
Submitted: 2000-11-27
Keywords: GEE, panel data, robust variance, inference
Abstract: (click to show/hide) Implicit in most statistical analyses is the assumption that observations are conditionally independent; this claim has important implications, both statistical and substantive, for the conclusions we draw. I outline and compare two alternatives for addressing heterogeneity due to correlated data: the use of "robust" (or "heteroskedasticity-corrected") standard errors, and application of the method of generalized estimating equations ("GEEs"). I provide an example, based on an earlier study of judicial voting in search and seizure cases (Segal 1986), and use the example to discuss practical considerations in choosing among the various variance estimators in the presence of correlated data.
Zorn, Christopher
Submitted: 2000-11-27
Keywords: GEE, panel data, robust variance, inference
Abstract: (click to show/hide) Implicit in most statistical analyses is the assumption that observations are conditionally independent; this claim has important implications, both statistical and substantive, for the conclusions we draw. I outline and compare two alternatives for addressing heterogeneity due to correlated data: the use of "robust" (or "heteroskedasticity-corrected") standard errors, and application of the method of generalized estimating equations ("GEEs"). I provide an example, based on an earlier study of judicial voting in search and seizure cases (Segal 1986), and use the example to discuss practical considerations in choosing among the various variance estimators in the presence of correlated data.
