Abstract It is understood that ensuring equation balance is a necessary condition for a valid model of times series data. Yet, the definition of balance provided so far has been incomplete and there has not been a consistent understanding of exactly why balance is important or how it can be applied. The discussion to date has focused on the estimates produced by the general error correction model (GECM). In this paper, we go beyond the GECM and beyond model estimates. We treat equation balance as a theoretical matter, not merely an empirical one, and describe how to use the concept of balance to test theoretical propositions before longitudinal data have been gathered. We explain how equation balance can be used to check if your theoretical or empirical model is either wrong or incomplete in a way that will prevent a meaningful interpretation of the model. We also raise the issue of “ $I(0)$ balance” and its importance.
{"title":"Balance as a Pre-Estimation Test for Time Series Analysis","authors":"Mark Pickup, Paul M. Kellstedt","doi":"10.1017/pan.2022.4","DOIUrl":"https://doi.org/10.1017/pan.2022.4","url":null,"abstract":"Abstract It is understood that ensuring equation balance is a necessary condition for a valid model of times series data. Yet, the definition of balance provided so far has been incomplete and there has not been a consistent understanding of exactly why balance is important or how it can be applied. The discussion to date has focused on the estimates produced by the general error correction model (GECM). In this paper, we go beyond the GECM and beyond model estimates. We treat equation balance as a theoretical matter, not merely an empirical one, and describe how to use the concept of balance to test theoretical propositions before longitudinal data have been gathered. We explain how equation balance can be used to check if your theoretical or empirical model is either wrong or incomplete in a way that will prevent a meaningful interpretation of the model. We also raise the issue of “ \u0000$I(0)$\u0000 balance” and its importance.","PeriodicalId":48270,"journal":{"name":"Political Analysis","volume":"31 1","pages":"295 - 304"},"PeriodicalIF":5.4,"publicationDate":"2022-04-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"44251727","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract While a difference-in-differences (DID) design was originally developed with one pre- and one posttreatment period, data from additional pretreatment periods are often available. How can researchers improve the DID design with such multiple pretreatment periods under what conditions? We first use potential outcomes to clarify three benefits of multiple pretreatment periods: (1) assessing the parallel trends assumption, (2) improving estimation accuracy, and (3) allowing for a more flexible parallel trends assumption. We then propose a new estimator, double DID, which combines all the benefits through the generalized method of moments and contains the two-way fixed effects regression as a special case. We show that the double DID requires a weaker assumption about outcome trends and is more efficient than existing DID estimators. We also generalize the double DID to the staggered adoption design where different units can receive the treatment in different time periods. We illustrate the proposed method with two empirical applications, covering both the basic DID and staggered adoption designs. We offer an open-source R package that implements the proposed methodologies.
{"title":"Using Multiple Pretreatment Periods to Improve Difference-in-Differences and Staggered Adoption Designs","authors":"Naoki Egami, S. Yamauchi","doi":"10.1017/pan.2022.8","DOIUrl":"https://doi.org/10.1017/pan.2022.8","url":null,"abstract":"Abstract While a difference-in-differences (DID) design was originally developed with one pre- and one posttreatment period, data from additional pretreatment periods are often available. How can researchers improve the DID design with such multiple pretreatment periods under what conditions? We first use potential outcomes to clarify three benefits of multiple pretreatment periods: (1) assessing the parallel trends assumption, (2) improving estimation accuracy, and (3) allowing for a more flexible parallel trends assumption. We then propose a new estimator, double DID, which combines all the benefits through the generalized method of moments and contains the two-way fixed effects regression as a special case. We show that the double DID requires a weaker assumption about outcome trends and is more efficient than existing DID estimators. We also generalize the double DID to the staggered adoption design where different units can receive the treatment in different time periods. We illustrate the proposed method with two empirical applications, covering both the basic DID and staggered adoption designs. We offer an open-source R package that implements the proposed methodologies.","PeriodicalId":48270,"journal":{"name":"Political Analysis","volume":"31 1","pages":"195 - 212"},"PeriodicalIF":5.4,"publicationDate":"2022-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42521787","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Lisa P. Argyle, E. Busby, Nancy Fulda, Joshua R Gubler, Christopher Rytting, Taylor Sorensen, D. Wingate
Pre-trained language models derive substantial linguistic and factual knowledge from the massive corpora on which they are trained, and prompt engineering seeks to align these models to specific tasks. Unfortunately, existing prompt engineering methods require significant amounts of labeled data, access to model parameters, or both. We introduce a new method for selecting prompt templates without labeled examples and without direct access to the model. Specifically, over a set of candidate templates, we choose the template that maximizes the mutual information between the input and the corresponding model output. Across 8 datasets representing 7 distinct NLP tasks, we show that when a template has high mutual information, it also has high accuracy on the task. On the largest model, selecting prompts with our method gets 90% of the way from the average prompt accuracy to the best prompt accuracy and requires no ground truth labels.
{"title":"An Information-theoretic Approach to Prompt Engineering Without Ground Truth Labels","authors":"Lisa P. Argyle, E. Busby, Nancy Fulda, Joshua R Gubler, Christopher Rytting, Taylor Sorensen, D. Wingate","doi":"10.1017/pan.2023.2","DOIUrl":"https://doi.org/10.1017/pan.2023.2","url":null,"abstract":"Pre-trained language models derive substantial linguistic and factual knowledge from the massive corpora on which they are trained, and prompt engineering seeks to align these models to specific tasks. Unfortunately, existing prompt engineering methods require significant amounts of labeled data, access to model parameters, or both. We introduce a new method for selecting prompt templates without labeled examples and without direct access to the model. Specifically, over a set of candidate templates, we choose the template that maximizes the mutual information between the input and the corresponding model output. Across 8 datasets representing 7 distinct NLP tasks, we show that when a template has high mutual information, it also has high accuracy on the task. On the largest model, selecting prompts with our method gets 90% of the way from the average prompt accuracy to the best prompt accuracy and requires no ground truth labels.","PeriodicalId":48270,"journal":{"name":"Political Analysis","volume":"31 1","pages":"337 - 351"},"PeriodicalIF":5.4,"publicationDate":"2022-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"45581735","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract In recent decades, political science literature has experienced significant growth in the popularity of nonlinear models with multiplicative interaction terms. When one or more constitutive variables are not binary, most studies report the marginal effect of the variable of interest at its sample mean while allowing the other constitutive variable/s to vary along its range and holding all other covariates constant at their means, modes, or medians. In this article, we argue that this conventional approach is not always the most suitable since the marginal effect of a variable at its sample mean might not be sufficiently representative of its prevalent effect at a specific value of the conditioning variable and might produce excessively model-dependent predictions. We propose two procedures to help researchers gain a better understanding of how the typical effect of the variable of interest varies as a function of the conditioning variable: (1) computing and plotting the marginal effects at all in-sample combinations of the values of the constitutive variables and (2) computing and plotting what we call the “Distribution-Weighted Average Marginal Effect” over the values of the conditioning variable.
{"title":"Taking Distributions Seriously: On the Interpretation of the Estimates of Interactive Nonlinear Models","authors":"A. Zhirnov, Mert Moral, Evgeny Sedashov","doi":"10.1017/pan.2022.9","DOIUrl":"https://doi.org/10.1017/pan.2022.9","url":null,"abstract":"Abstract In recent decades, political science literature has experienced significant growth in the popularity of nonlinear models with multiplicative interaction terms. When one or more constitutive variables are not binary, most studies report the marginal effect of the variable of interest at its sample mean while allowing the other constitutive variable/s to vary along its range and holding all other covariates constant at their means, modes, or medians. In this article, we argue that this conventional approach is not always the most suitable since the marginal effect of a variable at its sample mean might not be sufficiently representative of its prevalent effect at a specific value of the conditioning variable and might produce excessively model-dependent predictions. We propose two procedures to help researchers gain a better understanding of how the typical effect of the variable of interest varies as a function of the conditioning variable: (1) computing and plotting the marginal effects at all in-sample combinations of the values of the constitutive variables and (2) computing and plotting what we call the “Distribution-Weighted Average Marginal Effect” over the values of the conditioning variable.","PeriodicalId":48270,"journal":{"name":"Political Analysis","volume":"31 1","pages":"213 - 234"},"PeriodicalIF":5.4,"publicationDate":"2022-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"41839871","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
V o lu m e 28• N u m b er • A p rl20 20 V o l u m e 2 8•Number2•April2020 V o lu m e 28• N u m b er • A p rl20 20 ARICLES Dcrete Coice Data w ith U noerved H etereneity:A CondnalBinary Q untile M odel Xiao Lu M eauring he om petienessof Eltions Gary W .Cx,Jon H .iva nd DnielM Sm ith U neected Eveduring Srvey Dign: Prom se nd PitfallsfusalInference JrdiM uñoz,rtFalcó-Gim no nd Erique H erndez M eauring Sugroup Preencesin ConjtExperim ents Thom asJ.per,Sara .H obltand am esilley
V o lu m e 28•N u m b er•A p rl20 20 V o l u m e 2 8•数字2•2020年4月V o lu m e 28•N u m b r•A p rl20 20 ARICLES Dcrete Coice数据与u无服务的H确定性：一个CondnalBinary Q，直到m模型Xiao lu m获得了Gary w.Cx、Jon H.iva和DnielM Sm与u Neeted Evevide Srvey Dign的合作伙伴关系：Prom se nd Pitfallsfusal推理JrdiM uñoz，rtFalcó-Gim no nd Erique H erndez M eauring Sugroup Preencesin联合专家Thom asJ.per，Sara.H obltand am esilley
{"title":"PAN volume 30 issue 2 Cover and Front matter","authors":"Emily M. Farris","doi":"10.1017/pan.2022.6","DOIUrl":"https://doi.org/10.1017/pan.2022.6","url":null,"abstract":"V o lu m e 28• N u m b er • A p rl20 20 V o l u m e 2 8•Number2•April2020 V o lu m e 28• N u m b er • A p rl20 20 ARICLES Dcrete Coice Data w ith U noerved H etereneity:A CondnalBinary Q untile M odel Xiao Lu M eauring he om petienessof Eltions Gary W .Cx,Jon H .iva nd DnielM Sm ith U neected Eveduring Srvey Dign: Prom se nd PitfallsfusalInference JrdiM uñoz,rtFalcó-Gim no nd Erique H erndez M eauring Sugroup Preencesin ConjtExperim ents Thom asJ.per,Sara .H obltand am esilley","PeriodicalId":48270,"journal":{"name":"Political Analysis","volume":" ","pages":"f1 - f3"},"PeriodicalIF":5.4,"publicationDate":"2022-03-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"47064427","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract The link between objective facts and politically relevant beliefs is an essential mechanism for democratic accountability. Yet the bulk of empirical work on this topic measures objective facts at whatever geographic units are readily available. We investigate the implications of these largely arbitrary choices for predicting individual-level opinions. We show that varying the geographic resolution—namely aggregating economic data to different geographic units—influences the strength of the relationship between economic evaluations and local economic conditions. Finding that unemployment claims are the best predictor of economic evaluations, especially when aggregated at the commuting zone or media market level, we underscore the importance of the modifiable areal unit problem. Our methods provide an example of how applied scholars might investigate the importance of geography in their own research going forward.
{"title":"Geographic Boundaries and Local Economic Conditions Matter for Views of the Economy","authors":"James Bisbee, J. Zilinsky","doi":"10.1017/pan.2021.50","DOIUrl":"https://doi.org/10.1017/pan.2021.50","url":null,"abstract":"Abstract The link between objective facts and politically relevant beliefs is an essential mechanism for democratic accountability. Yet the bulk of empirical work on this topic measures objective facts at whatever geographic units are readily available. We investigate the implications of these largely arbitrary choices for predicting individual-level opinions. We show that varying the geographic resolution—namely aggregating economic data to different geographic units—influences the strength of the relationship between economic evaluations and local economic conditions. Finding that unemployment claims are the best predictor of economic evaluations, especially when aggregated at the commuting zone or media market level, we underscore the importance of the modifiable areal unit problem. Our methods provide an example of how applied scholars might investigate the importance of geography in their own research going forward.","PeriodicalId":48270,"journal":{"name":"Political Analysis","volume":"31 1","pages":"288 - 294"},"PeriodicalIF":5.4,"publicationDate":"2022-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49474478","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract A core part of political research is to identify how political preferences are shaped. The nature of these questions is such that robust causal identification is often difficult to achieve, and we are not seldom stuck with observational methods that we know have limited causal validity. The purpose of this paper is to measure the magnitude of bias stemming from both measurable and unmeasurable confounders across three broad domains of individual determinants of political preferences: socio-economic factors, moral values, and psychological constructs. We leverage a unique combination of rich Swedish registry data for a large sample of identical twins, with a comprehensive battery of 34 political preference measures, and build a meta-analytical model comparing our most conservative observational (naive) estimates with discordant twin estimates. This allows us to infer the amount of bias from unobserved genetic and shared environmental factors that remains in the naive models for our predictors, while avoiding precision issues common in family-based designs. The results are sobering: in most cases, substantial bias remains in naive models. A rough heuristic is that about half of the effect size even in conservative observational estimates is composed of confounding.
{"title":"Quantifying Bias from Measurable and Unmeasurable Confounders Across Three Domains of Individual Determinants of Political Preferences","authors":"Rafael Ahlskog, Sven Oskarsson","doi":"10.1017/pan.2022.2","DOIUrl":"https://doi.org/10.1017/pan.2022.2","url":null,"abstract":"Abstract A core part of political research is to identify how political preferences are shaped. The nature of these questions is such that robust causal identification is often difficult to achieve, and we are not seldom stuck with observational methods that we know have limited causal validity. The purpose of this paper is to measure the magnitude of bias stemming from both measurable and unmeasurable confounders across three broad domains of individual determinants of political preferences: socio-economic factors, moral values, and psychological constructs. We leverage a unique combination of rich Swedish registry data for a large sample of identical twins, with a comprehensive battery of 34 political preference measures, and build a meta-analytical model comparing our most conservative observational (naive) estimates with discordant twin estimates. This allows us to infer the amount of bias from unobserved genetic and shared environmental factors that remains in the naive models for our predictors, while avoiding precision issues common in family-based designs. The results are sobering: in most cases, substantial bias remains in naive models. A rough heuristic is that about half of the effect size even in conservative observational estimates is composed of confounding.","PeriodicalId":48270,"journal":{"name":"Political Analysis","volume":"31 1","pages":"181 - 194"},"PeriodicalIF":5.4,"publicationDate":"2022-02-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"47468189","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract We present a hierarchical Dirichlet regression model with Gaussian process priors that enables accurate and well-calibrated forecasts for U.S. Senate elections at varying time horizons. This Bayesian model provides a balance between predictions based on time-dependent opinion polls and those made based on fundamentals. It also provides uncertainty estimates that arise naturally from historical data on elections and polls. Experiments show that our model is highly accurate and has a well calibrated coverage rate for vote share predictions at various forecasting horizons. We validate the model with a retrospective forecast of the 2018 cycle as well as a true out-of-sample forecast for 2020. We show that our approach achieves state-of-the art accuracy and coverage despite relying on few covariates.
{"title":"Polls, Context, and Time: A Dynamic Hierarchical Bayesian Forecasting Model for US Senate Elections","authors":"Yehua Chen, R. Garnett, J. Montgomery","doi":"10.1017/pan.2021.42","DOIUrl":"https://doi.org/10.1017/pan.2021.42","url":null,"abstract":"Abstract We present a hierarchical Dirichlet regression model with Gaussian process priors that enables accurate and well-calibrated forecasts for U.S. Senate elections at varying time horizons. This Bayesian model provides a balance between predictions based on time-dependent opinion polls and those made based on fundamentals. It also provides uncertainty estimates that arise naturally from historical data on elections and polls. Experiments show that our model is highly accurate and has a well calibrated coverage rate for vote share predictions at various forecasting horizons. We validate the model with a retrospective forecast of the 2018 cycle as well as a true out-of-sample forecast for 2020. We show that our approach achieves state-of-the art accuracy and coverage despite relying on few covariates.","PeriodicalId":48270,"journal":{"name":"Political Analysis","volume":"31 1","pages":"113 - 133"},"PeriodicalIF":5.4,"publicationDate":"2022-01-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"41362685","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract Some scholars build models to classify documents into chosen categories. Others, especially social scientists who tend to focus on population characteristics, instead usually estimate the proportion of documents in each category—using either parametric “classify-and-count” methods or “direct” nonparametric estimation of proportions without individual classification. Unfortunately, classify-and-count methods can be highly model-dependent or generate more bias in the proportions even as the percent of documents correctly classified increases. Direct estimation avoids these problems, but can suffer when the meaning of language changes between training and test sets or is too similar across categories. We develop an improved direct estimation approach without these issues by including and optimizing continuous text features, along with a form of matching adapted from the causal inference literature. Our approach substantially improves performance in a diverse collection of 73 datasets. We also offer easy-to-use software that implements all ideas discussed herein.
{"title":"An Improved Method of Automated Nonparametric Content Analysis for Social Science","authors":"Gary King, Connor Jerzak, Anton Strezhnev","doi":"10.1017/pan.2021.36","DOIUrl":"https://doi.org/10.1017/pan.2021.36","url":null,"abstract":"Abstract Some scholars build models to classify documents into chosen categories. Others, especially social scientists who tend to focus on population characteristics, instead usually estimate the proportion of documents in each category—using either parametric “classify-and-count” methods or “direct” nonparametric estimation of proportions without individual classification. Unfortunately, classify-and-count methods can be highly model-dependent or generate more bias in the proportions even as the percent of documents correctly classified increases. Direct estimation avoids these problems, but can suffer when the meaning of language changes between training and test sets or is too similar across categories. We develop an improved direct estimation approach without these issues by including and optimizing continuous text features, along with a form of matching adapted from the causal inference literature. Our approach substantially improves performance in a diverse collection of 73 datasets. We also offer easy-to-use software that implements all ideas discussed herein.","PeriodicalId":48270,"journal":{"name":"Political Analysis","volume":"31 1","pages":"42 - 58"},"PeriodicalIF":5.4,"publicationDate":"2022-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"44479486","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：