Publications
“When Do Politicians Grandstand? Measuring Message Politics in Committee Hearings” Forthcoming at the Journal of Politics.
While congressional committee members sometimes hold hearings to collect and transmit specialized information to the floor, they also use hearings as venues to send political messages by framing an issue or a party to the public which I refer to as “grandstanding.” However, we lack clear understanding of when they strategically engage in grandstanding. I argue that when committee members have limited legislative power they resort to making grandstanding speeches in hearings to please their target audience. Using 12,820 House committee hearing transcripts from the 105th to 114th Congresses and employing a crowd-sourced supervised learning method, I measure a “grandstanding score” for each statement that committee members make. Findings suggest that grandstanding efforts are made more commonly among minority members under a unified government, and non-chair members of powerful committees, and in committees with jurisdiction over policies that the president wields primary power, such as foreign affairs and national security.
“Committee Chair’s Majority Partisan Status and Its Effect on Information Transmission via Hearings” 2019. The Journal of Legislative Studies
While US Congress assigns only the members of a majority party to committee chairs, some state legislatures and other legislative bodies using a proportional representation system also consider members of a minority party for the position to promote a bipartisan policy making practice. Although previous literature investigates the effects of bipartisan rules and practices exploiting such institutional variations, the informational benefit of having a minority partisan committee chair has not been explored. By extending a recent study by Park (2017), this research note theoretically examines the effect of the committee chair’s majority partisan status on information acquisition and transmission via committee hearings. Findings suggest that under some conditions, the floor can informationally benefit more from having a chair representing a minority party in the chamber with opposite bias call a hearing than with a chair representing a majority party.
“The Grievance Asymmetry in Economic Voting Revisited” 2019. Electoral Studies.
In principle, committees hold hearings to gather and provide information to their principals, but some hearings are characterized as political showcases. This article investigates conditions that moderate committee members’ incentives to hold an informative hearing by presenting a game-theoretic model and a lab experiment. Specifically, it studies when committees hold hearings and which types of hearing they hold by varying policy preferences of committee members and the principal and political gains from posturing. Findings provide new insights to how preferences and power distribution affect individuals’ incentives to be informed when they make decisions as members of a committee in many contexts.
“A Lab Experiment on Committee Hearings: Preferences, Power and a Quest for Information” 2017. Legislative Studies Quarterly.
It has been controversial whether incumbents are punished more for a bad economy than they are rewarded for a good economy due to mixed results from previous studies on one or handful number of countries. This paper makes an empirical contribution to this lingering question by conducting extensive tests on whether this asymmetry hypothesis is a cross-nationally generalizable phenomenon using all currently available modules of the Comparative Study of Electoral Systems survey from 122 elections in 42 representative democracies between 1996 and 2016, as well as macro-economic indicators and individual-level economic perception. In general, this paper finds little support for the asymmetry hypothesis; although the evidence of such asymmetric economic voting is found in some subpopulations using certain economic indicators, these conditional effects are largely inconsistent, suggesting that it is still safe to assume a linear relationship between economic conditions and support for the incumbent.
Working Papers
"Electoral Rewards for Political Grandstanding in Congressional Committee Hearings"
Sometimes members of Congress use committee hearings as a venue to send political messages to the public by framing an issue or branding a party which I refer to as “grandstanding.” However, we lack clear understanding of their incentives to engage in grandstanding and who the target audience is. This paper examines whether House members’ grandstanding efforts are electorally rewarded through increased vote shares or by soliciting more political donations using a novel dataset of 12,820 House committee hearing transcripts from the 105th to 114th Congresses. Findings suggest that if a committee member grandstands more in hearings than he normally does, he is likely to accrue more votes in the following election but do not solicit more political donations. In contrast, an opposite rewarding mechanism is at work for being more active in policy-making. This result suggests that grandstanding which is often considered wasteful can be a useful political tool for politicians to communicate electoral messages to their constituents and thus can change the electoral climate in favor of themselves. These findings further imply that, due to the asymmetric response from constituents and donors, the political representation may work unequally in favor of political donors because politicians can win voters’ minds only by verbally appealing to them while benefiting those affluent donors policy-wise.
"Measuring Concepts in Text: Using Supervised Learning to Estimate Continuous Latent Traits in Individual Texts" with Jacob Montgomery
This study explores how to best measure a latent trait in textual data using crowd-sourced supervised learning and the ensemble Bayesian model averaging which combines multiple learners by assigning weight to each to optimize performance. Using the original dataset on Senate confirmation hearings from the 105th to 115th Congresses, we measure the negativity of committee members’ questioning tones to showcase necessary validation steps that are important for justifying modeling choices as well as improving the model’s prediction performance but often ignored by researchers utilizing the supervised learning method. Furthermore, we compare our results to other supervised and unsupervised learning models commonly used in political science literature and show that our model outperforms the others.
"A Theory of the Supply of an Informative Signal in Competitive Elections" with Myunghoon Kang (Under Review)
We present a signaling model where two competing parties decide whether to spend their limited resources either on improving the precision of the signal on the incumbent party’s performance or on party brand building or a combination of both as their electoral campaign strategy. Our findings suggest that if one party has obviously higher quality than the other the parties’ strategies will polarize such that the one with higher quality will focus on improving the precision of the signal; however, if both parties have similar level of quality, the one with more resources will contribute to the precision of the signal but this effort will reduce as the qualities of both parties diverge.
"Testing Conditional Effects of a Moderator in Deliberation: A Lab Experiment" (Under Review)
Recent findings show that deliberation outcomes can be vulnerable to manipulations by moderators. However, what makes moderators influential over deliberation outcomes and how to minimize the chance of manipulation have not been investigated yet. To address this issue, this paper presents a lab experiment which examines the conditional effect of a moderator’s perceived expertise on his potential to persuade participants of deliberation. Subjects are asked to deliberate on whether to expand the subsidized jobs program of New York State or not. The result suggests that participants are more susceptible to a moderator’s influence when they perceive him as having expertise on the policy than not. They also tend to assume that a moderator would be more knowledgeable on the issue than themselves due to his leading position in the discourse. However, such a perception does not strengthen his persuasive message while his self-claimed expertise does. This paper contributes not only to deliberative democracy but also provides a broader implication to studies on leadership effects in legislative committees.
Works in Progress
"The Changing Role of Committee Hearings in the Era of Polarization and Party Government"
As the Congress historically experienced a shift from a committee government to a party government which empowered the party leadership while weakening the discretion that committees had, the role of the congressional committees in the legislative process has been believed to be changed. However, the literature lacks a systematic analysis of how their role has changed over time. Using congressional committee hearing transcripts in select Congresses from the early 1960s to the 114th Congresses, I examine whether the role of committees as an informational mediator in the legislative procedures has declined over time. In addition, I examine the causal effect of the congressional transformation led by Newt Gingrich, upon the Republican rule in both chambers after a long break, on the congressmen’s committee behaviors using regression discontinuity design on the data collected from the 103th and 104th Congress.
"Gendered Information: Gender Composition of Witness Appearances in Congressional Hearings
and Information Provision in the US Congress" with Pamela Ban and Hye Young You
How are politicians informed and who provides information to politicians? Congressional committee hearings have played a crucial role in acquiring policy relevant information. However, we know little who appears in those hearings and whether the composition of hearing witnesses varies depending on the topic of the hearings.
In particular, we are interested in gender composition of witnesses who appeared in congressional hearings. Existing literature documents that there is significant gender gap in activities that could influence the policy choices of legislators such as campaign contributions, staffing decisions, and the gender composition of lobbyists. Congressional members’ interactions with individuals and groups shape their perspectives and policy choices and some concerns are raised that women do not have equal footing on their policy inputs even on the issues that dominantly affect women. We argue that the lack of female witnesses in congressional hearings could explain the limited influence of women’s voice in policies. We assembled the most comprehensive dataset about information about witnesses (e.g., Their names and affiliations) invited to congressional committee hearings held for the period 1963-2018 and information specific to hearings (e.g., Dates of hearings, committees, a policy issue discussed in hearings, etc.). First, we provide the most comprehensive description of the gender composition of witnesses in congressional hearings over time and across the committees. Second, we examine factors that are associated with the appearance of female witnesses. Specifically, we consider the partisanship and gender of committee leadership as well as the gender composition of the whole committee. We also consider whether issues that hearings address are systematically associated with the proportion of female witnesses in congressional hearings.
"Politicizing Information in Policy-Making Procedures via Hearings" with Pamela Ban and Hye Young You
Committee members’ right to choose witnesses to hear on a bill provides political opportunities not only for the members themselves but also for interest groups to lobby to provide their expert knowledge in order to publicly frame the issue in their favor and affect policy outcomes. This mechanism may have been strengthened by Newt Gingrich’s reform of the House in 1994 which resulted in downsizing the research-oriented congressional staffs prompting committees to more heavily rely on interest groups for expert information. Using my novel hearing data back to 1960s, this study investigates whether this change has motivated committee members to increasingly invite and transmit their donor sponsored research products to legislate public policies and help the policy-making procedures as well as potentially policy outcomes dominated by special interest groups and affluent individuals. We will examine this question by analyzing whether the similarity between members’ statements and testimonies of witnesses representing their donor institutions has increased after the reform using regression discontinuity design. Findings of this study will help us better understand the sources of social inequality in American society.
"Better Targeted Messages for Get-Out-the-Vote Campaigns" with Chris Dawes and John Kuk
We study whether minority voters (e.g. female voters, racial minority and naturalized citizens) tend to respond to different GOTV messages than other main stream voters. Based on our preliminary finding that men are more likely to be motivated by civic duty while women are more interested in practicing their voting rights, we argue that voters whose voting rights have been obtained through a harder process historically or personally are more likely to be motivated to turnout when prompted with a campaign message emphasizing the importance of exercising their voting right than others and test the argument through a survey experiment in the 2019 local election in New Jersey as a pilot study and in the 2020 presidential election at a larger scale.
"Inversed Representation: Top Down Issue Framing through Twitter" with Yu-Ru Lin
This project studies whether politicians communicate the framing of an issue to the public via Twitter. We have collected tweets of 600K Twitter users from 2016 until now and the tweets of U.S. congressmen from 2008. Using the Twitter data and supervised-learning method, we are measuring the moral and emotional dimensions of individual tweets on each of the five most controversial issues (e.g. gun control, immigration, abortion, climate change and gay marriage). We will examine whether and how the average moral and emotional dimensions on each issue based on the Democrat and Republican politicians’ tweets, respectively, has changed over time during the period covered in the data and see if these changes are adopted by the liberal and conservative Twitter users’ discourse on each issue.
"Measuring Group Differences Across Party Factions: An Application of TRIBAL Model" with Yu-Ru Lin
In another project with Yu-Ru Lin, we validate our supervised-learning model which is designed to predict group biases using social media data by demonstrating that the model performs well on predicting multiple group membership as well as binary group membership. By extending our application of the model which was originally tested on classifying liberal and conservative tweets, this study predicts individual congressmen's membership of party factions in each Congress by measuring their moral dimensions on each of the five most controversial issues based on their tweets in a given Congress.
"A Model of Abstention with Known Asymmetric Information Quality and Public Messages" with Rebecca Morton
We study voters’ decisions to publicly communicate their private signals before voting when there is a known asymmetry in the quality of signals they receive. We present a game theoretic model and conduct laboratory experiments to test our theoretical predictions. The study suggests that as the asymmetry between the high quality and low quality signals increases sometimes there is a situation where only the voter with the high quality signal sends a message to persuade others while those with low quality signals prefer to just listen. Also, such an effort by the high quality signal holder to inform others tends to have an effect on their vote choices. The effect is strongest especially when both types of signals are not too precise because otherwise those with low types will follow their own signal ignoring the message of the high type if they are different. This study suggests that possessing relatively high quality information may provide the person with a political power to affect final policy outcomes, and this power can be amplified through communication before voting. The real world applications can be found in legislative committees where committee chairs tend to be better informed than other members.
“When Do Politicians Grandstand? Measuring Message Politics in Committee Hearings” Forthcoming at the Journal of Politics.
While congressional committee members sometimes hold hearings to collect and transmit specialized information to the floor, they also use hearings as venues to send political messages by framing an issue or a party to the public which I refer to as “grandstanding.” However, we lack clear understanding of when they strategically engage in grandstanding. I argue that when committee members have limited legislative power they resort to making grandstanding speeches in hearings to please their target audience. Using 12,820 House committee hearing transcripts from the 105th to 114th Congresses and employing a crowd-sourced supervised learning method, I measure a “grandstanding score” for each statement that committee members make. Findings suggest that grandstanding efforts are made more commonly among minority members under a unified government, and non-chair members of powerful committees, and in committees with jurisdiction over policies that the president wields primary power, such as foreign affairs and national security.
“Committee Chair’s Majority Partisan Status and Its Effect on Information Transmission via Hearings” 2019. The Journal of Legislative Studies
While US Congress assigns only the members of a majority party to committee chairs, some state legislatures and other legislative bodies using a proportional representation system also consider members of a minority party for the position to promote a bipartisan policy making practice. Although previous literature investigates the effects of bipartisan rules and practices exploiting such institutional variations, the informational benefit of having a minority partisan committee chair has not been explored. By extending a recent study by Park (2017), this research note theoretically examines the effect of the committee chair’s majority partisan status on information acquisition and transmission via committee hearings. Findings suggest that under some conditions, the floor can informationally benefit more from having a chair representing a minority party in the chamber with opposite bias call a hearing than with a chair representing a majority party.
“The Grievance Asymmetry in Economic Voting Revisited” 2019. Electoral Studies.
In principle, committees hold hearings to gather and provide information to their principals, but some hearings are characterized as political showcases. This article investigates conditions that moderate committee members’ incentives to hold an informative hearing by presenting a game-theoretic model and a lab experiment. Specifically, it studies when committees hold hearings and which types of hearing they hold by varying policy preferences of committee members and the principal and political gains from posturing. Findings provide new insights to how preferences and power distribution affect individuals’ incentives to be informed when they make decisions as members of a committee in many contexts.
“A Lab Experiment on Committee Hearings: Preferences, Power and a Quest for Information” 2017. Legislative Studies Quarterly.
It has been controversial whether incumbents are punished more for a bad economy than they are rewarded for a good economy due to mixed results from previous studies on one or handful number of countries. This paper makes an empirical contribution to this lingering question by conducting extensive tests on whether this asymmetry hypothesis is a cross-nationally generalizable phenomenon using all currently available modules of the Comparative Study of Electoral Systems survey from 122 elections in 42 representative democracies between 1996 and 2016, as well as macro-economic indicators and individual-level economic perception. In general, this paper finds little support for the asymmetry hypothesis; although the evidence of such asymmetric economic voting is found in some subpopulations using certain economic indicators, these conditional effects are largely inconsistent, suggesting that it is still safe to assume a linear relationship between economic conditions and support for the incumbent.
Working Papers
"Electoral Rewards for Political Grandstanding in Congressional Committee Hearings"
Sometimes members of Congress use committee hearings as a venue to send political messages to the public by framing an issue or branding a party which I refer to as “grandstanding.” However, we lack clear understanding of their incentives to engage in grandstanding and who the target audience is. This paper examines whether House members’ grandstanding efforts are electorally rewarded through increased vote shares or by soliciting more political donations using a novel dataset of 12,820 House committee hearing transcripts from the 105th to 114th Congresses. Findings suggest that if a committee member grandstands more in hearings than he normally does, he is likely to accrue more votes in the following election but do not solicit more political donations. In contrast, an opposite rewarding mechanism is at work for being more active in policy-making. This result suggests that grandstanding which is often considered wasteful can be a useful political tool for politicians to communicate electoral messages to their constituents and thus can change the electoral climate in favor of themselves. These findings further imply that, due to the asymmetric response from constituents and donors, the political representation may work unequally in favor of political donors because politicians can win voters’ minds only by verbally appealing to them while benefiting those affluent donors policy-wise.
"Measuring Concepts in Text: Using Supervised Learning to Estimate Continuous Latent Traits in Individual Texts" with Jacob Montgomery
This study explores how to best measure a latent trait in textual data using crowd-sourced supervised learning and the ensemble Bayesian model averaging which combines multiple learners by assigning weight to each to optimize performance. Using the original dataset on Senate confirmation hearings from the 105th to 115th Congresses, we measure the negativity of committee members’ questioning tones to showcase necessary validation steps that are important for justifying modeling choices as well as improving the model’s prediction performance but often ignored by researchers utilizing the supervised learning method. Furthermore, we compare our results to other supervised and unsupervised learning models commonly used in political science literature and show that our model outperforms the others.
"A Theory of the Supply of an Informative Signal in Competitive Elections" with Myunghoon Kang (Under Review)
We present a signaling model where two competing parties decide whether to spend their limited resources either on improving the precision of the signal on the incumbent party’s performance or on party brand building or a combination of both as their electoral campaign strategy. Our findings suggest that if one party has obviously higher quality than the other the parties’ strategies will polarize such that the one with higher quality will focus on improving the precision of the signal; however, if both parties have similar level of quality, the one with more resources will contribute to the precision of the signal but this effort will reduce as the qualities of both parties diverge.
"Testing Conditional Effects of a Moderator in Deliberation: A Lab Experiment" (Under Review)
Recent findings show that deliberation outcomes can be vulnerable to manipulations by moderators. However, what makes moderators influential over deliberation outcomes and how to minimize the chance of manipulation have not been investigated yet. To address this issue, this paper presents a lab experiment which examines the conditional effect of a moderator’s perceived expertise on his potential to persuade participants of deliberation. Subjects are asked to deliberate on whether to expand the subsidized jobs program of New York State or not. The result suggests that participants are more susceptible to a moderator’s influence when they perceive him as having expertise on the policy than not. They also tend to assume that a moderator would be more knowledgeable on the issue than themselves due to his leading position in the discourse. However, such a perception does not strengthen his persuasive message while his self-claimed expertise does. This paper contributes not only to deliberative democracy but also provides a broader implication to studies on leadership effects in legislative committees.
Works in Progress
"The Changing Role of Committee Hearings in the Era of Polarization and Party Government"
As the Congress historically experienced a shift from a committee government to a party government which empowered the party leadership while weakening the discretion that committees had, the role of the congressional committees in the legislative process has been believed to be changed. However, the literature lacks a systematic analysis of how their role has changed over time. Using congressional committee hearing transcripts in select Congresses from the early 1960s to the 114th Congresses, I examine whether the role of committees as an informational mediator in the legislative procedures has declined over time. In addition, I examine the causal effect of the congressional transformation led by Newt Gingrich, upon the Republican rule in both chambers after a long break, on the congressmen’s committee behaviors using regression discontinuity design on the data collected from the 103th and 104th Congress.
"Gendered Information: Gender Composition of Witness Appearances in Congressional Hearings
and Information Provision in the US Congress" with Pamela Ban and Hye Young You
How are politicians informed and who provides information to politicians? Congressional committee hearings have played a crucial role in acquiring policy relevant information. However, we know little who appears in those hearings and whether the composition of hearing witnesses varies depending on the topic of the hearings.
In particular, we are interested in gender composition of witnesses who appeared in congressional hearings. Existing literature documents that there is significant gender gap in activities that could influence the policy choices of legislators such as campaign contributions, staffing decisions, and the gender composition of lobbyists. Congressional members’ interactions with individuals and groups shape their perspectives and policy choices and some concerns are raised that women do not have equal footing on their policy inputs even on the issues that dominantly affect women. We argue that the lack of female witnesses in congressional hearings could explain the limited influence of women’s voice in policies. We assembled the most comprehensive dataset about information about witnesses (e.g., Their names and affiliations) invited to congressional committee hearings held for the period 1963-2018 and information specific to hearings (e.g., Dates of hearings, committees, a policy issue discussed in hearings, etc.). First, we provide the most comprehensive description of the gender composition of witnesses in congressional hearings over time and across the committees. Second, we examine factors that are associated with the appearance of female witnesses. Specifically, we consider the partisanship and gender of committee leadership as well as the gender composition of the whole committee. We also consider whether issues that hearings address are systematically associated with the proportion of female witnesses in congressional hearings.
"Politicizing Information in Policy-Making Procedures via Hearings" with Pamela Ban and Hye Young You
Committee members’ right to choose witnesses to hear on a bill provides political opportunities not only for the members themselves but also for interest groups to lobby to provide their expert knowledge in order to publicly frame the issue in their favor and affect policy outcomes. This mechanism may have been strengthened by Newt Gingrich’s reform of the House in 1994 which resulted in downsizing the research-oriented congressional staffs prompting committees to more heavily rely on interest groups for expert information. Using my novel hearing data back to 1960s, this study investigates whether this change has motivated committee members to increasingly invite and transmit their donor sponsored research products to legislate public policies and help the policy-making procedures as well as potentially policy outcomes dominated by special interest groups and affluent individuals. We will examine this question by analyzing whether the similarity between members’ statements and testimonies of witnesses representing their donor institutions has increased after the reform using regression discontinuity design. Findings of this study will help us better understand the sources of social inequality in American society.
"Better Targeted Messages for Get-Out-the-Vote Campaigns" with Chris Dawes and John Kuk
We study whether minority voters (e.g. female voters, racial minority and naturalized citizens) tend to respond to different GOTV messages than other main stream voters. Based on our preliminary finding that men are more likely to be motivated by civic duty while women are more interested in practicing their voting rights, we argue that voters whose voting rights have been obtained through a harder process historically or personally are more likely to be motivated to turnout when prompted with a campaign message emphasizing the importance of exercising their voting right than others and test the argument through a survey experiment in the 2019 local election in New Jersey as a pilot study and in the 2020 presidential election at a larger scale.
"Inversed Representation: Top Down Issue Framing through Twitter" with Yu-Ru Lin
This project studies whether politicians communicate the framing of an issue to the public via Twitter. We have collected tweets of 600K Twitter users from 2016 until now and the tweets of U.S. congressmen from 2008. Using the Twitter data and supervised-learning method, we are measuring the moral and emotional dimensions of individual tweets on each of the five most controversial issues (e.g. gun control, immigration, abortion, climate change and gay marriage). We will examine whether and how the average moral and emotional dimensions on each issue based on the Democrat and Republican politicians’ tweets, respectively, has changed over time during the period covered in the data and see if these changes are adopted by the liberal and conservative Twitter users’ discourse on each issue.
"Measuring Group Differences Across Party Factions: An Application of TRIBAL Model" with Yu-Ru Lin
In another project with Yu-Ru Lin, we validate our supervised-learning model which is designed to predict group biases using social media data by demonstrating that the model performs well on predicting multiple group membership as well as binary group membership. By extending our application of the model which was originally tested on classifying liberal and conservative tweets, this study predicts individual congressmen's membership of party factions in each Congress by measuring their moral dimensions on each of the five most controversial issues based on their tweets in a given Congress.
"A Model of Abstention with Known Asymmetric Information Quality and Public Messages" with Rebecca Morton
We study voters’ decisions to publicly communicate their private signals before voting when there is a known asymmetry in the quality of signals they receive. We present a game theoretic model and conduct laboratory experiments to test our theoretical predictions. The study suggests that as the asymmetry between the high quality and low quality signals increases sometimes there is a situation where only the voter with the high quality signal sends a message to persuade others while those with low quality signals prefer to just listen. Also, such an effort by the high quality signal holder to inform others tends to have an effect on their vote choices. The effect is strongest especially when both types of signals are not too precise because otherwise those with low types will follow their own signal ignoring the message of the high type if they are different. This study suggests that possessing relatively high quality information may provide the person with a political power to affect final policy outcomes, and this power can be amplified through communication before voting. The real world applications can be found in legislative committees where committee chairs tend to be better informed than other members.
