Replication data Conjoint and Trustgame

The data are related to two experimental studies dealing with question how politics influence inter-personal interactions. The first one is conjoint analysis with following hypotheses.

H1: Individuals prefer to talk to others who match a preferred political party and reject those who match the least preferred party
H2: Individuals prefer to talk to people with a similar attitude towards immigration and reject those with a different attitude
H3a: Individuals prefer to talk less to those people who frequently talk about politics and vote for the individual’s disliked parties, compared to voters of liked parties
H3b: Individuals prefer to talk less to those people with different attitudes on immigration who frequently talk about politics
H4: Individuals prefer to talk less to people with different values

The data used for test were obtained in online survey. The experiment was fielded by Focus Marketing and Social Research between May 22 and June 3, 2019, on a sample of 1032 adult Czech respondents (a non-probability sample using quotas on gender, education, age, region, and the size of settlement). Each respondent evaluated five pairs of fictional communication partners’ profiles; the total number of cases used in the analysis was 10,250.

Data are stored in file conjoint_data.csv.
Complete analysis is covered by document conjoint_trustgame_replication.r

The second study is trustgame with similar hypotheses as in the first study:
H5a. When Player 2 is a supporter of the same party as Player 1, the amount of money allocated to Player 2 by Player 1 increases in a trust game
H5b. When Player 2 is a supporter of Player 1’s least preferred party, the amount of money allocated to Player 2 by Player 1 decreases in a trust game
H6a. When Player 2 has the same attitude on immigration as Player 1, the amount of money allocated to Player 2 by Player 1 increases in a trust game
H6b. When Player 2 has the opposite attitude towards immigration as Player 1, the amount of money allocated to Player 2 by Player 1 decreases in a trust game
H7a. When Player 2 is a member of the same random nonpolitical group as Player 1, there is little effect on the amount of money allocated to Player 2 by Player 1
H7b. When Player 2 is a member of a different random nonpolitical group as Player 1, there is little effect on the amount of money allocated to Player 2 by Player 1

The study was conducted between December 4 and 17, 2019, on a sample of 946 adult respondents. The data was collected by the same company as in the previous study, based on the same quota measures. Respondents participating in the conjoint experiment were excluded from the pool of respondents. Each respondent played 9 rounds of the game, with resulted in a total of 8,461 cases. To estimate effects , we used multilevel tobit regression model for a dependent variable with its range limited to between 0 and 100 CZK.

Data are stored in file trust_stata.xlsx.

Replication file is trustgame.do.

本数据集关联两项围绕「政治因素如何影响人际互动」主题开展的实验研究。
第一项实验为联合分析（conjoint analysis），提出如下研究假设：
H1：个体更倾向于与所属党派契合自身偏好的他人交流，而排斥与自身最不偏好党派匹配的交流对象。
H2：个体更倾向于与在移民议题上态度相似的他人交流，而排斥态度相异的交流对象。
H3a：相较于支持自身偏好党派的对象，个体更不愿与频繁谈论政治且支持自身厌恶党派的人群交流。
H3b：相较于在移民议题上态度相似的对象，个体更不愿与在移民议题上态度相异且频繁谈论政治的人群交流。
H4：个体更不愿与价值观相异的他人交流。

用于检验上述假设的数据来源于线上调研。该实验由Focus Marketing and Social Research公司于2019年5月22日至6月3日期间开展，调研样本为1032名成年捷克受访者（采用基于性别、教育程度、年龄、地区及定居点规模的配额抽样非概率样本）。每位受访者需评价5组虚构的交流伙伴画像，最终用于分析的有效案例总数为10250例。

数据集存储于conjoint_data.csv文件中，完整的复现分析代码包含于conjoint_trustgame_replication.r文档内。

第二项实验为信任博弈（trust game），其研究假设与第一项实验类似：
H5a：在信任博弈中，当玩家2与玩家1同属某一党派的支持者时，玩家1分配给玩家2的资金数额会增加。
H5b：在信任博弈中，当玩家2为玩家1最不偏好党派的支持者时，玩家1分配给玩家2的资金数额会减少。
H6a：在信任博弈中，当玩家2与玩家1在移民议题上态度一致时，玩家1分配给玩家2的资金数额会增加。
H6b：在信任博弈中，当玩家2与玩家1在移民议题上态度相悖时，玩家1分配给玩家2的资金数额会减少。
H7a：在信任博弈中，当玩家2与玩家1同属某一随机非政治群体时，玩家1分配给玩家2的资金数额几乎不受影响。
H7b：在信任博弈中，当玩家2与玩家1分属不同的随机非政治群体时，玩家1分配给玩家2的资金数额几乎不受影响。

该实验于2019年12月4日至17日期间开展，调研样本为946名成年受访者，数据收集由与第一项实验相同的公司完成，采用相同的配额抽样标准。参与过联合分析实验的受访者被排除在本次样本池之外。每位受访者需参与9轮博弈，最终有效案例总数为8461例。为估计效应量，我们采用多层Tobit回归模型（multilevel Tobit regression model）对取值范围限定为0至100捷克克朗（CZK）的因变量进行分析。

数据集存储于trust_stata.xlsx文件中，复现分析代码为trustgame.do文件。