In this project we have embedded three different survey experiments to tackle three different theoretical problems. (1) Social media effects on affective polarization; (2) the effects of priming political polarization in individual affective polarization; (3) Social sorting and affective polarization.

In the codebook and questionnaire all the variable names  from the three experiments described below start by the prefix 'esmP' which indicates experimental variables. 

Experiment Wave 1

Experiment embedded in the first wave of the TRI-POL project. The purpose of this experiment was to test the effect of exposure to different Twitter accounts on a set of relevant political attitudes, such as political interest, affective and ideological polarization and political trust. Participation was restricted via invitation. Specifically, respondents were invited to follow one or two Twitter accounts from a list provided to them during the next seven days. Two experimental groups were created with a different list of Twitter accounts. Assignment to the first list, containing the political leaders’ accounts of the main parties, or the second one, with a list of institutional accounts, was randomized by a computer algorithm. After seven days, respondents who participated in the experiment were re-contacted, answered some question about their exposure to and the content of the selected Twitter accounts, and completed the survey questionnaire about their political attitudes and opinions. To verify respondents’ activity in Twitter, information is collected with a passive behavioural metre. The prefix 'esmP' indicates experimental variables.

Experiment Wave 2

This experiment is design to study examines the effects of priming political polarization or populist political frames on political polarization as measured in interpersonal trust discrimination via behavioral games (ie. trust game) and measures of political affect (feeling thermometers).

Experiment Wave 3

Conjoint experiment embedded in the third wave of the TRI-POL project and carried out in five different countries: Italy, Spain, Portugal, Chile and Argentina. The purpose of the experiment is to prove the social sorting behind social partisan identity.

 Respondents are asked to choose the basic characteristics of a hypothetical family unit moving respondents’ next door. Specifically, I use a fully randomized conjoint experiment that varies the attributes presented with respect to 10/11 (depending on the country) dimensions shared by the neighboring families: territorial identity; ideology; immigrant; sex orientation; party supporter; education; environmentalist; pet owner; religion; politicisation; and language (for the Spanish case) or attitudes towards vaccination (for the Italian case).

In each round or task, respondents are shown two neighbor’s profiles, which both display the same dimensions but then vary the attributes within each dimension. For each task, respondents are required to choose between the two proposals presented to them.