Excuses and Social Image (Part 2: Interpretation of Stigmatized Expression)

2020-02-18

2020-02-24

Respondent's guesses as to their matched respondent's score on the Foreign Culture Tolerance Scale and the Gullibility Scale (experiment 1)

Respondent's decision of whether or not to authorize a $1 bonus to their matched respondent and their guess of whether or not their matched respondent donated to the organization (follow-up)

Open-ended text: "Why do you think your matched respondent chose to donate to Fund the Wall"?

See the text analysis pre-analysis plan and Python code file attached to this submission.

For the first experiment, we recruit respondents through Lucid, a survey provider. For the follow-up experiment, we recruit them through Prolific, another survey provider.

We randomize respondents into one of two conditions: "excuse" or "no excuse". We provide respondents with information about a recent study (Lott 2018) that finds that undocumented immigrants in Arizona commit crimes at substantially higher rates than comparable US citizens. We (truthfully) tell participants that they have been matched with another participant who chose to authorize a donation to Fund the Wall (www.fundthewall.com), an organization that supports the proposed US-Mexico border wall. Participants in the "no excuse" group are told that their matched respondent did not see the information about Lott's study before donating, whereas participants in the "excuse" group are told that their matched respondent did see the information.

Participants are then cross-randomized into one of two conditions: "bias" or "gullibility". Participants in the "bias" group are asked to guess their matched respondent's score on a Foreign Culture Tolerance Scale, which measures "tolerance toward foreign values and traditions" on a scale from 0-100. Participants in the "gullibility" group are asked to guess their matched respondent's score on a Gullibility Scale, which measures "how easily people are manipulated by evidence from untrustworthy sources" on a scale from 0 to 100. Both guesses are incentivized: participants are told that if they guess correctly, they will be entered into a lottery for a $50 Amazon gift card.

Our follow-up experiment is similar, but differs in the following respects. First, we tell participants that they have been matched with another participant who chose to join a campaign either to illegally deport all illegal immigrants or to oppose defunding the police, and authorized a Tweet encouraging their followers to do the same (see AEARCTR-0007379 for additional details on this Tweet). Participants in the "no excuse" group see a Tweet indicating that the matched respondent saw the rationale -- a video or article in support of the action -- after making the decision to join the campaign, while those in the "excuse" group see a Tweet indicating that the matched respondent saw the rationale beforehand. They then have the opportunity to authorize a $1 bonus to their matched respondent. Finally, they guess whether their matched respondent donated to an organization related to the cause.

References: Lott, John R., Undocumented Immigrants, U.S. Citizens, and Convicted Criminals in Arizona (February 10, 2018). Available at SSRN: https://ssrn.com/abstract=3099992 or http://dx.doi.org/10.2139/ssrn.3099992

Qualtrics will randomize respondents into one of the four conditions. We use the "Evenly Presents Elements" option in the Qualtrics Randomizer to secure equal group sizes.

Individual

No

In the first experiment, we target 3000 individuals (5163 individuals when pooling with pilot)

In the second experiment, we target 1000 individuals per topic (illegal immigration or defunding the police) for a total of 2000 individuals.

In the first experiment, we target 3000 individuals (5163 individuals when pooling with pilot) In the second experiment, we target 1000 individuals per topic (illegal immigration or defunding the police) for a total of 2000 individuals.

In the first experiment, we target 750 individuals in excuse/gullibility, 750 individuals excuse/bias, 750 individuals no excuse/gullibility, 750 individuals no excuse/bias. To gain statistical precision, we also report some specifications in which we pool with additional pilot data (N=2163, with 517 excuse/gullibility, 510 excuse/bias, 502 no excuse/gullibility, and 511 no excuse/bias). In the second experiment, we target 500 individuals per treatment arm (excuse/no excuse) per topic (illegal immigration/defunding the police).

In the first experiment, we ask our survey provider to restrict the survey to respondents who had not taken our pilot. To further ensure that our sample does not include repeat respondents, we also include a post-outcome question asking respondents whether they have taken a previous online survey that discussed the Lott study. We expect this number to be small and plan to exclude respondents who respond in the affirmative from our main specifications (though we will also report results with the entire sample).

Based on pilot results, our MDE is approximately 3.6 on our cultural and gullibility scales (which range from 0 to 100).

Text analysis preanalysis plan