Experimental Design Details
Recruitment:
Participants will be recruited by Amazon’s Mechanical Turk; however, only participants with an IP address from Turkey will be eligible to participate. Participants will be directed to Qualtrics to complete the survey.
HITs (what the assignment is called on MTurk that workers can choose to accept) will be sent out in batches of 9 every hour in order to attempt to account for temporal bias.
The proposed length of the survey is as long as it is to account for the possible difficulties in obtaining responses. Unfortunately, there are only around 2000 mTurk workers that operate in Turkey so it will be more difficult to recruit participants than if I were using mTurk in the United States. It is possible that I experience no issues and recruit 250 participants quickly. However, if this is not the case the following protocol for recruitment will be used in attempt to recruit the missing number of participants.
1. mTurk will be used again with different recruitment wording.
2. Other crowdsourcing websites will be used, such as CrowdFlower and Prolific Academic.
Consent:
Implied consent will be given if the participant selects ‘agree’ on the first page, which asks if the participant is willing to participate. Participants will then be randomized into either the treatment group or control group and will subsequently complete their respective task.
Treatment:
Participants will be provided with a list of 10 values, which they will be asked to rank in order of most important to them to least important to them. The initial order of the values will be randomized to ensure there is no effect of initial order. Participants will then be asked to describe why the most important value to them is important to them. According to affirmation theory, this will allow participants to experience the affirmation effect.
Control:
Participants will be asked to describe everything that they have eaten or drank in the past 48 hours. This is meant to serve as a placeholder for the treatment experience so that there is no effect on the amount of time the survey takes overall on the results.
After completing their respective tasks, participants will move to the section where the dependent variables are measured.
Dependent Variables:
There are two dependent variables of interest, which measure prejudice towards Kurds and prejudice towards Syrians, which will be measured by asking participants six questions, stated below. In addition to the two questions for the dependent variables, participants will also be asked four questions that are not of particular interest to the researcher. However, they are meant to mask which questions are of interest to the researcher to avoid demand characteristics. All questions will involve participants ranking how much they agree with each statement on a one-hundred-point Likert scale. The first question, which is a modified version of two questions developed by Meertens and Pettigrew (1995), will state “Kurds fail to be successful in Turkey because they do not attempt to learn Turkish values and skills.” The second question, modified from Pereira et al. (2010), will state “Turkey’s cultural life is generally undermined by people coming here from Syria.” The following four statements will be used to attempt to mask the questions of interest: “European states should allow Turkey to join the European Union,” “States that have a shoreline to the Black Sea should cooperate economically,” “Turkey should try to host the Olympics,” and “The Mediterranean states should cooperate economically.” The order of the dependent variable questions will be randomized to ensure that the order has no effect on results.
Demographic Questions:
The following covariates will be collected after the treatment and dependent variables are collected: age, sex, how long it took participants to complete either the control or treatment exercise, how much they identify as a Turk, how content they are with their economic condition, how religious they are, in the province in Turkey in which they reside, education level and ethnicity. These will be collected after the dependent variables are measured so that these identity questions do not inadvertently prime participants’ identities.
After the demographic questions are collected, participants will be debriefed using vague language (but adequately according to Skidmore’s IRB). Vague language is necessary due to the controversial nature of this experiment.