|Title||Conspiracy and Misperception Beliefs in the Middle East and North Africa|
|C1 Background and Explanation of Rationale||
Misperceptions about foreign powers and different religious and ethnic groups frequently inflame tensions between groups and distort public opinion. In doing so, they may exacerbate intergroup conflict, civil wars, and even interstate wars. In particular, misconceptions about American, Jewish, and/or Israeli actions often play an important role in shaping foreign and domestic policy in the developing world, especially the Middle East and North Africa (MENA). However, we still know relatively little about how people acquire these false or unsupported beliefs and why they seem to hold on to them so tenaciously. This study reports the results of a novel survey experiment examining conspiracy belief using the YouGov MENA panel. We test whether experimentally manipulated feelings of a lack of control increase people's vulnerability to popular conspiracy theories and examine how those conspiracy beliefs are related to other attitudes and characteristics.
|C2 What are the hypotheses to be tested?||
We examine some of the most common and influential conspiracy theories in the MENA region, including anti-Semitic misperceptions denying the Holocaust and alleging Jewish control of the U.S. government as well as foreign policy conspiracy theories claiming that 9/11 was an "inside job" and that the U.S. is supporting the Islamic State in Iraq and Syria (ISIS). This experiment tests whether feelings of a lack of control contribute to belief in conspiracy and misperception beliefs and whether those effects vary by country or the misperception in question. We will also examine the observational relationship between conspiracy and misperception beliefs and other relevant attitudes and characteristics.
The effects of feelings of control - associations in observational data and experimental effects:
H1: Anti-Western attitudes will be positively associated with belief in conspiracy theories and misperceptions.
H2a: Self-reported feelings of control will be negatively correlated with belief in conspiracy theories and misperceptions.
H3a: The association between self-reported feelings of control and belief in the conspiracy theories and misperceptions will vary depending on respondents' attitudes toward the West.
Examining potential country-level differences in conspiracy and misperception belief:
RQ3a: Will conspiracy and misperception belief differ between respondents in Egypt and Saudi Arabia?
The potential effects of the low control treatment on other outcomes of interest:
RQ4a: Will being reminded of feelings of low control decrease support for democracy? (compared to respondents in the high control condition)
|C3 How will these hypotheses be tested? *||
[The study protocol is attached.]
Eligibility and exclusion criteria for participants
Participants from Saudi Arabia are residents of Saudi Arabia (i.e., both citizens and non-citizen residents) who are 18 years or older and members of the YouGov MENA panel. Likewise, participants from Egypt are citizens of Egypt who are 18 years or older and members of the YouGov MENA panel.
We will randomly assign respondents to one of the following conditions (p=1/3 for each):
Data collection and blinding
YouGov MENA will conduct the survey and data collection.
Primary and secondary outcome measures
Our primary outcome variable is a composite measure of conspiracy and misperception belief. This variable combines questions about Holocaust denial and conspiracy theories related to Jewish world domination, the 9/11 terrorist attacks, and the Islamic State/ISIS:
– The United States is secretly trying to help the Islamic State (ISIS) take power in Syria and Iraq.
We will take the mean of these eight outcome measures which are measured on a four-point scale from "Very accurate" to "Not at all accurate" after reverse-coding as appropriate so that higher values indicate greater misperceptions. If the questions used for this measure scale together using principal components factor analysis, we will create a composite measure of conspiracy and misperception belief and use it as our primary outcome measure. However, regardless of how well the questions scale together, we will also analyze responses to each question separately and group them by the misperception or conspiracy in question - i.e., 9/11, Holocaust/Jewish conspiracy, and ISIS. (Note: Though we will analyze the component questions of our primary outcome measure using principal component factor analysis, we will construct any combined outcome measure as a mean for ease of interpretation.)
We will also examine three secondary outcome measures: 1) attitudes towards democracy, 2) sympathy with individuals who fight for the Islamic State, and 3) and a post-treatment measure of anti-Western attitudes.
First, we will construct a composite measure of attitudes toward democracy using the four measures of support for democracy below plus the question about the trade-off between security and stability. Each will be measured on a four-point scale from "Strongly agree" to "Strongly disagree" after reverse-coding as appropriate so that higher values indicate greater opposition to democracy. (If the questions used for this measure do not scale together, we may analyze them separately or consider a subset.)
To what extent do you agree or disagree with the following statements?
Please indicate whether you agree or disagree with the following statement: Respect for human rights in (Egypt/Saudia Arabia) is important, but promoting security and stability is more important.
Sympathy for fighters in the Islamic State will be measured based on responses to the following question, which is again measured on a four-point scale from "Strongly agree" to "Strongly disagree" so that higher values indicate greater sympathy.
Please indicate whether you agree or disagree with the following statement: Even though I don't agree with everything they do, I can understand why someone would go to fight for the Islamic State.
Finally, we will create a composite measure of anti-Western attitudes using responses to the questions below after response-coding as appropriate so that higher values indicate greater anti-Western attitudes. (If the questions used for this measure do not scale together, we may analyze each of them separately.)
Which comes closer to describing your view?
Which of these characteristics do you associate with people in Western countries such as the United
Both attitudes towards democracy and our post-treatment measure of anti-Western attitudes will be analyzed and constructed using principal components factor analysis, which allows for differential weights and also accounts for differing numbers of response options. Because sympathy with individuals who fight for the Islamic State is only one question, we will analyze responses to that measure directly.
For each of our analyses, we will use OLS with robust standard errors. Unless otherwise noted, all experimental treatment effects will be estimated as intent to treat effects.
Our primary model will be pooled regressions for respondents from Egypt and Saudi Arabia with country fixed effects, but we will also estimate pooled regressions fully interacting the country of the respondent with our main explanatory variables. If we cannot reject the null of no difference between countries, we will present the pooled model. If the effects of key explanatory variables differ between countries, we will present separate models by country in the main text for expositional clarity and present interactive models in an appendix. We will estimate marginal effects as appropriate when interaction terms are included in our models. (We will also estimate our treatment effect models with standard demographic controls that include demographic controls for age, sex, religion (Egyptian Coptic Christians), socioeconomic status, education, and an indicator for non-citizen residents of Saudi Arabia along with country fixed effects and report them if the treatment effect estimate is substantively different from the model only including country fixed effects.)
For H1, we estimate the correlation between pre-treatment anti-Western attitudes and conspiracy and misperception belief in a model that controls for fixed effects as well as demographic controls for age, sex, religion (Egyptian Coptic Christians), socioeconomic status, education, and an indicator for non-citizen residents of Saudi Arabia. To measure anti-Western attitudes, we will analyze responses to the following pre-treatment questions using principal component factor analysis and construct a composite measure as a factor score:
Please indicate if you have a very favorable, somewhat favorable, somewhat unfavorable or very
How favorable or unfavorable is your opinion of the following groups?
If the country or group-based measures of anti-Western attitudes do not scale well together, we will analyze each group separately, considering attitudes towards Christians, Jews, Americans, the US, the UK, and Israel separately. (We will also evaluate RQ3a using this model, which allows us to estimate whether belief in conspiracy theories and misperceptions differs between respondents in Egypt and Saudi Arabia conditional on the observable characteristics described above. We will also estimate a version of the model in which we omit the control, religion, socioeconomic status, and education variables, which are arguably post-treatment for country-level difference. If the results are substantively different in this model, we will report it instead.)
For RQ1, we will estimate the same model as H1 but also interact our measure of pre-treatment anti-Western attitudes with either a measure of political knowledge or a measure of respondent education in separate models. The knowledge score will be computed as the sum of correct answers to the following battery measuring both current and historical political knowledge:
Which country is Bashar al-Assad the leader of?
Who is the Supreme Leader of Iran?
Who is the prime minister of the United Kingdom?
Which country first used the atomic bomb?
Which country tore down a wall and reunified itself after the end of the Cold War?
Which leader helped India achieve independence from Britain in 1947?
We will analyze responses to the knowledge scale using factor analysis and construct separate current events and historical knowledge scale measures if two subscales do not scale well together. The education measure will be provided by YouGov from their panel demographics.
For H2a, we estimate the correlation between self-reported feelings of control and conspiracy and misperception belief in a model that controls for fixed effects as well as demographic controls for age, sex, religion (Egyptian Coptic Christians), socioeconomic status, education, and an indicator for non-citizen residents of Saudi Arabia. To measure feelings of control, we will analyze responses to the following pre-treatment questions on a four-point agree/disagree scale using principal components factor analysis and construct a composite measure as a factor score:
-I can solve the problems I have.
(If the items do not scale well together, we may instead construct the control measure using a subset of items with better measurement properties.)
For H2b and RQ2, we will use OLS with robust standard errors to estimate the effects of the low control manipulation on our composite measure of conspiracy and misperception belief. The effect of the low control manipulation will be estimated relative to both the high control manipulation (for H2b) and the placebo (for RQ2). The model will be specified as y=b0+b1*low_control+b2*high_control+b3*saudi_arabia. We will then compute the low control effect as b2-b1 for H2 and examine b1 for RQ2. (We may also estimate the model with the standard set of demographic covariates described above in addition to the KSA fixed effect if including those demographic measures has a substantively important effect on the precision of our treatment effect estimate but in that case both models will be reported.)
In evaluating these findings and H3a/H3b, we will also consider the effects of the treatment on our manipulation check measure:
Some people feel they have completely free choice and control over their lives, while other people feel that what they do has no real control over what happens to them. In general, how much control do you feel you have over what happens in your life?
This outcome variable will be evaluated using the same model/approach as H2b and RQ2. If the treatment has the expected effect, it may increase people's reported feelings of control. However, we are not confident that the treatment will necessarily affect this measure. It is possible that people who feel less in control after the manipulation may seek to alleviate that uncomfortable feeling by reasserting their feelings of control over their life. We will therefore use the manipulation check to provide context to our main tests of H2b and RQ2 described above.
If the low control manipulation has a large and statistically significant effect on our post-treatment measure of self-reported feelings of control (i.e., meets the criteria for a strong instrument), we will also estimate an instrumental variables regression that uses assignment to the low control condition as an instrument for post-treatment self-reported feelings of control. We will then estimate a local average treatment effect (LATE) of variation in feelings of control for people whose feelings of control were affected by the experimental manipulation.
For H3a, we estimate whether the association between pre-treatment self-reported feelings of control and conspiracy and misperception belief is moderated by pre-treatment anti-Western attitudes in a model that also controls for fixed effects as well as demographic controls for age, sex, religion (Egyptian Coptic Christians), socioeconomic status, education, and an indicator for non-citizen residents of Saudi Arabia. (This model will again use the measures of feelings of control and anti-Western attitudes constructed according to the procedure described above for H2a and H1, respectively.)
For H3b, we will estimate a similar model to H2b but interact our indicators for our low and high control conditions with our measure of pre-treatment anti-Western attitudes as used in H1. We will use this interaction to see if the effect of our treatments on conspiracy and misperception belief is moderated by anti-Western attitudes. The model will be specified as y=b0+b1*low_control+b2*high_control+b3*anti_Western_attitudes+b4*low_controlXanti_western+b5*high_controlXanti_western+b6*saudi_arabia. We will then compute the marginal effect of low versus high control over the range of anti-Western attitudes using a procedure similar to that described in Brambor et al. 2006 but implementing the robustness checks and recommendations of Hainmeuller et al. (N.d.). (Note: We may again estimate the model with the standard set of demographic covariates described above in addition to the KSA fixed effect if including those demographic measures has a substantively important effect on the precision of our treatment effect estimate but in that case both models will be reported.)
To test RQ3b, we will also evaluate whether the effect of reminders of feelings of low control on belief in conspiracy theories and misperceptions differs between respondents in Egypt and Saudi Arabia (compared to respondents in the high control condition). This model will be specified as y=b0+b1*low_control+b2*high_control+b3*saudi_arabia+b4*low_controlXsaudi_arabia+b5*high_controlXsaudi_arabia. We will estimate the marginal effect of low control (versus high) for both Egypt and Saudia Arabia and also evaluate whether the difference between the marginal effects is statistically significant.
Finally, our analyses for RQ4a-RQ4c will repeat the analysis described above for H2b but with the dependent variable as attitudes towards democracy (RQ4a), sympathy for fighting for the Islamic State (RQ4b), and attitudes toward the West (RQ4c), which will be measured as described above.
(All results will be unweighted. Regression results for individual ordered dependent variables will be verified for robustness using ordered probit. The order of hypotheses and analyses in the final manuscript may be altered for expositional clarity.)
|C4 Country||Egypt, Saudi Arabia|
|C5 Scale (# of Units)||n=1000 in Egypt and n=1000 in KSA|
|C6 Was a power analysis conducted prior to data collection?||No|
|C7 Has this research received Insitutional Review Board (IRB) or ethics committee approval?||Yes|
|C8 IRB Number||Dartmouth College STUDY00029363, American University IRB-2016-234|
|C9 Date of IRB Approval||Dartmouth 3/2/16, American 2/18/16|
|C10 Will the intervention be implemented by the researcher or a third party?||YouGov MENA|
|C11 Did any of the research team receive remuneration from the implementing agency for taking part in this research?||No|
|C12 If relevant, is there an advance agreement with the implementation group that all results can be published?||not provided by authors|
|C13 JEL Classification(s)||not provided by authors|