Witchcraft beliefs around the world: An exploratory analysis

This paper presents a new global dataset on contemporary witchcraft beliefs and investigates their correlates. Witchcraft beliefs cut across socio-demographic groups but are less widespread among the more educated and economically secure. Country-level variation in the prevalence of witchcraft beliefs is systematically linked to a number of cultural, institutional, psychological, and socioeconomic characteristics. Consistent with their hypothesized function of maintaining order and cohesion in the absence of effective governance mechanisms, witchcraft beliefs are more widespread in countries with weak institutions and correlate positively with conformist culture and in-group bias. Among the documented potential costs of witchcraft beliefs are disrupted social relations, high levels of anxiety, pessimistic worldview, lack of entrepreneurial culture and innovative activity.

Notes. Summary statistics are shown for the sample of people who gave a "yes" or "no" response to the witchcraft question. In addition to missing data for some respondents, several questions were not asked in certain survey waves. Specifically, the personal economic situation question was not asked in Central and Eastern Europe and the U.S., the urban location and household size variables are missing in the Western Europe wave, and the belief in god question is phrased differently and missing in the World's Muslims and the U.S. surveys, respectively.

Witchcraft beliefs
Personal belief in witchcraft. A dummy variable coding "yes" (1) and "no" (0) answers to the following question: "Do you believe in the evil eye, or that certain people can cast curses or spells that cause bad things to happen to someone?" Source: Pew Research Center surveys.
Prevalence of witchcraft beliefs at the country level. The fraction of respondents who claim to believe "in the evil eye, or that certain people can cast curses or spells that cause bad things to happen to someone" relative to the total number of respondents. Computed at the country level using individual-level survey weights provided for aggregation purpose. Source: Pew Research Center surveys.

Socio-demographic characteristics
All socio-demographic characteristics of respondents are constructed and harmonized based on the original surveys listed in table A.1.
Age. Age of respondent in tens of years.

Gender. A dummy variable equal to 1 (0), if female (male).
Location of residence. A dummy variable equal to 1 (0) for urban (rural) locations.

Education.
A categorical variable classifying the data on self-reported educational attainment into three categories: primary or less, secondary, above secondary.
Personal economic situation. A categorical variable reflecting respondents' assessment of their personal economic situation on the following scale: very bad, somewhat bad, somewhat good, very good.
Religious affiliation. A categorical variable capturing religious affiliation or its absence: Christian, Muslim, unaffiliated (including agnostics and atheists). About 0.5% of respondents representing all other religions are excluded from the sample when using this variable.

Importance of religion.
A categorical variable capturing self-reported importance of religion in life: not at all important, not too important, somewhat important, very important.
Belief in god. A dummy variable equal to 1, if the respondent claims to believe in god, and 0, if not.

Baseline control variables
Continental fixed effects. A set of dummy variables indicating the belonging of a given country to one of the following world regions (total number of countries indicated in parentheses): Africa (24), Americas (20), Asia (18), Europe (33).
Absolute latitude. Absolute latitude of the country centroid. Source: Nunn and Puga (2012).
Distance to the coast. Average distance (in thousands of kilometers) to the nearest ice-free coast.
Agricultural suitability of land. Caloric suitability index capturing average potential agricultural output (measured in calories) based on crops that were available for cultivation in the post-1500CE era.
Source: https://ozak.github.io/Caloric-Suitability-Index/, based on Galor andÖzak (2016). Efficiency of the tax administration. Captures the efficiency of collecting corporate and household income taxes, the ability to collect taxes across the entire state territory and limit tax evasion. Source:

Religiosity
Institutional Profiles Database (2012). Confidence in local police, judicial system and courts, national government. The share of survey respondents expressing confidence in respective institutions; averages of the available data up to 2020. Source: own calculations based on the Gallup World Poll data.
Autonomy vs. embeddedness. A scale capturing the extent to which people are autonomous rather than embedded in their groups. Calculated as the difference between the average of "affective" and "intellectual" autonomy scores and embeddedness score. Autonomous cultures "encourage people to cultivate and express their own preferences, feelings, ideas, and abilities, and to find meaning in their own uniqueness." Intellectual autonomy "encourages individuals to pursue their own ideas and intellectual directions Individualism "can be defined as a preference for a loosely-knit social framework in which individuals are expected to take care of only themselves and their immediate families." Collectivism "represents a preference for a tightly-knit framework in society in which individuals can expect their relatives or members of a particular ingroup to look after them in exchange for unquestioning loyalty." Source: Hofstede et al.

Migrant acceptance index. Gallup's migrant acceptance index is based on three questions. Respondents
are asked whether the following situations are "good things" or "bad things": immigrants living in their country, an immigrant becoming their neighbor and immigrants marrying into their families. "A good thing" response is worth three points in the index calculation, a volunteered response of "it depends" or "dont know" is worth one point, and "a bad thing" is worth zero points. The index is a sum of the points across the three questions. The higher the score, the more accepting the population is of migrants. Source: Gallup World Poll, 2016-2017.

Social relations, anxiety, and worldview
Generalized trust. Share of respondents replying that "people can be trusted" in the generalized trust question: "Generally speaking would you say that most people can be trusted or that you cant be too careful in dealing with people?" Averages across available years. Source: own calculations based on Pew

Research Center surveys, WVS/EVS, and Gallup World Poll (as recorded in the 2019 World Happiness
Report database).

Trust in neighbors.
Based on the following survey question: "How much do you trust the people in your neighborhood?" Possible answers are: a lot (4), some (3), not much (2), not at all (1); data for the year 2018. Source: own calculations based on the Gallup World Poll.
Out-group trust. See the definition of the "in-vs. out-group trust" variable above.
Trusted business partner. The share of respondents who believe they can find someone outside their own family to be a trusted business partner; average across available years. Source: own calculations based on the Gallup World Poll.
Generalized fairness. Based on the WVS/EVS question: "Do you think that most people would try to take advantage of you if they got the chance, or would they try to be fair?" Answers range on a 0-10 scale, from "most people would try to take advantage of me" (0) to "most people would try to be fair" UNDP Human Development Report (2020) database.

Exposure to misfortunes
Exposure to natural disasters. Share of population physical exposed to earthquakes, storms, floods, droughts, and sea-level rise. Source: WorldRiskReport (2020).
Exposure to agricultural drought. An index of exposure to agricultural drought based on historical climate conditions. Source: Meza et al. (2020).

Pathogen richness. The number of all infectious diseases listed for a given country in the Global
Infectious Disease and Epidemiology Network; April-August 2007. Source: Fincher and Thornhill (2008).    Notes. The binary dependent variable is personal belief in witchcraft. Ordinary least-squares estimates from the linear probability regressions are reported in all columns. Standard errors clustered by country are shown in parentheses. * * * , * * , and * denote statistical significance at the 1, 5, and 10 percent level, respectively. Country fixed effects are included in all specifications. Age is measured in tens of years. The number of observations and countries for each specification reflects data availability constraints.