global preferences survey
elicit risk, time, and social preferences
About the Global Preferences Survey (GPS)
The Global Preferences Survey is a globally representative dataset on risk and time preferences, positive and negative reciprocity, altruism, and trust. We collected these preference data as well as a rich set of covariates for 80,000 individuals, drawn as representative samples from 76 countries around the world, representing 90 percent of both the world’s population and global income. Data collection was financially supported by the European Research Council through ERC Starting Grant No. 209214, which is gratefully acknowledged.
The data is described in: Falk, A., Becker, A., Dohmen, T., Enke, B., Huffman, D., & Sunde, U. (2018). Global evidence on economic preferences. Quarterly Journal of Economics, 133 (4), 1645–1692.
The GPS data were collected within the framework of the 2012 Gallup World Poll, a survey that includes representative population samples in a large number of countries, and asks about social and economic issues, on an annual basis.
The data collection process consisted of four steps. First, an experimental validation procedure was conducted to select the survey items. Second, there was a pre-test of the selected survey items in a variety of countries to ensure implementability in a culturally diverse sample. Third, the survey items (for details see: Armin Falk, Anke Becker, Thomas Dohmen, David Huffman, and Uwe Sunde. The preference survey module: A validated instrument for measuring risk, time, and social preferences. IZA Discussion Paper No. 9674, 2016) were translated and quantitative amounts were adjusted to ensure comparability across countries. Fourth, the final data set was collected through the regular professional data collection efforts in the framework of the World Poll 2012.
One important feature of the GPS data is that it measures preferences for a nationally representative sample for each of the 76 countries covered. Thus, it is possible to study how preferences vary within the population of a given country, and also to construct country level averages, shedding light on how preferences vary across countries. The median sample size was 1,000 participants per country. Respondents were selected through probability sampling; ex post representativeness of the data can be achieved using weights provided by Gallup. In total, the sample involves preference measures for more than 80,000 participants worldwide.
The 76 countries included in the GPS constitute a geographically and culturally diverse set of nations. They were chosen with the aim of providing a globally representative sample. The collection of countries spans all continents, various cultures, and different levels of development. Specifically, it includes 15 countries from the Americas, 25 from Europe, 22 from Asia and Pacific, as well as 14 African countries, 11 of which are Sub-Saharan. This set of countries covers about 90% of both the world population and global income.
Development of Preference Module
The preference module is a concise, experimentally-validated survey module for measuring risk aversion, time discounting, trust, altruism, positive and negative reciprocity. These preferences affect individuals’ choices in myriad situations. The module is a convenient tool for obtaining standardized measures in all popular methods of data collection. It is therefore a useful tool in a wide range of applications, not least because preference measures can allow for improved prediction of many important economic behaviors, or can provide control variables if researchers want to identify causal effects of other factors that are correlated with preferences.
The development of the preference module is described in:
Falk, A., Becker, A., Dohmen, T. J., Huffman, D., & Sunde, U. (2016). The preference survey module: A validated instrument for measuring risk, time, and social preferences. IZA Discussion Paper No. 9674, forthcoming in Management Science. [download paper] [web appendix.]
Time Preference/Patience. The measure of time preference is derived from the combination of responses to two survey measures, one with a quantitative and one with a qualitative format. The quantitative survey measure consists of a series of five interdependent hypothetical binary choices between immediate and delayed financial rewards. In each of the five questions, participants had to decide between receiving a payment today or larger payments in 12 months. The qualitative measure of patience is given by the respondents’ self-assessment regarding their willingness to wait on an 11-point Likert scale, asking “how willing are you to give up something that is beneficial for you today in order to benefit more from that in the future?”
Risk Preference. Risk preferences were elicited through a series of related quantitative questions as well as one qualitative question. Just as with patience, the quantitative measure consists of a series of five binary choices. Choices were between a fixed lottery, in which the individual could win x or zero, and varying sure payments, y. Choice of the lottery resulted in an increase of the sure amount being offered in the next question, and vice versa, thereby zooming in around the individual’s certainty equivalent. The qualitative item and the outcome of the quantitative staircase measure were combined through roughly equal weights.
Positive Reciprocity. Respondents’ propensities to act in a positively reciprocal way were measured using one quantitative item and one qualitative question. First, respondents were presented a choice scenario in which they were asked to imagine that they got lost in an unfamiliar area and that a stranger – when asked for directions – offered to take them to their destination. Respondents were then asked which out of six presents (worth between 5 and 30 euros, or the respective country-specific equivalents) they would give to the stranger as a “thank you”. Second, respondents were asked to provide a self-assessment about how willing they are to return a favor on an 11-point Likert scale. These two items receive roughly equal weights.
Negative Reciprocity. Negative reciprocity was elicited through three self-assessments. First, respondents were asked how willing they are to take revenge if they are treated very unjustly, even if doing so comes at a cost (Likert scale, 0-10). The second and third item probed respondents about their willingness to punish someone for unfair behavior, either towards themselves or towards a third person. This last item captures prosocial punishment and hence a concept akin to norm enforcement. These three items receive weights of about one third each.
Altruism. Altruism was measured through a combination of one qualitative and one quantitative item, both of which are related to donations. The qualitative question asked respondents how willing they would be to give to good causes without expecting anything in return on an 11-point scale. The quantitative scenario depicted a situation in which the respondent unexpectedly received 1,000 euros and asked them to state how much of this amount they would donate. These two items were weighted about equally.
Trust. The trust measure is based on one item, which asked respondents whether they assume that other people only have the best intentions (Likert scale, 0-10). The item was a strong predictor of trusting behavior in incentivized trust games, in the survey design stage. Time constraints and the fact that there already exists a global measure of trust in the World Values Survey (WVS) data set determined the choice to have only one item measuring trust.
The preference module provides a valuable contribution in several respects. First, while incentivized experiments are generally viewed as the gold standard for eliciting preferences, because they measure actual behavior in a controlled way, they are expensive and time consuming; our survey module is developed based on its ability to capture behavior in incentivized experiments. It thus provides a low cost way to measure preferences in large representative samples, while retaining key advantages of the experimental approach.
Second, while some existing survey measures have been shown to predict behavior in experiments, these have typically been developed for different individual studies, based on intuitive plausibility, which has led to a large set of diverse measures. Our approach takes the natural next step, assessing which of a wide array of candidate measures are the best predictors. Specifically, for each preference, we evaluate the ability of roughly 30 different survey measures to predict behavior in corresponding incentivized choice experiments. We include measures developed in previous studies, as well as novel measures. We select the survey items that are jointly the best predictors.
Third, the previous literature has used a wide variety of different types of preference measures, with different formats, wordings, and elicitation modes. This poses a substantial obstacle to the comparability of results across studies. Our preference module provides a new tool for measuring preferences in a standardized way. In psychology, this type of problem has been at least partially addressed by the development of a standard set of measures - in particular the so-called Big Five and Locus of Control - which are conceptualized to be key personality traits and relevant for many settings.
Our module captures a set of preferences identified by economic theory as being fundamental determinants of behavior in many contexts. If the module is widely adopted, it will help enhance comparability across studies and may thereby accelerate scientific progress.
The selected module involves two survey items for the elicitation of each preference. The preference module is symmetric, in that most preferences are measured with one quantitative and one qualitative item. The single best predictor of behavior for a given preference tends to be quantitative: a hypothetical version of the experiment itself. The second survey item that is typically selected is a qualitative question, asking about a general orientation in the relevant preference dimension. The module thus offers an attractive balance between measures that allow for inferring (cardinal) preference parameters, and subjective measures that capture other contexts besides choices about financial rewards.
A notable feature of the preference module is its symmetry: For most preference dimensions, it contains a measure based on a hypothetical choice experiment and a qualitative item. These two types of measures are complementary in the sense that the quantitative measure is akin to the standard revealed preference approach whereas the qualitative item is a subjective self-assessment. Previous research has shown that subjective assessments with abstract framings can lead to strong all around predictors of life choices across many different life contexts.
Quantitative survey measures that involve explicit monetary stakes are no exception, as they are somewhat tied to the context of financial decision making by construction; they may be better predictors of financial decisions in life than qualitative measures of a general disposition, but less predictive of choice in other domains. The preference module has an attractive balance between both approaches.
