Fig. 6: Study 4, norm vs monetary condition in India, by language prime (English or Hindi).

Effects of a monetary incentive (green) and a social norm treatment (blue) in India (N = 2,065 participants recruited on Facebook), by assigned language (English or Hindi). a, The central tendency and distribution of effort by language and incentive conditions. The black line within each box represents the median and the red dot shows the mean; upper and lower bounds show the third and first quartiles, respectively; whiskers represent 1.5× the interquartile range, with black dots showing observations outside of this range. The width of each violin corresponds to the frequency of observations at any given number of images rated on the y axis. The interaction between language and incentive in a multiple linear regression model is statistically significant (b = 10.69, t(2,061) = 3.31, P = 0.001, 95% CI 4.35–17.02). b, The money advantage, that is, how much more effective the monetary condition is compared to the social norm condition. c, The central tendency and distribution of the cost-effectiveness (effort per dollar spent) by language and incentive. Graph elements are analogous to those in a, with the width of each violin corresponding to the frequency of observations at any given level of cost-effectiveness (effort per dollar spent) rated on the y axis. The monetary incentive is more cost-effective than the social norm condition in English (two-sided Welch’s t(919.21) = 4.37, P < 0.001, P_Bonf < 0.001, Mean_difference = 4.34, d = 0.27, 95% CI 2.39–6.28), but the two incentives do not significantly differ in their cost-effectiveness in Hindi (two-sided Welch’s t(915.62) = 0.30, P = 0.761, P_Bonf = 1.000, Mean_difference = 0.30, d = 0.02, 95% CI −1.64 to 2.24). In b, error bars are bootstrapped 95% CIs for the mean relative difference in the number of images rated in the norm vs monetary condition.