Skip to main content
. 2024 Jan 8;6:e2. doi: 10.1017/ehs.2023.30

Table 3.

Raters’ guesses about players’ cooperative behaviour.

Panel A: Summary statistics
Treatment (N) Guessed split Average correct Accuracy
Round 1 guesses (actual Split = 74.5%)
None 63.6 56.1 −0.115
(108) (27.2) (14.0) (0.546)
Label 61.0 58.7 0.238
(101) (15.9) (8.6) (0.363)
Photo 56.1 55.9 0.214
(108) (16.8) (8.7) (0.348)
Video 54.0 53.4 0.113
(105) (13.2) (8.5) (0.362)
All 58.7 56.0 0.110
(422) (19.4) (10.4) (0.436)
Difference 16.77*** 13.49** 48.40***
Round 2 guesses (actual Split = 53.2%)
None 60.0 60.2 0.536
(108) (18.4) (8.2) (0.451)
Label 60.8 60.7 0.561
(101) (14.6) (7.9) (0.438)
Photo 59.0 62.5 0.674
(108) (12.1) (6.3) (0.351)
Video 55.4 61.6 0.606
(105) (11.1) (6.5) (0.352)
All 58.8 61.3 0.595
(422) (14.5) (7.3) (0.402)
Difference 8.58* 6.55 7.42

Note: Values for guessed Split, correctness, and actual Split are percentages. Standard deviations are in parentheses. Difference reports the results of the Wald test that the treatment dummy coefficients in generalised linear model regression are equal: chi-squared with 3 degrees of freedom reported; *** p < 0.001, ** p < 0.01, * p < 0.05. These test results are robust using Kruskal–Wallis.