“…However, the small sample size might overestimate the effect size. Therefore, as a replication convention [ 9 , 10 ], we halved the effect size of the original experiment (Cohen's f = 0.151), and used G * Power [ 11 ] to conduct a power analysis (Cohen's f = 0.151, α = 0.05, 1− β = 0.99, number of groups = 3), as a result of which we set N = 942 as the required sample size. Moreover, considering that many participants are often excluded in online surveys because of satisficing [ 12 – 14 ], and the dropout rate from the second wave, we have set almost double the required sample size ( N = 1890) as the maximum sample size.…”