| Abstract: | The deleterious effects of joint bias in favor of statistical inference and against replication are becoming well known. The acceptance of numerous Type I errors into the literature is by far the most serious of these. Data on the contents of three major journals support the contention that a joint bias for statistical significance tests, for rejections, and against replication exists in modern sociology. This finding replicates that of Sterling (1959) for psychology. A speculative analysis of the dynamics of publication decisions suggests that a compact format for reporting replications might make their publication more attractive to editors, and thus increase their frequency in the literature. A possible format for briefly reporting replication studies is suggested. |
