摘要：Some scholars have argued for fixed standards for null hypothesis significance testing, using only .05 as the criterion for assessing or reporting statistical significance. One might get an impression that no credible scholars have ever succeeded in publishing results in high-impact factor, peer-reviewed journals using any other criteria. Here, we point out that not only are there sound theoretical arguments for flexibility in the choice of alpha or the reporting of statistical results, but that peer-reviewed articles have been published both recently and in the past in major social science research journals using more flexible statistical standards. In particular, we review the reporting of statistical trends (p < .10) in seven major scholarly journals from 2005 to 2009, as well as in 2013 for two of those journals. Sample size was not significantly correlated with whether or not articles in one journal reported results for which p < .10. The use of less conservative levels of alpha or the reporting of statistical trends should not be used against the credibility or scientific soundness of either scholars or their research.
