To correct this mistake and confirm the statistical significance of the originally reported changes in blood pressure, new telemetry data have been acquired in additional animals to increase the sample size, and the data have been reanalyzed without REML transformation. The error was noted after publication of the article when an unrelated analysis led to conspicuously low p values. Whereas REML is disabled in glmer, it is enabled by default in lmer, which led to incorrect p-values. However, in such a setting glmer automatically refers the data to the lmer function. This error occurred because a glmer-based script previously used by the authors to analyze data with non-normally distributed errors was run under the assumption it was equally applicable to a normally distributed error model for use in the analysis in this article. The approach called “restricted maximum likelihood” (REML) was inadvertently and erroneously used during the analysis, causing spuriously low p-values to be generated by the likelihood ratio test model comparisons.
The original version of this article contained an error in the statistical analysis of the telemetry data comparing male WT and male R180Q/+ mice, which were analyzed by mixed effect modeling using the lme4 R package and the lme4::glmer function.
0 kommentar(er)
