Retraction Watch story on the five Stewart articles:
It includes links to the anonymous emails with the irregularities:
https://retractionwatch.com/wp-content/uploads/2019/07/Smith-email-1.pdf
https://retractionwatch.com/wp-content/uploads/2019/07/Smith-email-2.pdf
The emails are wild. The first one starts:
In Stewart et al. (2018), tables 2-4 include regression results. The standard errors are identical to the third decimal place across models. The web links are to pictures of the tables. Stable standard errors are in yellow. Some stability in standard errors is normal, especially with a sample of this size, but this level of stability is unusual given the observed changes both in the regression coefficients and in the amount of explained variance. Combined there are 548 regression coefficients and standard errors in these three tables, but just one ends with a zero in the third decimal place. This is unusual because the distribution of third-decimal-place numbers (with rounding) should be close
to uniform.
The standard errors are also identical across models in Stewart et al. (2019). The following web links are to the tables. Stable standard errors are in yellow. None of the 348 regression coefficients and standard errors in these two tables end with a zero in the third decimal place, even though the distribution of third-decimal-place numbers (with rounding) should be nearly uniform.