std_selected(): Should bootstrap SEs be reported if bootstrapping is requested?

[sfcheung]

Note

• Although the t statistic is valid as a test of the null hypothesis in terms of Type I error rate, the OLS SEs are not recommended as a measure of sampling variability and forming the CI when a coefficient involves variables that are standardized and the population value is not equal to zero. Therefore, it may sound strange to report the OLS SEs, ts, and p-values alongside bootstrapping CIs.
• However, some users may want to use bootstrapping only to form CIs. They may want OLS SEs, ts, and p-values, especially because the p-values are the same for a coefficient before and after standardization (except for some of them in moderated regression). Giving two sets of p-values may be confusing to readers. Having different p-values for unstandardized and standardized coefficients is even more confusing.
• Maybe just let users to decide themselves? This only need to change print.std_selected().

[sfcheung]

Moved.