Choose all of the following scenarios where the researcher would need to formally adjust for the "problem of multiple comparisons":

For a study examining the effect of smoking on health, researchers conduct analyses to compare the difference among seven subgroups of smokers with different race/ethnicity (i.e. American Indian, Asian, Black, Native Hawaiian, White, Hispanic, and Multiple) and only report the significant comparisons using uncorrected p values.

The researchers have a sample size of N=35 and obtain non-statistically-significant results. They then decide to keep recruiting more participants and re-run the analyses until they get significant results.

The researchers run a study with multiple outcome variables (i.e, over 50 genetic variables) and only reports the outcomes that have statistically significant results, using uncorrected p values.

The researchers try multiple different methods to define a binary variable, and only report the final definition used as the one that produces the lowest p-value in their model of interest.

A study examining the effect of smoking on health pre-planned to only compare two subgroups of smokers with different race/ethnicity (i.e. Hispanic vs. White).

The researchers conduct multiple sub-group comparisons and transparently present the results from all comparisons with uncorrected p-values. The researchers then reminder the reader that they may informally account for multiple comparisons themselves.