mr_ruckers missing correction for underdispersion leads to p-values = 0

[arendma]

This relates to the discussion in #381. There @mightyphil2000 suggests to correct for underdispersion (underlined by the given literature sources).

However, checking the implementations of ivw:

TwoSampleMR/R/rucker.R

Line 141 in 6268c2c

se_ivw_fe <- stats::coefficients(mod_ivw)[1,2] / max(mod_ivw$sigma, 1)

TwoSampleMR/R/rucker.R

Line 153 in 6268c2c

se_ivw_re <- stats::coefficients(mod_ivw)[1,2]

and MR egger:

TwoSampleMR/R/rucker.R

Line 179 in 6268c2c

se1_egger_fe <- stats::coefficients(mod_egger)[2,2] / max(mod_egger$sigma, 1)

TwoSampleMR/R/rucker.R

Line 191 in 6268c2c

se1_egger_re <- stats::coefficients(mod_egger)[2,2]

in the mr_rucker() function I could not find any underdispersion correction.

For the fixed effect IVW results this can lead to much smaller standard errors and pvalues then reported by the ivw_fe() implementation for the same IVs/ summary statistics (in some extreme cases 0 p-values).

What's the reason for the different implementation in mr_rucker() & mr()?