Closed aiorazabala closed 8 years ago
Or it might be an issue arisen directly from qmethod()
exporting the TRUE/FALSE
values incorrectly.
Fixed. The issue did not affect any other calculation
The exact issues were two, and both were in the calculation of the overall bootstrap stats (they didn't affect any of the individual steps, neither any other results!):
FALSE
rather than as NA
(which would be the correct value), this decreased the proportion of TRUE
versus FALSE
in all cases.
The fix is:
https://github.com/aiorazabala/qmethod/commit/48de489820a0fd96766ac4cf78546c6baf852379#diff-6dc9760350d2ae99aa5aeaa001318fa2L18
replaced with
https://github.com/aiorazabala/qmethod/commit/48de489820a0fd96766ac4cf78546c6baf852379#diff-6dc9760350d2ae99aa5aeaa001318fa2R18
As per the title... some recent change in the code I made, broke the correct construction of tables in
boots$full.bts.res$factor_***$flagged
. This does not affect the results, as they remain the same. I suspect that the bit that exportsTRUE/FALSE
values into this table after the analysis, is mistaken.