Although imputation techniques enable researchers to fill in gaps in data sets, GBS approaches typically suffer from large amounts of missing data, making it challenging to accurately call heterozygotes.
However, if that happens on a voice call, though, those bits of missing data make the human voice sound broken, like it's coming from a poor cellular phone connection.
First, the selected data might exclude samples because of missing data, making the data here incomplete.
Imputation approaches to manage missing data made no difference to the results.
These missing data make attributing increased radon levels to a particular source difficult.
Multiple imputation for missing data made no differences to our findings (see online supplementary appendix table S6).
Sensitivity analyses will be conducted to investigate the potential impact of missing data, making different assumptions such as 'best' and 'worst' case scenarios as well as imputation models.
As there were no material differences between the results and because imputing missing data made maximum use of the actual responses provided, we report here only the results from the analyses incorporating imputed data.
However, since the sample was not randomly selected we cannot be sure that dropping cases with missing data made the sample less representative of the low-income female primary care population.
