It's difficult to impute the behaviour of a small number of individuals to a whole company, and, to repeat, there are two separate companies here.
Of the removed SNPs, 40.6% had a MAF of 0, which could have been easily imputed with a 100% accuracy, 56.1% had a MAF between 0 and 0.1 and their imputation accuracy could have been affected by their low MAF only, and the remaining 3.3% had a MAF above 0.1, which could have been difficult to impute for other reasons than their low MAF.
The simulation design described in this section relates to a setting where the ideal imputation method is unclear: the presence of x and x in the analysis model means it is difficult to find a compatible model for imputing x| y[ 35].
Missing outcome data will not be imputed, as it is difficult to make assumptions about the missingness of supermarket till receipt data, or the direction of the imputed values.
It is difficult to identify in advance the number of multiple imputed data sets we will need to construct but it is likely to be between 5 and 10.
Also, because of the varying estimation samples it is difficult to present nested models with equal number of observations for each estimated specification when exploiting multiply imputed data.
We find that it is more difficult to impute non-European samples, which is due to our imputation towards the GoNL reference panel.
We focused mainly on the rare variants, as these are more difficult to impute and most can be gained in terms of imputation quality when using a better reference set.
The DoM estimator allows for more straightforward modeling of additional unmatchable covariates and is easier to apply when matched pairs are difficult to select or impute.
Although many loci with a low MAF in the observed genotypes were removed, among the remaining SNPs those with a lower MAF were more difficult to impute correctly and the reliability of imputation varied more than for the SNPs with a higher MAF.
Surprisingly, typical imputation error rates (∼2%–6%) lead to a large increase in the required sample size (∼10%–60%), and in some African populations whose genotypes are particularly difficult to impute, the required sample-size increase is as high as ∼30%150%%.
