This study shows inference of association to decreased risk of being bullied and experience of violence/injury for healthy lifestyle behaviour, the effect remaining even after controlling for risk behaviour, exhibiting a buffering effect.
Fig. 3 An IGV screenshot showing inference of somatic meiotic-like recombination at scaffold 278.
We first describe the dataset and then show inference on HWE using single and multiple imputation.
As shown, inference is the same irrespective of the allele coding method when a fixed general mean is in the model.
Table 3 and Figure 5 show inference is insensitive to this, indeed for plausible delta the estimate moves back towards the complete case estimate, consistent with what would be expected if the additional MNAR dependence does not materially change the lack of dependence of the chance of observing genotype 3 on severe liver disease.
Table 3 and Figure 5 show inference is sensitive to this, suggesting that if the mechanism is MNAR with increased chance of observing HIV status for those who are positive, the association in the model of interest is stronger and more significant than analysis under MAR would suggest.
Table 3 and Figure 5 show inference is relatively insensitive to this, which is consistent with what would be expected if the additional MNAR dependence does not materially change the dependence of the chance of observing alcohol consumption on severe liver disease.
Also, this simple model still shows that inferences of the static model hold in the dynamic extension.
A simple counterexample using an additional copollutant variable shows that inferences based on this method could be misleading.
An obvious example is Anna Wintour's refusal to carry a bag at fashion shows (the inference being that she has minions to do the heavy lifting); another is a fondness for ludicrously high heels (translation: I only take taxis, darling).
