New answers tagged

1

tl;dr your analysis removes missing data. You lose a bit of power, but the analysis remains unbiased when model assumptions are met. If an assumption is violated, there's not much a missing data method can do for you. LMER in R will remove missing observations, i.e. a complete case analysis is performed. As with complete case analysis in a likelihood based ...


-1

I wrote a paper arxiv.org/abs/1907.09090 that describes how the pseudo-marginal approach can impute missing data. 400 covariates sounds tough, though, to be completely honest. Depends on what kind of distributions you want to put on the columns, the number of rows, how you program everything. Intractable in your case? Probably, yes. In section 3.3, we ...


0

Are you familiar with censorship in survival analysis modeling? Censorship means that a patient was removed from the study without explicitly dying (at least in the context of survival analysis.) There are appropriate architectures to handle censorship; however, it's more a question of model architecture than the underlying MCMC sampling algorithm. For ...


2

Your first question is not really on-topic on this site, but what you decide to think about it might have consequences for the modeling. You should also ask yourself why gender is missing at such a large rate, might it have to do with people actively refusing to answer? So, depending on context and goals, missing might be informative, so just dropping might ...


0

No, there should not be a multiplicity correction. Recall that if we perform two independent hypothesis tests and the null hypothesis is true for both of them, the probability that we incorrectly reject the null hypothesis at 0.05 on at least one of the tests is 0.0975, much higher than the intended confidence level. Multiplicity corrections adjust for that ...


1

This situation would readily be handled by survival analysis. Age at onset or at last follow-up would be the outcome variable, further coded as to whether the time represented an event (onset at that time) or not (censored time, no onset as of last follow-up). The other variable(s) would be included as predictor(s) in the model. Choice between a continuous-...


0

You've correctly found that the data being NMAR is not a sufficient condition for there to be bias in the estimated parameters. To make the example more obvious, consider a simple linear regression with only 1 covariate and no correlation between $x$ and $y$. set.seed(1) n <- 10000 x <- rnorm(n) y <- rnorm(n) mod1 <- lm(y ~ 0 + x) summary(mod1) # ...


0

It isn't possible to answer this question generally. It will depend on the type and objectives of the analysis and on the particulars of the data. Sometimes a partially observed variable is very useful, in which case the decision needs to be made based on domain knowledge. In cases when the variable is mostly missing and doesn't significantly change the ...


0

I think nnet supports the all the standard na.action arguments. So, you can set na.action = na.exclude in the nnet call. Then predict will insert NAs into the predicted values at the appropriate places and your code should work. Alternatively, you could try something like i = which(is.na(df)) return(data.frame(pred=psm_pred,truth=psm_truth[-i])) (assuming ...


Top 50 recent answers are included