Reputation
Top tag
Next privilege 250 Rep.
View close votes
Badges
8
Impact
~2k people reached

  • 0 posts edited
  • 0 helpful flags
  • 6 votes cast
Jul
24
comment Computing Standard Errors in EM algorithm
I am not sure if this theory carries straight over to the complete data observed information matrix too. I imagine it does since the EM algorithm by maximising the complete data LL also maximises the ordinary LL. However I would want to see some theory on this and I am no expert.
Jul
24
comment Computing Standard Errors in EM algorithm
The Observed information matrix under certain "regularity conditions" can be a consistent estimator of the Fisher Information matrix. I tried to prove this in a post here
Jul
23
revised How to simulate informative censoring in a Cox PH model?
To clarify some comments I made
Jul
23
comment How to simulate informative censoring in a Cox PH model?
Sorry I don't know how to post the code here in a comment! I think I need to assume a model for the event times beyond the point of censoring - this would be a counterfactual event. I think if this counterfactual model is the same as the event times model then censoring will only affect the baseline survival function estimate (shifts it down?), but the regression parameters remain well estimated. Only if the counterfactual model is different does censoring make a difference - i.e. running a coxph on the "full" data if we could see it would produce a different result than on the observed data.
Jul
23
comment How to simulate informative censoring in a Cox PH model?
Thanks @DWin. I looked at the link you gave, and as far as I could understand, the code therein showed how the baseline hazard function estimate becomes worse as time increases (due I think to the ever decreasing numbers of subjects at risk). I amended this code keeping the same censoring function. I also simulated covariate effects. If I choose the sample size high enough I can get unbiased estimates from coxph even though the same degradation of the hazard function occurs for increasing time. Thus the bias you speak of I think does not relate to estimation of the regression parameters.
Jul
22
revised How to simulate informative censoring in a Cox PH model?
improving notation
Jul
21
asked How to simulate informative censoring in a Cox PH model?
Jun
24
asked covariate-adjusted analysis for time to event endpoint in a cross over design RCT
Jun
8
revised Sample size calculation for crossover design using dependent t-test - role of $\rho$?
Clarified the difference between the index j for both models (treatments versus periods)
Jun
6
asked Sample size calculation for crossover design using dependent t-test - role of $\rho$?
Jun
4
awarded  Nice Question
Jun
1
comment Residual Diagnostics and Homogeneity of variances in linear mixed model
Great questions - a possible answer to your number 2 can be found here comp.soft-sys.sas.narkive.com/7Qmrgufe/…
May
28
awarded  Popular Question
Feb
11
comment Picking cases to label for classification
The parameter estimates obtained could then be used to build prior distributions on the model paramters so that you could then estimate within a Bayesian framework the mixture model on the labelled and unlabelled data. In this way once the expert has done their classification, the classification of subjects to diseased/un-diseased would be model-based. The accuracy of this classification could of course be checked against the cases the expert has reviewed.
Feb
11
comment Picking cases to label for classification
You could use mixture models assuming your subjects are drawn from one of 2 subgroups in the population: diseased or not. If your response variable is say normally distributed (maybe some measure of kidney function over time) then you could fit a 2-component normal mixture distribution to the data where each group is modelled with a (for example) linear mixed effects model. You could then incorporate the expert by fitting 2 separate mixed models to only those cases reviewed (hence you know the 2 groups).
Jan
26
asked Relationship between Gumbel and Weibull distribution, accelerated failure time models, and Survreg using R
Jan
12
comment Is it valid to include a baseline measure as control variable when testing the effect of an independent variable on change scores?
measurement as a response, and conditioning on a fixed baseline value, and secondly that the point estimate variance from the ANCOVA model is always greater than or equal to that from the unconditional one. It turns out this variance difference will typically be small due to randomization ensuring baseline mean responses between the groups are small. The authors conclude the unconditional model is appropriate for modelling baseline as a random variable, but ANCOVA as appropriate when viewing it as fixed.
Jan
12
comment Is it valid to include a baseline measure as control variable when testing the effect of an independent variable on change scores?
A clear discussion of this paper can be found in (“Should baseline be a covariate or dependent variable in analyses of change from baseline in clinical trials?” by Liu, Mogg, Mallick and Mehrotra 2009). They refer to this model as an unconditional model (i.e. it does not condition on baseline response). In the Liu (2009) paper they discuss the main results of the Zeger (2000) paper. These are firstly that with no missing data the point estimates of $B_{1}$ from the unconditional model are the same as those from the conditional approach of ANCOVA using the post-baseline
Jan
12
comment Is it valid to include a baseline measure as control variable when testing the effect of an independent variable on change scores?
The model I am talking about does indeed imply $Y_{1}$ is a function of treatment, but only from the viewpoint that despite randomization there will always be slight differences between the treatment and control group with respect to their baseline means. Thus $\beta_{1}$ will capture this difference as well as the effect of treatment. The reference for this is ("Longitudinal Data Analysis of Continuous and Discrete Responses for Pre-Post Designs" by Zeger and Liang, 2000).
Jan
12
comment Is it valid to include a baseline measure as control variable when testing the effect of an independent variable on change scores?
In the equation $Y_2 = \beta_1T + \beta_2X + \beta_3Y_1 + (e + Y_1)$ if, as is standard practice, we assume all the covariates are not random variables, then $Y_1$ is not correlated with $e + Y_1$. Thus I think there is only a problem if you view $Y_1$ as random, in which case (again just my opinion) you should model $(Y_1,Y_2)$ jointly but without $Y_1$ as a covariate. In this respect without missing data I have been informed that this approach is equivalent to $Y_1$ being a fixed covariate (I will try and find some references for this).