# Wolfgang

less info
reputation
928
bio website wvbauer.com location Netherlands age member for 3 years, 1 month seen 4 hours ago profile views 383

# 208 Actions

 1d comment Sum of variances from regression coefficients, larger then total variance. Why? If I understand you correctly, you want to determine the relative importance of a particular predictor. This is not a simple problem. There are some suggestions on how to do that in the literature. As an example: ncbi.nlm.nih.gov/pubmed/12924811 You could also consider looking into partial and semi-partial correlations: en.wikipedia.org/wiki/Partial_correlation 1d comment Sum of variances from regression coefficients, larger then total variance. Why? Okay, Ymod is $Var(\hat{y})$, so the variance of the predicted values. That is indeed equal to $\sum b_j^2 Var(x_j)$ if the x's are all perfectly uncorrelated. But not when they are correlated. If $n$ is large (relative to the number of x's), then the difference will be negligible, but otherwise $\sum b_j^2 Var(x_j)$ can be easily 50% larger than $Var(\hat{y})$. 1d comment Sum of variances from regression coefficients, larger then total variance. Why? I don't understand what you mean by 'Ymod' and 'the results that come from the linear fit, when i use it on the original samples.' Can you actually write down how 'Ymod' is computed? 1d comment Sum of variances from regression coefficients, larger then total variance. Why? If $n$ is large, then this is probably not the real source of the issue. Let me see if I understand you correctly. You are computing $\sum b_j^2 Var(x_j)$ and find that it does not equal $Var(y)$, correct? Even if all of the X's were perfectly uncorrelated, then this is not surprising. You need to compute $\sum b_j^2 Var(x_j) + Var(e)$ (i.e., you also need to add the variance of the residuals) if you want to get $Var(y)$. 1d answered Sum of variances from regression coefficients, larger then total variance. Why? Dec1 revised Determine individual consistency from mixed model? Expanded on my answer based on the OPs comment. Dec1 comment Determine individual consistency from mixed model? The ICC is a correlation, so you can interpret it that way. Whether a correlation of .35 is small, medium, or large for your particular application is something you will have to think about. As for testing its significance - you can just test whether the intercept variance is larger than zero. That is the same as testing whether the ICC is larger than zero. You can also allow the correlation to differ for the different task pairs. I'll update my answer to show you how that can be done. Dec1 answered Determine individual consistency from mixed model? Nov25 comment Simulating longitudinal lognormal data in R That would be useful. Also useful would be an idea what kind of model you assume would provide a good fit to your data. That is essentially the basis for simulating the data. I just gave one simple example, but ultimately that needs to be tweaked based on your application. Nov25 comment Statistical comparison of 2 independent Cohen's ds If you do a regular meta-analysis on those 2 $d$ values, then the $Q$ test for heterogeneity will just be the square of the $z$ statistic above. So, those two approaches are exactly equivalent. Nov25 answered Simulating longitudinal lognormal data in R Nov24 awarded Nice Answer Nov23 answered Statistical comparison of 2 independent Cohen's ds Nov9 awarded Yearling Sep30 comment Main effects and interaction in multivariate meta-analysis (network meta-analysis) in R I edited my answer to address your question. Sep30 revised Main effects and interaction in multivariate meta-analysis (network meta-analysis) in R Provided answer to comment by OP asking for additional methods. Sep29 revised Main effects and interaction in multivariate meta-analysis (network meta-analysis) in R Fixed grammar. Sep29 answered Main effects and interaction in multivariate meta-analysis (network meta-analysis) in R Sep29 comment Main effects and interaction in multivariate meta-analysis (network meta-analysis) in R 1) Your example data make it seem as if each effect size estimate comes from a different study ($studylab$). Is that correct? 2) You mention a continuous variable, but this is not part of the example data. Sep20 comment How to estimate a standardized mean difference from two samples' quartiles? Regarding (1.): True, but using the medians will make the loss of precision smallest. Combining different options would be another option.