tag:blogger.com,1999:blog-3240271627873788873.post1539987704375634879..comments2024-03-26T06:46:11.752-04:00Comments on Doing Bayesian Data Analysis: Trade-off of between-group and within-group variance (and implosive shrinkage)John K. Kruschkehttp://www.blogger.com/profile/17323153789716653784noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-3240271627873788873.post-91548259146398848082017-04-15T22:25:59.426-04:002017-04-15T22:25:59.426-04:00Don: Shrinkage comes from the hierarchical model, ...Don: Shrinkage comes from the hierarchical model, not from Bayesian estimation. A maximum likelihood estimate of the hierarchical model would also show shrinkage. A major benefit of the Bayesian approach is the explicit posterior distribution on all the parameters -- no need for auxiliary assumptions to generate sampling distributions for p values and confidence intervals. (I'd have to delve into the details of lme4 to know if it's the very same model structure -- might be. But in JAGS, it's easy to generalize to non-normal distributions and heterogeneous variances.)John K. Kruschkehttps://www.blogger.com/profile/17323153789716653784noreply@blogger.comtag:blogger.com,1999:blog-3240271627873788873.post-558547001902092882017-04-14T16:44:58.107-04:002017-04-14T16:44:58.107-04:00Hi John:
What is the difference between your model...Hi John:<br />What is the difference between your model and using a so-called random effect ANOVA using lme4 or nlme, both of which are frequentist? The shrinkage here is also obtained by bayes theorem for the random effects. If you summarize the posterior with the mode, I would imagine the estimates are equivalent to those from a frequentist model, which also uses the mode. Anonymoushttps://www.blogger.com/profile/13045015065678590524noreply@blogger.comtag:blogger.com,1999:blog-3240271627873788873.post-31349414356558759562017-04-10T21:10:03.030-04:002017-04-10T21:10:03.030-04:00In the high-level script, one of the lines defines...In the high-level script, one of the lines defines fileNameRoot. Be sure you've executed that line.John K. Kruschkehttps://www.blogger.com/profile/17323153789716653784noreply@blogger.comtag:blogger.com,1999:blog-3240271627873788873.post-75205818164160623442017-04-10T21:07:21.885-04:002017-04-10T21:07:21.885-04:00Hello, just starting this chapter and encountered ...Hello, just starting this chapter and encountered this error below. Is it something obvious I have misspelt? Thanks, Graeme<br /><br />myDataFrame=read.csv("FruitflyDataReduced.csv")<br />yName="Longevity"<br />xName="CompanionNumber"<br />source("Jags-Ymet-Xnom1fac-MnormalHom.R") <br />mcmcCoda=genMCMC(datFrm=myDataFrame,yName=yName, xName=xName,<br /> numSavedSteps=1100, thinSteps=10, saveName=fileNameRoot)<br /><br /><br />Error in genMCMC(datFrm = myDataFrame, yName = yName, xName = xName, numSavedSteps = 1100, : <br /> object 'fileNameRoot' not found<br /><br /><br /><br />OUTPUT<br />Reading parameter file inits1.txt<br />. Initializing model<br />. Adapting 500<br />-------------------------------------------------| 500<br />++++++++++++++++++++++++++++++++++++++++++++++++++ 100%<br />Adaptation successful<br />. Updating 1000<br />-------------------------------------------------| 1000<br />************************************************** 100%<br />. . . . . . Updating 3670<br />-------------------------------------------------| 3650<br />************************************************** 100%<br />* 100%<br />. . . . Updating 0<br />. Deleting model<br />. <br />All chains have finished<br />Simulation complete. Reading coda files...<br />Coda files loaded successfully<br />Finished running the simulationAnonymousnoreply@blogger.com