Skip to content
Prev 15885 / 20628 Next

effective sample size in MCMCglmm

Hello again,


Thank you so much, I will change the code and specify pl=FALSE!

I ran the model based on Matthew's suggestion and I got the output below. What effective sample should I aim for in regards to the  G structure (studyid) - should I try again tweaking the specifications based on higher burnins until I reach an effective sample of 100 for studyid?


On a separate note, I managed to plot the model, but it looks illegible, the graphs are so small given the large number of groups. Should I plot just a random subsample then?


Thank you all so much,

D

Iterations = 3001:102901
Thinning interval  = 100
Sample size  = 1000

DIC: 2928.214

 G-structure:  ~studyid

           post.mean  l-95% CI u-95% CI eff.samp
studyid   0.06139 5.424e-16   0.4621    21.16

               ~class

      post.mean l-95% CI u-95% CI eff.samp
class    0.7278    0.132    1.232    154.9

               ~idv(l_lfvcspn)

           post.mean l-95% CI u-95% CI eff.samp
l_lfvcspn.     752.6    50.16     2352     1000

 R-structure:  ~units

      post.mean l-95% CI u-95% CI eff.samp
units     2.533    1.907    3.204    186.5

 Location effects: y ~ f_newage_c + x2n + x8n + x9n + x5n + l_lfvcspo + x3n + x4n + x6n + x7n + offset

            post.mean  l-95% CI  u-95% CI eff.samp  pMCMC
(Intercept) -7.033381 -7.535696 -6.524339    487.7 <0.001 ***
f_newage_c   0.009066 -0.025116  0.041159    909.3  0.596
x2nM        -0.107655 -0.403837  0.165664   1000.0  0.440
x8n1         0.413878  0.119578  0.745171    891.1  0.002 **
x9n1        -0.287382 -0.600003  0.016531    892.7  0.070 .
x5n         -0.001037 -0.006182  0.004417    864.8  0.670
l_lfvcspo    0.397952 -0.941309  1.455719   1000.0  0.426
x3n          0.079109 -0.001274  0.150664    914.1  0.042 *
x4n          0.058303 -0.067089  0.174404   1000.0  0.344
x6n         -0.013916 -0.067176  0.049719   1000.0  0.610
x7n         -0.014880 -0.081243  0.057582    710.6  0.654
offset       0.999452  0.977609  1.018083   1000.0 <0.001 ***


<http://aka.ms/weboutlook>