I submitted a paper to a conference. In the paper, I created a model namely lm1:
lm1 <- brm(data=ys1, ys ~ vec+delta+deltachi+deltahmix+deltasmix, seed=123456)
The lm1 is summarized in the first revision of the paper as (I do not have full info of the model as code output):
Estimate Est. Error CI95 low CI95 high
Intercept 267.98 311.28 -352.65 867.58
vec -160.01 22.73 -204.49 -115.98
delta -0.13 18.18 -35.95 36.63
deltachi 346.74 455.77 -522.00 1230.91
deltahmix -14.30 5.14 -24.58 -4.34
deltasmix 130.83 20.43 91.78 170.99
(don’t remember brms version, could be 2.18.0 or 2.19.0)
In the paper’s second revision, I used the same code with the same seed but the model coefficients are different:
Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
Intercept 275.30 315.56 -337.46 893.49 1.00 5645 3379
vec -160.73 22.79 -205.42 -116.02 1.00 5163 3352
delta -0.67 18.39 -37.15 35.09 1.00 3649 2697
deltachi 360.13 462.72 -546.49 1265.08 1.00 5150 3152
deltahmix -14.41 5.22 -24.63 -4.33 1.00 3756 2877
deltasmix 130.61 21.14 89.13 171.87 1.00 5357 2911
(brms version: 2.19.0)
The first model was created on Fedora, the second was created on Ubuntu, on the same physical home desktop. I reinstalled the operating system. My data is unchanged.
How could I reproduce the the model?
Thank you for you advice.
Please also provide the following information in addition to your question:
- Operating System: Fedora, Ubuntu
- brms Version: 2.18.0, 2.19.0
Don’t forget to add relevant tags to your topic (top right of this form) especially for application area. Delete this text before posting your question :-) Thx!
Looking forward to your topic!