Hey,

I am new to bayesian statistics, stan, and brms - so I have a long way to go, but all this seems very useful for my data, so I am trying to see how far I get.

I am trying to analyze the effect of parents on their offspring, taking into account (co-)dependencies between parents, maternal effects and developmental processes; using phenotypic data of a lab-pedigree. Individual models (`lm`

and `gam`

) I ran before using `brm`

indicate significant correlations between some variables (e.g. `Clutchsize_Lengthmom`

or `Pigmentation_Length`

, however, with low R2). Now my goal is to conduct a path analysis with `brm`

; seems more meaningful to have one model that reflects the biology at works than 5 separate regression analyses.

However, in my `brm`

fit (see below) the credibility interval for these estimates is extremely wide and within zero range and I am not sure what to make of this - I trust my visual inspections of the data, as well as the results I got from the individual models.

Why do my individual regression models detect these effects and the `brm`

fit did not? Is this simply too little data for such a complex model?

And, more philosophically, given that the estimates have the right sign and a meaningful magnitude, just how bad are wide intervals and zero-overlap?

Thanks!

**EDIT: I changed the title of this post after realizing the real problem over the course of the discussion.**

Model specs (all data was scaled and divided by 2):

```
path_mod_mates_clutch_juv = brm(
bf(Lengthmom ~ Lengthdad) +
bf(Pigmentationdad ~ Lengthdad) +
bf(Pigmentationmom ~ Lengthmom + Pigmentationdad) +
bf(Clutchsize ~ Lengthdad + Lengthmom + Pigmentationdad + Pigmentationmom) +
bf(Length ~ Clutchsize + Lengthmom + Lengthdad) +
bf(Pigmentation ~ Length + Clutchsize + Pigmentationmom + Pigmentationdad)
,data = mod_data,family=gaussian,
chains=4, cores=8,seed=17, iter=10000, warmup = 500, thin = 10, inits=0, silent=F,
prior=prior(normal(0, 1), b), save_all_pars=T,control = list(adapt_delta = 0.99,max_treedepth=20))
```

Summary:

```
Family: MV(gaussian, gaussian, gaussian, gaussian, gaussian, gaussian)
Links: mu = identity; sigma = identity
mu = identity; sigma = identity
mu = identity; sigma = identity
mu = identity; sigma = identity
mu = identity; sigma = identity
mu = identity; sigma = identity
Formula: Lengthmom ~ Lengthdad
Pigmentationdad ~ Lengthdad
Pigmentationmom ~ Lengthmom + Pigmentationdad
Clutchsize ~ Lengthdad + Lengthmom + Pigmentationdad + Pigmentationmom
Length ~ Clutchsize + Lengthmom + Lengthdad
Pigmentation ~ Length + Clutchsize + Pigmentationmom + Pigmentationdad
Data: mod_data (Number of observations: 62)
Samples: 4 chains, each with iter = 10000; warmup = 500; thin = 10;
total post-warmup samples = 3800
Population-Level Effects:
Estimate Est.Error l-95% CI u-95% CI Eff.Sample Rhat
Lengthmom_Intercept 0.06 0.05 -0.04 0.16 3738 1.00
Pigmentationdad_Intercept 0.18 0.06 0.06 0.30 3940 1.00
Pigmentationmom_Intercept 0.11 0.09 -0.06 0.29 3918 1.00
Clutchsize_Intercept 0.13 0.13 -0.13 0.40 3618 1.00
Length_Intercept 0.12 0.11 -0.11 0.34 3799 1.00
Pigmentation_Intercept 0.23 0.14 -0.05 0.52 3815 1.00
Lengthmom_Lengthdad 0.52 0.09 0.33 0.70 3834 1.00
Pigmentationdad_Lengthdad 0.22 0.11 -0.01 0.43 3818 1.00
Pigmentationmom_Lengthmom 0.19 0.23 -0.27 0.63 3442 1.00
Pigmentationmom_Pigmentationdad 0.27 0.39 -0.54 1.02 3509 1.00
Clutchsize_Lengthdad 0.13 0.31 -0.46 0.77 3864 1.00
Clutchsize_Lengthmom 0.34 0.54 -0.75 1.39 3841 1.00
Clutchsize_Pigmentationdad 0.15 0.50 -0.84 1.15 4052 1.00
Clutchsize_Pigmentationmom -0.05 0.55 -1.10 1.10 3803 1.00
Length_Clutchsize -0.34 0.52 -1.30 0.74 3659 1.00
Length_Lengthmom 0.03 0.56 -1.12 1.15 3707 1.00
Length_Lengthdad -0.20 0.31 -0.82 0.40 3751 1.00
Pigmentation_Length 0.39 0.43 -0.49 1.27 3813 1.00
Pigmentation_Clutchsize -0.27 0.40 -1.06 0.53 3814 1.00
Pigmentation_Pigmentationmom -0.30 0.50 -1.30 0.69 3739 1.00
Pigmentation_Pigmentationdad -0.48 0.45 -1.34 0.44 3769 1.00
Family Specific Parameters:
Estimate Est.Error l-95% CI u-95% CI Eff.Sample Rhat
sigma_Lengthmom 0.39 0.04 0.32 0.47 3512 1.00
sigma_Pigmentationdad 0.46 0.04 0.38 0.56 3476 1.00
sigma_Pigmentationmom 0.41 0.07 0.31 0.58 3899 1.00
sigma_Clutchsize 0.52 0.11 0.38 0.81 3993 1.00
sigma_Length 0.54 0.10 0.41 0.80 3622 1.00
sigma_Pigmentation 0.46 0.12 0.31 0.76 3927 1.00
Residual Correlations:
Estimate Est.Error l-95% CI u-95% CI Eff.Sample Rhat
rescor(Lengthmom,Pigmentationdad) 0.03 0.12 -0.21 0.27 3684 1.00
rescor(Lengthmom,Pigmentationmom) 0.17 0.21 -0.27 0.56 3607 1.00
rescor(Pigmentationdad,Pigmentationmom) 0.09 0.36 -0.59 0.73 3516 1.00
rescor(Lengthmom,Clutchsize) 0.08 0.34 -0.58 0.69 3679 1.00
rescor(Pigmentationdad,Clutchsize) 0.01 0.36 -0.65 0.68 3817 1.00
rescor(Pigmentationmom,Clutchsize) -0.03 0.34 -0.67 0.62 3757 1.00
rescor(Lengthmom,Length) 0.03 0.34 -0.61 0.65 3821 1.00
rescor(Pigmentationdad,Length) -0.19 0.13 -0.43 0.08 3879 1.00
rescor(Pigmentationmom,Length) -0.21 0.16 -0.50 0.15 3748 1.00
rescor(Clutchsize,Length) -0.05 0.32 -0.64 0.56 3784 1.00
rescor(Lengthmom,Pigmentation) 0.09 0.20 -0.33 0.45 3959 1.00
rescor(Pigmentationdad,Pigmentation) 0.09 0.34 -0.56 0.68 3760 1.00
rescor(Pigmentationmom,Pigmentation) 0.04 0.34 -0.62 0.67 3841 1.00
rescor(Clutchsize,Pigmentation) 0.16 0.29 -0.45 0.66 3830 1.00
rescor(Length,Pigmentation) -0.07 0.33 -0.65 0.59 3717 1.00
Samples were drawn using sampling(NUTS). For each parameter, Eff.Sample
is a crude measure of effective sample size, and Rhat is the potential
scale reduction factor on split chains (at convergence, Rhat = 1).
```

- Operating System: Ubuntu 18.04
- brms Version: 2.7.0