Я использую бета-регрессии с использованием пакетов betareg и glmmTMB в R. Значения коэффициента и AIC получаются одинаковыми, но значения p отличаются.Почему это так?
Betareg: beta1 <-betareg(eagtrans ~ Interspace + cover + elevation + Basal,
data = rall)
> summary(beta1)
Call:
betareg(formula = eagtrans ~ Interspace + cover + elevation +
Basal, data = rall)
Standardized weighted residuals 2:
Min 1Q Median 3Q Max
-2.8165 -0.8454 0.2053 0.5900 3.9175
Coefficients (mean model with logit link):
Estimate Std. Error z value Pr(>|z|)
(Intercept) -2.418009 0.341760 -7.075 1.49e-12 ***
Interspace 0.012614 0.004326 2.916 0.00354 **
cover 0.039379 0.012271 3.209 0.00133 **
elevation -0.433384 0.174386 -2.485 0.01295 *
Basal -0.033827 0.018478 -1.831 0.06715 .
Phi coefficients (precision model with identity link):
Estimate Std. Error z value Pr(>|z|)
(phi) 5.9271 0.8248 7.186 6.65e-13 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Type of estimator: ML (maximum likelihood)
Log-likelihood: 87.12 on 6 Df
Pseudo R-squared: 0.1687
Number of iterations: 12 (BFGS) + 2 (Fisher scoring)
glmmTMB: beta2 <-glmmTMB(eagtrans ~ Interspace + cover + elevation + Basal,
data = rall, family=list(family="beta",link="logit"))
> summary(beta2)
Family: beta ( logit )
Formula: eagtrans ~ Interspace + cover + elevation +
Basal
Data: rall
AIC BIC logLik deviance df.resid
-162.2 -146.3 87.1 -174.2 100
Overdispersion parameter for beta family (): 5.93
Conditional model:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -2.41801 0.34402 -7.029 2.08e-12 ***
Interspace 0.01261 0.00444 2.841 0.00450 **
cover 0.03938 0.01222 3.224 0.00127 **
elevation -0.43338 0.19421 -2.232 0.02565 *
Basal -0.03383 0.01944 -1.740 0.08178 .
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Эти пакеты рассчитывают значение p по-разному?