atahk / pscl

Political Science Computational Laboratory
65 stars 14 forks source link

Error running zeroinfl() with dist = "negbin": NA for log(theta) std. error #12

Open shailee93 opened 4 years ago

shailee93 commented 4 years ago

Hello!

I am trying to run a zero-inflated negative binomial for my data on lifetime inclusive fitness of birds. This is the distribution of fitness (dependent variable):

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
 0.0000  0.0000  0.0000  0.3321  0.2072  3.8041 

The independent variables are "prop.breeding": number of times a bird attempts to breed in its lifetime (it can nest multiple times a year), and "birthstatus": natal (N) or immigrant (I)

This is the model summary:

Call:
pscl::zeroinfl(formula = as.integer(inclusive.fitness_positive) ~ prop.breeding * birthstatus, 
    data = inclusive.fitness.all, dist = "negbin", link = "logit", maxit = 5e+06)

Pearson residuals:
    Min      1Q  Median      3Q     Max 
-0.9914 -0.2128 -0.2128 -0.1003  4.4865 

Count model coefficients (negbin with log link):
                           Estimate Std. Error z value Pr(>|z|)    
(Intercept)                  1.3126     0.6628   1.980   0.0477 *  
prop.breeding               -0.3385     0.2552  -1.327   0.1846    
birthstatusN                -4.4074     0.8316  -5.300 1.16e-07 ***
prop.breeding:birthstatusN   2.4936     0.4214   5.918 3.27e-09 ***
Log(theta)                  10.4413         NA      NA       NA    

Zero-inflation model coefficients (binomial with logit link):
                           Estimate Std. Error z value Pr(>|z|)    
(Intercept)                  4.3586     1.2901   3.378 0.000729 ***
prop.breeding               -1.6734     0.6668  -2.509 0.012092 *  
birthstatusN               -18.5287    13.4420  -1.378 0.168074    
prop.breeding:birthstatusN   9.4203     6.8572   1.374 0.169510    
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 

Theta = 34244.0508 
Number of iterations in BFGS optimization: 36 
Log-likelihood: -56.06 on 9 Df

I can't figure out why the standard error of log(theta) is NA.

Thanks for you help!