[R-sig-ME] Mixed Multinomial model built with npmlt function: coefficients significance and Goodness of fit

[Romaine TCHEDJI]

Hello all,
I fitted (using npmlt function) a mixte multinomial regression model which
ran without errors.
I expected that the outputs of the model include p-values that I didn't
find.
Please, find below the outputs of the model:

Call:  npmlt(formula = three levels categorical variable ~ variable1 +
variable2 ,
formula.npo = ~1 + variable1 , random = ~1, id = Patient_id, k = 2)

Coefficients:

```
|                             |Estimate |Std. Error|
---------------------------------------------------
| (Intercept) 1                   | -1.029e+01|2.995e+00  |
|(Intercept) 2                    |1.024e+01   |1.544e+00|
|(Intercept) 3                    |2.351e+01   |1.352e+00|
|variable1           level1 1      |-7.995e+00  | 2.444e+00|
|variable1           level1 2      |-2.711e+00  | 7.896e-01|
|variable1           level1 3    |-1.375e+01|   7.093e-01|
|variable1           level2 1    |-1.970e+01   |7.507e-07|
|variable1           level2 2    |-2.172e+01   |1.171e+00|
|variable1           level2 3    |-1.688e+01  | 9.573e-01|
|variable1           level3 1    |-1.976e+01   |5.782e-15|
|variable1           level3 2    |-2.835e+01   |3.827e-15|
|variable1           level3 3    |-4.295e+01   |8.260e-16|
|variable2 1                     | 1.527e+00   |7.733e-01|
|variable2 2                      |2.288e-01   |1.865e-01|
|variable2 3                      |2.760e-01   |2.915e-01|
```

Please, could someone here help me address the following concerns:
1)
 How do I know whether a coefficient (the outputs above) is significant or
not?  I mean, Retrieve p-values or calculate confidence intervals based on
the above output.
2)
How do I assess the Goodness of fit of such a model?
I unsuccessfully tried some common tools which seem not to support this
type of model.

Thanks in advance for your answers.
Best

	[[alternative HTML version deleted]]