[Kjell Weyde]

Sorry, I should have shown the complete commands before. I am investigating whether there is an association between night-time road traffic noise levels outside the home and inattention, as reported by parents (ADD part of the ADHD symptom list). The model behind the figures in #14, is:

glm frac_inattention road_traffic_noise i.gender age i.railnoise i.urbanity i.income_cat i.education i.ethnicity depression anxiety if gest_age>=259 & gest_age<300 & birthweight>=2500, link(logit) vce(robust) f(binomial)

The dependent variable was made according to your previous suggestions:

(inattention-9)/27

[Nick Cox]

I am reposting this to make it a little more readable:

glm frac_inattention road_traffic_noise i.gender age i.railnoise   ///
i.urbanity i.income_cat i.education i.ethnicity depression anxiety ///
if gest_age>=259 & gest_age<300 & birthweight>=2500,               ///
link(logit) vce(robust) f(binomial)

[Kjell Weyde]

What are we looking for in the plots

twoway function _b[_cons] + _b[stoy10_8] * x , ra(stoy10_8)
twoway function invlogit(_b[_cons] + _b[stoy10_8] * x) , ra(stoy10_8)

What is considered "aceptable" concerning model fit?
And do you think the rvfplot2 looked OK enough?

[Nick Cox]

There is precisely no evidence in those plots about model fit. They just show logit p as a function of the intercept and predictor and p as a function of the intercept and predictor. Your earlier question showed that you didn't understand what the coefficient was, so the graphs may help to explain that.

As said, residual versus fitted is just one kind of check. It doesn't mean that your model is good any more than the fact that there is petrol in the tank means that your car will drive without problems. The model needs to make sense too and pass other kinds of checks.

[Santosh Pathak]

Nick Cox Please kindly suggest how can we extend the fractional regression estimation in unbalanced panel data (T = 24 years, n = 64) with multiple independent variables.

With regards,
Santosh Pathak

[Nick Cox]

@Santoosh Pathak You have asked that on How to deal with panel dataset with variables in fractional form? - Statalist The question doesn't fit here. Further, I really don't know more about this than the people you're asking there.