Interpretation for Gologit2

[Valerie Aulia]

Hello everyone

In my recent study, I wanna estimate using Gologit as my dependent var is in categorical (Keluhan Kesehatan/Health Complaints (Y) 1 (lowest) and 4 (highest or complaining about health, illness, interference with work, to outpatient care)). I choose Gologit as my model doesnt fit the parallel trend assumption for the usual ordered logit. I have an independent var of Car Ownership (binary, 1 refer to owned car, and 0 otherwise). These are the details:

dependent var (Y) : HealthComplaints
Category Y = 1 if the respondent has not experienced any health complaints during the past month (No complaints)
Category Y = 2 if the respondent has experienced health complaints during the past month, but they have not interfered with work, school, or daily activities (Mild)
Category Y = 3 if the respondent has experienced health complaints during the past month that have interfered with work, school, or daily activities, but have not sought outpatient treatment (Moderate)
Category Y = 4 if the respondent has experienced health complaints that have interfered with daily activities during the past month (Severe) 

Independent var (X): CarOwnership
CarOwnership = 1 = Owning car
CarOwnership = 0 = Didnt own car

I have some control variables: married (0 1), female (0 1), educ (0 1(JHS) 2(SHS) 3(Bachelor dst)), working (0 1), AgeCat (0 1(<15) 2 (15-65) 3(>65)), Urban (0 1) toilet (0 1), Water Source condition (0 1 good quality), smoking (0 1), commuting frequency (0 1 (2-10) 2(>10)), working hour (0 1 (21-40) 2(41-60) 3(>60)), health insurance (0 1), social insurance (0 1), water deprivation (0 1), food deprivation (0 1) ==> All recorded as 1 for the name of the variables (ex: Female=1=Female, Female=0=Male)

I'm still struggling on understanding how to interpret my gologit result. As we know, gologit will have J-1 groups, so for my result I have 3 groups with the result details below: (KeluhanKesehatan means HealthComplaints)

1. How should I interpret for each groups? What does (1 vs 2,3,4) for group 1, (1,2 vs 3,4) for group 2, and (1,2,3 vs 4) for group 3 actually means?
2. Base of Richard Williams handout about gologit, the interpretation for negative and positive are like this:

(Positive result) Higher likelihood of being in a higher category of Y than the current one as the explanatory variable (X) increase, and (Negative result) Higher likelihood of being in a lower category of Y than the current one as the explanatory variable (X) increase

For my result, does this interpretation correct?

a. (Car Ownership) Car owners have a higher probability of complaining of minor health problems compared to non-car owners (as the sign are all negative for 3 groups)
b. (Marital Status) A married individu have a higher probability of complaining severe health problems compared to single individu (as the sign are all positive for 3 groups)

3. It is much easier when the result for all of each groups have the same sign (whether it is negative or positive), but I found it difficult to interpret when there is a different sign for one of the group, like this one, as group 3 is the only one who has negative sign:

How we should interpret for this case?

I really apreciate if anyone can help me to understand this 🤗

Thank you, have a nice day!

[Maarten Buis]

In theory, it is possible to interpret those: they are log odds ratios. At the very least you will want to use the or to get odds ratios instead of log odds ratios. In practice, it is just a huge headache to keep track of all the comparisons that these tables imply. I am not the biggest fan of marginal effects, but this is one case where marginal effects are really useful. So my advise would be to just report and interpret the marginal effects and not the raw coefficients or odds ratios. Also, at least look at graphs of predicted probabilities to figure out what is going on in your model. These are complex models where lots of different things are happening at different levels. Figuring out what your model is telling you is often quite a puzzle.

[Valerie Aulia]

In theory, it is possible to interpret those: they are log odds ratios. At the very least you will want to use the or to get odds ratios instead of log odds ratios. In practice, it is just a huge headache to keep track of all the comparisons that these tables imply. I am not the biggest fan of marginal effects, but this is one case where marginal effects are really useful. So my advise would be to just report and interpret the marginal effects and not the raw coefficients or odds ratios. Also, at least look at graphs of predicted probabilities to figure out what is going on in your model. These are complex models where lots of different things are happening at different levels. Figuring out what your model is telling you is often quite a puzzle.

Yaps, I also calculate the marginal effect, but my supervisor told me that I should interpret those positive/negative result from gologit to know the relationship between my dependen and indept var (which after thinking about it, those sign also shown in marginal effect.....). But that's alright, I choose to explain the result using marginal effect instead of the gologit coeff.

Thank you very much!

Regards,
Aulia Valerie

[Richard Williams]

I like the techniques described in this handout for making results more interpretable.

https://academicweb.nd.edu/~rwilliam.../Margins05.pdf