One critical thing that we need to consider is called the proportional odds assumption (or parallel lines assumption). This means that the coefficients that describe the relationship between, for example, the lowest versus all higher categories of the outcome variable are the same as those that describe the relationship between the next lowest category and all higher categories, and so on. Because the relationships between all pairs of categories are assumed to be the same, we only get one estimate for each x-variable. 

For this purpose, we can use the brant command. This command requires that you install a user-written package first. So, if you have not installed it already, type: 

search spost13_ado

Click on the first link in the list, and then choose Click here to install.

More information help brant

Practical example

The first step is re-run the multiple ordinal regression model. The quietly option is included in the beginning of the command to suppress the output. 

quietly ologit educ gpa bullied ib1.bestfriends if pop_ordinal==1

Then we use the brant command, which produces the following output: 

brant, detail

We can see from the test that we have an occurrence of a significant test statistic: gpa has a p-value (p>chi2) which is below 0.05 (0.000), thus violating the proportional odds assumption. While we will not explore this matter further here, a possible solution would be to transform this variable (e.g. through categorisation, or log transformation), or to choose some other type of analysis (e.g. transforming the outcome into a binary version and conduct logistic regression instead). 

Additional alternatives

If the proportional odds assumption is violated, it might be interesting to explore other alternatives to ologit. Among these, we have the gologit2 and omodel commands. Both of them are user-written packages.