Ordered probit and goodness of fit

[Barbora Sedova]

Dear all,

is there any goodness of fit test for an ordered probit model?
Which other diagnosis tests would you recommend me for an ordered probit model?
Many thanks.

Best,

Barbora

[Barbora Sedova]

lrtest should do the trick, no??

[Richard Williams]

linktest is one option.

If you were using ologit, you could use the user-wriitten brant command (findit spost13_ado) to test the parallel lines assumption. For oprobit, you can do something similar with the user-written gologit2 and omodel commands. See pp. 10-11 of

https://www3.nd.edu/~rwilliam/xsoc73994/Ologit01.pdf

The leastlikely command (also part of spost13) can help to identify relatively deviant cases that may warrant closer inspection.