Brant Test Error

[pgr451]

Using Stata 13
trying to test proportional odds assumption
I've downloaded the -Brant- command

(command 1)

omodel logit var1 var2 var3 var4
(no errors)

(Command 2)

brant, detail

Error
varible _cut1 not found


What am i missing?
-sean

[Richard Williams]

There are multiple versions of brant floating around and you are probably using one of the hopelessly antiquated ones. Uninstall it. Then, from within Stata type

findit spost13_ado

and then install it. This will give you the Brant command and several others. You may also wish to get this book:

http://www.stata.com/bookstore/regre...ent-variables/

You can also get the -oparallel- command from SSC, which does Brant and several other tests of proportional odds.

Hope this helps. Rich

[Michael Sykuta]

I am receiving the same error using the -brant- command. I have spost13_ado fully installed.

Any other suggestions as to why I'm getting the "variable _cut1 not found" error and possible solutions?

[Richard Williams]

Use the -which- command to make sure that the most current version of brant is the version that is actually being executed:

. which brant
c:\ado\plus\b\brant.ado
*! version 3.3.0 2014-02-14 | long freese | spost13 release

If it is, then tell us what version of Stata you have and show us your actual commands and output. Use code tags. See the FAQ if you don't know how to do this.

[Jennifer Roberts]

Hello,
I have followed the instruction above and am still having the same error. I am using Stata/MP 14.0 for Mac (64-bit Intel) Revision 03 Aug 2015.

The error I receive is as follows:
. . which brant
/Users/jenniferroberts/Library/Application Support/Stata/ado/plus/b/brant.ado
*! version 3.3.0 2014-02-14 | long freese | spost13 release

. brant, detail
variable _cut1 not found
r(111);

Please advise

[Richard Williams]

Show the commands that preceded brant. Also, see if this example works:

webuse nhanes2f, clear
ologit health height weight
brant

The last command yields

. brant

Brant test of parallel regression assumption

              |       chi2     p>chi2      df
 -------------+------------------------------
          All |      19.56      0.003       6
 -------------+------------------------------
       height |      18.78      0.000       3
       weight |       6.01      0.111       3

A significant test statistic provides evidence that the parallel
regression assumption has been violated.

Also, do -update all- and make sure your Stata is up to date. For that matter, make sure you are running Stata 14 and not some earlier version.

[Nam Tran]

Hi,

I got similar error in running brant test and have just found how to solve it. Just run brant test after ologit command. The brant test does not work if you run it following omodel. Please noted that I use Stata 13.

Hope this helps.

[Benito D Isaac]

Thank you Richard Williams. Your comments have just been helpful to me either...

[Benito D Isaac]

I have a situation where the null hypothesis is rejected. I try to estimate a "oglm" after, but the results with the Ologit are strikingly the same...

[Richard Williams]

How did you run oglm? Did you include a heteroskedasticity equation? By default, oglm runs the same model as ologit, so if you did something like

ologit y x1 x2 x3
oglm y x1 x2 x3

the results should be the same give or take a few minor differences in, say, the 4th decimal place.

If you did have a herero equation -- Especially with large samples, statistically significant differences can be substantively trivial.

If you show commands and output using code tags (see pt. 12 in the FAQ) we might be able to better advise.

[Lilly Frost]

The original post in this chain captures the same problem I am currently having. I am using Stata/IC 15 on Windows 10 (64-bit). I have checked Stata for updates multiple times as well as re-installed spost13_ado (I also tried to re-install oparallel). In addition, I have tried running the command with different data sets, including the example provided in Richard's post dated "20 Nov 2015, 13:10" Specifically, after running:

webuse nhanes2f, clear
ologit health height weight
brant

I received the following response:

. brant
variable cut1 not found
r(111);

In addition, when I checked which "brant" I had, it is up-to-date:

. which brant
c:\ado\plus\b\brant.ado
*! version 3.3.0 2014-02-14 | long freese | spost13 release


Does anyone have other ideas for how to fix this problem?

Thank you!

[Richard Williams]

See

https://www.statalist.org/forums/for...ata-15-for-now

I know Long is working on a fix.

[Lilly Frost]

Dear Richard,

Thank you very much. I apologize for not seeing the other forum. The advice worked well and is much appreciated!

Best,
Lilly