JB Resid command

[Rose Banda]

Hi everyone,

I am trying the following in stata 16

jb resid

The command is not working. Can someone share latest command. it tried

findit jb resid

It brings the old file which I have installed but still get the

command jb is unrecognized

Thanks

[William Lisowski]

Apparently you are trying to run a Jarque-Bera test using a community contributed program available from SSC.

When I run

findit jb resid

I see

----------------------------------------------------------------------------------------------
search for jb resid                                                      (manual:  [R] search)
----------------------------------------------------------------------------------------------

Search of official help files, FAQs, Examples, and Stata Journals

Search of web resources from Stata and other users

(contacting http://www.stata.com)

1 package found (Stata Journal and STB listed first)
----------------------------------------------------

lmnjb from http://fmwww.bc.edu/RePEc/bocode/l
    'LMNJB': module to compute Lagrange Multiplier LM Jarque-Bera Normality
    Test / lmnjb computes Lagrange Multiplier Jarque-Bera normality test / for
    OLS residuals after regression.  / KW: normality / KW: regression / KW:
    OLS / KW: Lagrange Multiplier / Requires: Stata version 10 /

(click here to return to the previous screen)

(end of search)

which suggests the command you want is lmnjb rather than jb.

When you do

which jb
which lmnjb

what are the results?

[Nick Cox]

William Lisowski shows good technique, but the explanation is simpler. Rose has a variable resid and was trying to run the command jb from SSC, but either hasn't installed it yet or has installed it somewhere that Stata can't see. The answer to the former is to install it

ssc install jb

and the answer to the latter is to check your

adopath

and to use your operating system to see where jb.ado and jb.hlp are lurking on your machine.

That said, enthusiasm for this test is hard to fathom. It hinges on using asymptotic standard errors for skewness and kurtosis for all sample sizes, A few simulations will throw doubt on that procedure. in addition to the usual reservations about testing residuals for normality, there are better tests if you feel compelled to use one, which in essence is why Stata does not support this one. Wanting to use it perhaps comes down to some imitative ritual in a sub-field.

Going back to #1, the way to search would be one of

search jb

findit jb 

Oddly, or otherwise, the original findit command was my idea in an update to Stata 7, as a simple wrapper for separate commands for machine search and internet search for the same stuff. Those commands have long since been put together so in Stata 16 findit is disparaged.

[Rose Banda]

William and Nick.

Thank you both