I've been looking for a command like inteff. But it does not seem to work for me, neither for mlogit or mprobit. I've got STATA 11. OnT: Sorry, can't help.
Hello,
(For people who are familiar with the inteff-command, skip the first two paragraphs and just read the third paragraph)
I run a probit model with the interaction term GindexHHI and the separate terms Gindex and HHI (Gindex = governance strength; HHI = industry concentration). All other variables are performance control variables. The dependent variable takeovercompleted equals takeover likelihood, coded as a binary dummy variable.
-> probit takeovercompleted SalesGrowth Liquidity DEratio MBratio PEratio Size Gindex HHI GindexHHI
After the probit model, I compute the marginal effect of the interaction term GindexHHI. I do this by applying the inteff-command after the probit model which computes the cross-partial derivative of the interaction term (which corresponds to the correct marginal effect of the interaction term)
->inteff takeovercompleted SalesGrowth Liquidity DEratio MBratio PEratio Size Gindex HHI GindexHHI
I get a graph with the marginal effect of the interaction term (interaction effect in %) and a graph with z-statistics of the interaction term to derive the stat. sign.
Firstly, Stata shows the graph with the interaction effect and then overwrites this graph with the graph for the z-statistics of the interaction effect. However, I need to show both graphs in my research. I can only copy the graph with z-statistics at the moment. How can I tell Stata to show both graphs or to show them separately?
Example of the output (standard error omitted)
Variable | Obs Mean Std. Dev. Min Max
-------------+--------------------------------------------------------
interaction effect | 136 -.2959831 .1422207 -.3679685 .290872
z-statistic | 136 -1.288604 .800488 -2.093989 2.513241
I've been looking for a command like inteff. But it does not seem to work for me, neither for mlogit or mprobit. I've got STATA 11. OnT: Sorry, can't help.
ok, thanks anyway...I just use a probit model and it works, disregarding the small problem that I described above... I do not know what is mprobit at all...
There is a paper called: "Computing interaction effects and standard errors in logit and probit models", The Stata Journal (2004), Vol. 4, No. 2, by Norton, Wang and Ai
Maybe this paper can help you...
Thanks, I'll check it out. mprobit is a multinomial probit model. A model where the DV takes on more than two values.
Ok, I think this paper only refers to "normal" probit and logit models...but I am not totally sure..
I haven't used -inteff- but it looks like you need to use the savegraph1() and savegraph2() options documented in -help inteff-
Hi bukharin,
I tried that with savegraph1 and savegraph2, but it does not work, I get the error message r199 below.
I use "F:\MASTER THESE MAASTRICHT Oct2011 - Backup\Most essential data\Stata11.2" as my current working directory...
Output:
inteff takeovercompleted Gindex HHI GindexHHI SalesGrowth Liquidity DEratio MBratio PEratio Size Buseq Chems Durbl Enrgy Hlth Manuf Mo
> ney Nodur Other Shops Telcm Utils
Probit with two continuous variables interacted
(10359 observations deleted)
Variable | Obs Mean Std. Dev. Min Max
-------------+--------------------------------------------------------
_probit_ie | 100 -.9049926 .5553251 -1.990601 1.025768
_probit_se | 100 .6795854 .2501268 3.65e-21 1.491133
_probit_z | 100 -1.243099 .7676063 -2.290685 1.018358
.
. savegraph1(F:\MASTER THESE MAASTRICHT Oct2011 - Backup\Most essential data\Stata11.2\Figure1ieinperc, replace)
unrecognized command: savegraph1
r(199);
.
. savegraph2(F:\MASTER THESE MAASTRICHT Oct2011 - Backup\Most essential data\Stata11.2\Figure1ieinperc, replace)
unrecognized command: savegraph2
r(199);
No, savegraph1() and savegraph(2) are options for -inteff- - you need to use them after a comma on the same line as the -inteff- command. See -help language-
yes indeed. It works now. Thanks a lot.
Tweet |