Problem reporting multivariate multiple regression

[Qi vd Kolk]

Hi all,

For my master thesis I conducted a multivariate multiple regression analysis as I have three dependent variables (moderately correlated with each other) and several predictor variables (independent, controls and moderation).
Now, I would like to report three models for each dependent variable as one would show with a multiple regression, thus:
Model 1: DV + controls,
Model 2: DV + controls + independent variable
Model 3: DV + controls + independent variable + interaction effect.

I ran three separate regressions with mvreg to get model 1,2 and 3 and used the estout command to make a regression table but I got the output for the dependent variables underneath each other instead of getting them side by side. Below I added my regression table.

Does someone know how to create a regression table with three different models for each dependent variable and then display the output side by side ??
I've looked at different commands like outreg2, mkcorr and asdoc but I can't find what I'm looking for.


The dependent variables in this table are: post-crisis reputation and post-crisis Anger and post-crisis CPI

Multivariaat Regression results
Independent variable	Model 1	Model 2	Model 3
DV: Postcrisisreputation
CrisisResponsibility	-0.060	-0.060	-0.011
	(0.124)	(0.122)	(0.122)
CrisisInvolvement	0.017	-0.014	-0.021
	(0.092)	(0.092)	(0.088)
Gender	-0.112	-0.127
	(0.187)	(0.185)
Employment	0.138	0.111	0.126
	(0.128)	(0.128)	(0.123)
Age	-0.130*	-0.134*	-0.121*
	(0.055)	(0.054)	(0.052)
Crisis strategy		-0.301	-0.265
		(0.160)	(0.157)
Priorreputation			0.343
			(0.180)
Crisis strategy=1 #Priorreputation			-0.091
			(0.251)
Constant	3.149***	3.300***	3.283***
	(0.079)	(0.111)	(0.110)
DV: Post-crisis Anger
Crisis Responsibility	0.313*	0.312*	0.263
	(0.137)	(0.136)	(0.136)
CrisisInvolvement	0.182	0.214*	0.223*
	(0.102)	(0.102)	(0.097)
Gender	0.161	0.176
	(0.208)	(0.206)
Employment	0.044	0.073	0.048
	(0.142)	(0.142)	(0.137)
Age	0.015	0.019	0.001
	(0.061)	(0.060)	(0.058)
Crisis strategy=1		0.309	0.265
		(0.178)	(0.174)
Priorreputation			-0.449*
			(0.200)
Crisis strategy=1#Priorreputation			0.224
			(0.279)
Constant	3.327***	3.173***	3.196***
	(0.087)	(0.124)	(0.122)
DV: Post-crisis CPI
CrisisResponsibility	-0.073	-0.073	-0.036
	(0.087)	(0.087)	(0.088)
CrisisInvolvement	0.004	-0.006	0.010
	(0.065)	(0.066)	(0.063)
Gender	0.105	0.100
	(0.132)	(0.132)
Employment	0.089	0.080	0.067
	(0.090)	(0.091)	(0.089)
Age	-0.010	-0.012	-0.015
	(0.039)	(0.039)	(0.037)
Crisis strategy=1		-0.101	-0.087
		(0.114)	(0.113)
Priorreputation			0.181
			(0.130)
Crisis strategy=1#Priorreputation			0.002
			(0.181)
Constant	2.680***	2.731***	2.721***
	(0.055)	(0.080)	(0.079)
R-squared	0.0576 0.1193 0.0218	0.0901 0.1454 0.0295	0.1354 0.1913 0.0620
R-squared, adjusted
Number of observations	106	106	106
* p<0.05, ** p<0.01, *** p<0.001

Thanks in advance!

[William Lisowski]

The output of help estout describes the unstack option

      [no]unstack                    place equations from multiple-
                                       equation models in separate columns

[Qi vd Kolk]

Thank you very much for your reply!
However, I can't find how I can use unstack in combination with estout.
I tried several things, for example the following and Stata says that unstack is unrecognised:

estout mmr_1 mmr_2 mmr_3 using regression.txt, cells(b(star fmt(%9.3f)) se(par)) stats(r2 r2_a N, fmt(%9.3f %9.3f %9.0g) labels("R-squared" "R-squared, adjusted" "Number of observations")) layout (unstack) legend label noomitted nobaselevels varlabels(_cons Constant)


How I can use [no] unstack in combination with estout?

Thanks in advance!

[William Lisowski]

The unstack option is not part of the layout() option. (The estout command and documentation are confusing, the word "layout" is both a general descripton of a group of options and the name of a specific option.)

In what you have written in post #3, replace "layout(unstack)" with "unstack".

[Qi vd Kolk]

Thank you very much, I tried unstack and it worked!!!
However, I'm wondering if this regression table is clear or confusing to other people.

	M1			M2			M3
	Post-Rep	Post-Ang	PostCPI	Post-rep	PostAng	PostCPI	Post-rep	PostAng	PostCPI
	b/se	b/se	b/se	b/se	b/se	b/se	b/se	b/se	b/se
Crisis responsibility	-0.060	0.313*	-0.073	-0.060	0.312*	-0.073	-0.011	0.263	-0.036
	(0.124)	(0.137)	(0.087)	(0.122)	(0.136)	(0.087)	(0.122)	(0.136)	(0.088)
Crisis Involvement	0.017	0.182	0.004	-0.014	0.214*	-0.006	-0.021	0.223*	0.010
	(0.092)	(0.102)	(0.065)	(0.092)	(0.102)	(0.066)	(0.088)	(0.097)	(0.063)
cGender	-0.112	0.161	0.105	-0.127	0.176	0.100
	(0.187)	(0.208)	(0.132)	(0.185)	(0.206)	(0.132)
Employment	0.138	0.044	0.089	0.111	0.073	0.080	0.126	0.048	0.067
	(0.128)	(0.142)	(0.090)	(0.128)	(0.142)	(0.091)	(0.123)	(0.137)	(0.089)
Age	-0.130*	0.015	-0.010	-0.134*	0.019	-0.012	-0.121*	0.001	-0.015
	(0.055)	(0.061)	(0.039)	(0.054)	(0.060)	(0.039)	(0.052)	(0.058)	(0.037)
Crisis strategy=1				-0.301	0.309	-0.101	-0.265	0.265	-0.087
				(0.160)	(0.178)	(0.114)	(0.157)	(0.174)	(0.113)
Prior reputation							0.343	-0.449*	0.181
							(0.180)	(0.200)	(0.130)
Crisis strategy#Priorreputation							-0.091	0.224	0.002
							(0.251)	(0.279)	(0.181)
Constant	3.149***	3.327***	2.680***	3.300***	3.173***	2.731***	3.283***	3.196***	2.721***
	(0.079)	(0.087)	(0.055)	(0.111)	(0.124)	(0.080)	(0.110)	(0.122)	(0.079)
R-squared	0.058	0.119	0.022	0.090	0.145	0.030	0.135	0.191	0.062
R-squared, adjusted
Observations	106			106			106

Therefore, I am wondering whether there is a command that could generate a table with all three models underneath each dependent variable?
For example something like this:

	Post-rep			Post-anger			Post-CPI
	M1	M2	M3	M1	M2	M3	M1	M2	M3

Furthermore, I can't find how to display the adjusted R-squared and I think it is important to report in my thesis.
Is there a command to also display the adjusted r-squared when conducting a multivariate multiple regression?


Next to this, I have some questions about multicollinearity and heteroscedascity:
1. I can't find a way to show the VIF scores to test for multicollinearity. For a multiple regression one uses estate vif, but is there also a command for that when using mvreg?
2. I can't find a way to test for heteroscedascity when conducting a multivariate regression, with a multiple regression one uses vce(robust) but what should I use when using mvreg?


Lastly, the three dependent variables post-crisis reputation, post-crisis anger and post-crisis cpi are measured after the treatment (post-test) but I also measured these variables before the treatment (pre-test). As my pretest variables (pre reputation, pre-anger, pre-cpi) also correlate with each other and with the post-test dependent variables, I was wondering if someone could give me advice on which multivariate pre-test post test technique I should use as I can't find any good sources or examples on the internet. Now, I used independent t-test gain score technique but I don't think that is the right method given the correlations between those variables.

Table 4: Correlation results
	1	2	3	4	5	6	7	8	9	10	11	12	13
(1) Crisis strategy	1
(2) Pre-reputation	-.012	1
(3) Pre-anger	-.042	-.453**	1
(4) Pre-cpi	-.081	.207*	-.405**	1
(5) Post-reputation	-.167	.394**	-.064	.135	1
(6) Post-CPI	-.101	.204*	-.176	.622**	.444**	1
(7) Post- anger	.102	-.415**	.479**	-.305**	-.536**	-.311**	1
(8) Crisis responsibility	-.036	-.356**	.394**	-.221*	-.055	-.084	.274**	1
(9) Crisis involvement	-.194*	-.300**	.512**	-.254**	.015	.012	.258**	.251**	1
(10) Gender	-.040	-.224*	.292**	-.059	-.003	.052	.141	.133	.285**	1
(11) Employment	-.148	.101	-.086	-.011	-.012	.083	.003	-.100	-.010	-.322**	1
(12) Age	-.089	-.089	-.105	.069	-.192*	.008	-.005	-.037	-.095	-.363**	.578**	1
(13) Prior reputation	-.065	.276**	-.053	.135	.246*	.215*	-.282**	-.207*	-.061	.035	.018	-.017	1

Note: Crisis strategy 0= compensation, 1= apology, *p<0.05, **p<0.01

Thanks in advance!