Displaying ivqregress results

[Alberto Ruben Osella]

Dear All,
I'm trying to display ivqregress results by using the suite of commands -collect-. My outcome variable is categorized BMI then I have three equations in the model: Overweight, Obesity and Pathologic Obesity. These models have been fitted to four surveys and I want to display results with equations in the rows and surveys in the columns. I'm using the following code:

collect create iv_bmi, replace

collect _r_b _r_ci, tag(model[(ENFR2005)]): ivqregress smooth imc (i.ni_cat2inv = i.ingresocat) i.age_cat5 i.sexo i.hta i.dbt if enfr==2005, quantile(52 84 95)

collect _r_b _r_ci, tag(model[(ENFR2009)]): ivqregress smooth imc (i.ni_cat2inv = i.ingresocat) i.age_cat5 i.sexo i.hta i.dbt if enfr==2009, quantile(47 82 93)

collect dims

collect layout (coleq#colname) (results _r_b _r_ci)

and the table looks like this:

	Coefficient	95% CI
q52
Primary/High	School	0.00
Graduated	1.90	1.35 2.45
	0.00
30-39	2.00	1.84 2.16
40-49	3.05	2.87 3.23
50-59	3.53	3.33 3.73
60 or >	2.93	2.71 3.14
Male	0.00
Female	-2.13	-2.24 -2.02
No	0.00
Yes	1.56	1.38 1.74
No	0.00
Yes	1.19	0.97 1.40
Intercept	23.18	22.97 23.38
q84
Primary/High	School	0.00
Graduated	0.28	-0.69 1.25
	0.00
30-39	2.42	2.17 2.68
40-49	3.75	3.45 4.04
50-59	3.67	3.32 4.02
60 or >	2.43	2.08 2.77
Male	0.00
Female	-0.99	-1.18 -0.80
No	0.00
Yes	2.18	1.89 2.48
No	0.00
Yes	1.96	1.57 2.35
Intercept	27.41	27.06 27.77
q95
Primary/High	School	0.00
Graduated	-3.75	-5.32 -2.17
	0.00
30-39	2.31	1.73 2.88
40-49	3.09	2.47 3.72
50-59	2.79	2.03 3.55
60 or >	0.78	0.07 1.48
Male	0.00
Female	0.15	-0.30 0.60
No	0.00
Yes	2.65	1.97 3.33
No	0.00
Yes	2.99	2.11 3.86
Intercept	33.25	32.22 34.29
q47
Primary/High	School	0.00
Graduated	1.74	0.80 2.67
	0.00
30-39	1.30	1.06 1.53
40-49	1.54	1.30 1.78
50-59	1.73	1.47 1.99
60 or >	1.55	1.28 1.83
Male	0.00
Female	-2.15	-2.28 -2.02
No	0.00
Yes	2.02	1.86 2.19
No	0.00
Yes	1.46	1.21 1.71
Intercept	23.62	23.22 24.02
q82
Primary/High	School	0.00
Graduated	0.06	-1.96 2.08
	0.00
30-39	1.70	1.34 2.07
40-49	2.01	1.62 2.40
50-59	2.06	1.65 2.48
60 or >	1.36	0.92 1.81
Male	0.00
Female	-0.81	-1.00 -0.61
No	0.00
Yes	2.48	2.23 2.74
No	0.00
Yes	2.14	1.81 2.47
Intercept	27.88	27.12 28.63
q93
Primary/High	School	0.00
Graduated	-1.89	-3.77 -0.00
	0.00
30-39	1.55	0.83 2.28
40-49	2.19	1.48 2.91
50-59	2.17	1.42 2.92
60 or >	0.93	0.16 1.69
Male	0.00
Female	-0.02	-0.37 0.33
No	0.00
Yes	2.63	2.21 3.04
No	0.00
Yes	2.51	1.79 3.22
Intercept	31.67	30.58 32.75

With the code the results are displayed one model after the other and I want one by side the other.
Can someone help me?
Thank you in advance

[Andrew Musau]

You need to interact the dimensions defining the stored models with the column dimensions. Here is an example (see highlighted):

webuse assets2, clear
collect clear
collect _r_b _r_ci, tag(model[(married)]): ivqregress iqr assets (i.p401k = i.e401k) income age familysize i.ira i.pension i.ownhome educ if married
collect _r_b _r_ci, tag(model[(notmarried)]): ivqregress iqr assets (i.p401k = i.e401k) income age familysize i.ira i.pension i.ownhome educ if !married
collect dims
collect layout (colname) (cmdset[1]#(_r_b ) cmdset[2]#(_r_b))

Res.:

. collect layout (colname) (cmdset[1]#(_r_b ) cmdset[2]#(_r_b))

Collection: default
      Rows: colname
   Columns: cmdset[1]#(_r_b ) cmdset[2]#(_r_b)
   Table 1: 13 x 2

--------------------------------------------
                   |           1           2
                   | Coefficient Coefficient
-------------------+------------------------
No                 |           0           0
Yes                |    6653.299    3582.471
Household income   |    .1893788    .0827741
Age                |    188.5805    29.66423
Household size     |   -295.8322   -102.7572
No                 |           0           0
Yes                |    24685.66    18706.49
No pension         |           0           0
Receives pension   |   -1399.455    19.17392
No                 |           0           0
Yes                |   -232.1314    243.4921
Years of education |   -182.4329   -14.01141
Intercept          |   -9527.379   -2142.279
--------------------------------------------

[Alberto Ruben Osella]

Dear Andrew,
thank you very much. I've added to my code the following row: collect layout (colname) (cmdset[1]#(_r_b _r_ci) cmdset[2]#(_r_b _r_ci)) and now I've obtain the following messagge:
. collect layout (colname) (cmdset[1]#(_r_b _r_ci) cmdset[2]#(_r_b _r_ci))

Collection: iv_bmi
Rows: colname
Columns: cmdset[1]#(_r_b _r_ci) cmdset[2]#(_r_b _r_ci)

Your layout specification does not uniquely match any items. One or more of the following
dimensions might help uniquely match items: age_cat5, coleq, dbt, hta, model, ni_cat2inv,
sexo.

[Andrew Musau]

From #1, you are estimating across quantiles. These are levels within the dimension coleq. The following is not tested:

local quantiles
collect levelsof coleq
foreach q in `s(levels)'{
    local quantiles `quantiles' coleq[`q']#(_r_b _r_ci)
}
collect layout (colname) (`quantiles')

If this does not work, please provide a data example using dataex that creates an example of the tables in #1.

[Andrew Musau]

This example confirms #4 works.

webuse assets2, clear
collect clear
collect _r_b _r_ci: ivqregress iqr assets (i.p401k = i.e401k) income age familysize, quantile(52 84 95)

local quantiles
collect levelsof coleq
foreach q in `s(levels)'{
    local quantiles `quantiles' coleq[`q']#(_r_b _r_ci)
}
collect layout (colname) (`quantiles')

Res.:

Collection: default
      Rows: colname
   Columns: coleq[q52]#(_r_b _r_ci) coleq[q84]#(_r_b _r_ci)
            coleq[q95]#(_r_b _r_ci)
   Table 1: 6 x 6

------------------------------------------------------------------------------------------------------------------
                 |         q52                 q52         q84                 q84         q95                 q95
                 | Coefficient        95% CI       Coefficient        95% CI       Coefficient        95% CI      
-----------------+------------------------------------------------------------------------------------------------
No               |           0                               0                               0                    
Yes              |    5740.978  4507.304  6974.652    9701.633  5756.123  13647.14    3487.239 -5126.108  12100.59
Household income |    .3174947   .283479  .3515104    .9791422  .9125421  1.045742    2.012705  1.717391  2.308018
Age              |    224.4077  195.8628  252.9527    905.9105  791.2425  1020.578    1582.251  1287.617  1876.884
Household size   |   -737.6436 -873.8362 -601.4511   -1763.951 -2140.192 -1387.709   -2574.577 -3239.373  -1909.78
Intercept        |   -12472.43 -13892.06 -11052.81   -35600.73 -39089.47 -32111.98   -56004.47 -63912.87 -48096.07
------------------------------------------------------------------------------------------------------------------

.

[Alberto Ruben Osella]

I've modified the code a bit, and now it works. Thank you