etable acting strangely

[Thijs van Rens]

I have just switched to the new "etable" from "estimates table", but I am having a bit of trouble.

.est table bl_ols bl_wls bl_wlw, b(%7.2f) se(%7.2f) t(%7.1f) p(%7.2f) stats(N N_clust) stf(%7.0f) varlab

runs fine and gives the following output:

--------------------------------------------------------
Variable | bl_ols bl_wls bl_wlw
-------------------------+------------------------------
Eco label | -0.02 -0.22 0.24
| 0.39 0.36 0.34
| -0.1 -0.6 0.7
| 0.96 0.53 0.49
Constant | 28.15 28.25 27.29
| 0.29 0.26 0.25
| 98.0 109.4 108.7
| 0.00 0.00 0.00
-------------------------+------------------------------
N | 1836 1836 1835
N_clust |
--------------------------------------------------------
Legend: b/se/t/p

However, while

.etable, estimates(bl_ols bl_wls bl_wlw) cstat(_r_b, nformat(%7.2f)) cstat(_r_se, nformat(%7.2f)) cstat(_r_z_abs, nformat(%7.1f)) cstat(_r_p, nformat(%7.2f)) stars(.1 "*" .05 "**" .01 "***", attach(_r_b)) mstat(N, label(Nr obs)) mstat(N_clust, label(Nr clusters) nformat(%7.0f)) col(estimates) varlab

runs too and gives nicer-looking output (and allows me to export to Excel, which is great), it puts the same independent variables in different rows of the table:

------------------------------------------
bl_ols bl_wls bl_wlw
------------------------------------------
Eco label -0.02 -0.22
(0.39) (0.36)
0.1 0.6
0.96 0.53
Intercept 28.15 *** 28.25 ***
(0.29) (0.26)
98.0 109.4
0.00 0.00
Eco label 0.24
(0.34)
0.7
0.49
Intercept 27.29 ***
(0.25)
108.7
0.00
Nr obs 1836 1836 1835
------------------------------------------
*** p<.01, ** p<.05, * p<.1

The only difference between model bl_wls and bl_wlw is that the dependent variable is different.

Thank you for your help!

Thijs

[Jeff Pitblado (StataCorp)]

Interactively, you can use the Tables Builder to remove coleq from the layout specification to get the row alignment you want. Here is a simplified example using the publicly available auto data.

sysuse auto
gen logprice = log(price)

regress logprice mpg turn trunk
est store regress

glm price mpg turn trunk, link(log)
est store glm

etable, estimates(regress glm)
db tables

Click on the down arrow in the first term of the rows specification.

Click on the dimension coleq to remove it from the term.

Now the coefficients are aligned in the rows.

Alternatively, etable has option eqrecode() that gives you the ability to recode the column equations to a common level/string. For example,

. etable, estimates(regress glm) eqrecode(logprice=price)

---------------------------------------
                       logprice  price
---------------------------------------
Mileage (mpg)            -0.035  -0.077
                        (0.010) (0.019)
Turn circle (ft.)        -0.013  -0.023
                        (0.014) (0.016)
Trunk space (cu. ft.)     0.014  -0.002
                        (0.012) (0.014)
Intercept                 9.716  11.252
                        (0.696) (0.907)
Number of observations       74      74
---------------------------------------

[Thijs van Rens]

Thank you for your quick and very adequate reply. The eqrecode() option is exactly what I needed.