Tables and stored results
table, tablefull, store() with esttab, and the returned matrices
Every balanceplot call returns its statistics, and the table options print them beneath the plot. The examples use nlsw88 and compare union and nonunion workers.
sysuse nlsw88, clear(NLSW, 1988 extract)table and tablefull
table prints one compact table per group comparison: the two group means, the standardized imbalance, and its p-value.
balanceplot wage age i.married i.race tenure, group(union) tableNOTE: 10 observations were excluded due to missing data on
at least one covariate, group(), or outcome() variable.
Base category = 0_Nonunion
Base selected by the 0/1 two-group default.
N used in balance calculations
- N for union = 0_Nonunion: 1408
- N for union = 1_Union: 460
Balance results: Nonunion (N=1,408) vs Union (N=460)
| union=0 | union=1 | Standard~d | _
Covariate | Mean | Mean | Imbalance | p-value
-------------------------+------------+------------+------------+-----------
| | | |
Hourly wage | 7.227 | 8.685 | 0.352 | 0.000
Age in current year | 39.205 | 39.276 | 0.023 | 0.664
Married (ref = Single) | 0.665 | 0.607 | -0.121 | 0.023
-------------------------+------------+------------+------------+-----------
Race | | | |
White(ref) | 0.743 | 0.654 | 0.000 | 1.000
Black (ref = White) | 0.246 | 0.328 | 0.183 | 0.001
Other (ref = White) | 0.011 | 0.017 | 0.051 | 0.319
-------------------------+------------+------------+------------+-----------
| | | |
Job tenure (years) | 6.141 | 7.888 | 0.303 | 0.000 tablefull adds the standard error and confidence limits.
balanceplot wage age i.married i.race tenure, group(union) tablefullNOTE: 10 observations were excluded due to missing data on
at least one covariate, group(), or outcome() variable.
Base category = 0_Nonunion
Base selected by the 0/1 two-group default.
N used in balance calculations
- N for union = 0_Nonunion: 1408
- N for union = 1_Union: 460
Balance results: Nonunion (N=1,408) vs Union (N=460)
| union=0 | union=1 | Standard~d | _
Covariate | Mean | Mean | Imbalance | SE
-------------------------+------------+------------+------------+------------
| | | |
Hourly wage | 7.227 | 8.685 | 0.352 | 0.053
Age in current year | 39.205 | 39.276 | 0.023 | 0.054
Married (ref = Single) | 0.665 | 0.607 | -0.121 | 0.053
-------------------------+------------+------------+------------+------------
Race | | | |
White(ref) | 0.743 | 0.654 | 0.000 | 0.000
Black (ref = White) | 0.246 | 0.328 | 0.183 | 0.052
Other (ref = White) | 0.011 | 0.017 | 0.051 | 0.051
-------------------------+------------+------------+------------+------------
| | | |
Job tenure (years) | 6.141 | 7.888 | 0.303 | 0.052
| 95% CI | _
Covariate | Lower Upper | p-value
-------------------------+------------------------+-----------
| |
Hourly wage | 0.247 0.457 | 0.000
Age in current year | -0.082 0.129 | 0.664
Married (ref = Single) | -0.226 -0.017 | 0.023
-------------------------+------------------------+-----------
Race | |
White(ref) | 0.000 0.000 | 1.000
Black (ref = White) | 0.080 0.286 | 0.001
Other (ref = White) | -0.049 0.150 | 0.319
-------------------------+------------------------+-----------
| |
Job tenure (years) | 0.201 0.405 | 0.000 Formatting the table: decimals(), width(), and labwidth()
decimals() sets the digits after the decimal point (default 3), width() the width of each statistic column (default 10), and labwidth() the width of the label column (default 24; longer labels are abbreviated).
balanceplot wage age i.married i.race tenure, group(union) table decimals(2) width(8) labwidth(30)NOTE: 10 observations were excluded due to missing data on
at least one covariate, group(), or outcome() variable.
Base category = 0_Nonunion
Base selected by the 0/1 two-group default.
N used in balance calculations
- N for union = 0_Nonunion: 1408
- N for union = 1_Union: 460
Balance results: Nonunion (N=1,408) vs Union (N=460)
| union=0 | union=1 | Standa~d | _
Covariate | Mean | Mean | Imbala~e | p-value
-------------------------------+----------+----------+----------+---------
| | | |
Hourly wage | 7.23 | 8.68 | 0.35 | 0.00
Age in current year | 39.21 | 39.28 | 0.02 | 0.66
Married (ref = Single) | 0.66 | 0.61 | -0.12 | 0.02
-------------------------------+----------+----------+----------+---------
Race | | | |
White(ref) | 0.74 | 0.65 | 0.00 | 1.00
Black (ref = White) | 0.25 | 0.33 | 0.18 | 0.00
Other (ref = White) | 0.01 | 0.02 | 0.05 | 0.32
-------------------------------+----------+----------+----------+---------
| | | |
Job tenure (years) | 6.14 | 7.89 | 0.30 | 0.00 Returned matrices
The first nonbase comparison is returned in r(bias1) (the second in r(bias2), and so on), and an export-ready version of each table in r(table1), r(table2), …
return listscalars:
r(threshold) = .
r(stored) = 0
r(absolute) = 0
r(labwidth) = 30
r(width) = 8
r(fadealpha) = .05
r(base_sumw) = 1408
r(base_n_unweighted) = 1408
r(base_n) = 1408
r(weighted) = 0
r(cohensh) = 0
r(fadens) = 0
r(level) = 95
r(nrows) = 7
r(ngroups) = 2
r(base) = 0
macros:
r(headings) : " 1.race = `"{bf:Race}"'"
r(tablemode) : "compact"
r(sortmode) : "original"
r(xtitle) : "Standardized Imbalance"
r(measure) : "standardized_difference"
r(plotcommand) : " (matrix(bias_0_1[,4]), ci((bias_0_1[,6] bias_0_1[,7])) label(`"Nonu.."
r(coeflabels) : " 1.married = `"Married (ref = Single)"' 1.race = `"White (ref)"' 2.rac.."
r(rownames) : "wage age 1.married 1b.race 2.race 3.race tenure"
r(tablefullmatrices) : "tablefull_0_1"
r(tablematrices) : "table_0_1"
r(matrices) : "bias_0_1"
r(groups) : "0 1"
r(mode) : "group"
matrices:
r(tablefull1) : 7 x 7
r(table1) : 7 x 4
r(bias1) : 7 x 7matrix list r(table1)r(table1)[7,4]
mean_g0 mean_g1 std_imbala~e p_value
wage 7.2269385 8.6845396 .35213133 5.918e-11
age 39.205256 39.276087 .0233567 .66430587
1.married .66477273 .60652174 -.12117473 .02291971
1b.race .74289773 .65434783 0 1
2.race .24573864 .32826087 .18305107 .00050025
3.race .01136364 .0173913 .05061108 .3192061
tenure 6.1407434 7.8882246 .3028704 6.944e-09Stored results for esttab: store()
store(stub) saves the results as stored estimates for esttab and estout (Jann). For a two-group comparison the names are stub_mean_g# for each group and stub_imbalance; the imbalance set carries standard errors, so esttab reports significance for it. replace overwrites results stored earlier under the same names.
balanceplot wage age i.married i.race tenure, group(union) store(bp, replace)NOTE: 10 observations were excluded due to missing data on
at least one covariate, group(), or outcome() variable.
Base category = 0_Nonunion
Base selected by the 0/1 two-group default.
N used in balance calculations
- N for union = 0_Nonunion: 1408
- N for union = 1_Union: 460esttab bp_mean_g0 bp_mean_g1 bp_imbalance, se label mtitles--------------------------------------------------------------------
(1) (2) (3)
bp_mean_g0 bp_mean_g1 bp_imbalance
--------------------------------------------------------------------
Hourly wage 7.227 8.685 0.352***
(0.0535)
Age in current year 39.21 39.28 0.0234
(0.0538)
Married 0.665 0.607 -0.121*
(0.0532)
White 0.743 0.654 0
(.)
Black 0.246 0.328 0.183***
(0.0525)
Other 0.0114 0.0174 0.0506
(0.0508)
Job tenure (years) 6.141 7.888 0.303***
(0.0520)
--------------------------------------------------------------------
Observations 1408 460 1868
--------------------------------------------------------------------
Standard errors in parentheses
* p<0.05, ** p<0.01, *** p<0.001With three or more groups each nonbase comparison is stored as stub_imb_g#:
balanceplot wage age i.married tenure, group(race) store(br, replace)NOTE: 15 observations were excluded due to missing data on
at least one covariate, group(), or outcome() variable.
Base category = 1_White
Base selected as the largest complete-case group.
N used in balance calculations
- N for race = 1_White: 1627
- N for race = 2_Black: 578
- N for race = 3_Other: 26esttab br_mean_g1 br_mean_g2 br_mean_g3 br_imb_g2 br_imb_g3, se label mtitles----------------------------------------------------------------------------------------------------
(1) (2) (3) (4) (5)
br_mean_g1 br_mean_g2 br_mean_g3 br_imb_g2 br_imb_g3
----------------------------------------------------------------------------------------------------
Hourly wage 8.106 6.876 8.551 -0.222*** 0.0795
(0.0502) (0.210)
Age in current year 39.27 38.81 39.31 -0.150** 0.0124
(0.0488) (0.193)
Married 0.702 0.471 0.692 -0.483*** -0.0207
(0.0474) (0.195)
Job tenure (years) 5.808 6.502 4.949 0.125** -0.161
(0.0481) (0.202)
----------------------------------------------------------------------------------------------------
Observations 1627 578 26 2205 1653
----------------------------------------------------------------------------------------------------
Standard errors in parentheses
* p<0.05, ** p<0.01, *** p<0.001The stored sets are ordinary estimation results, so estimates dir lists them and esttab can write them to a file with its using clause. The matchweight() and contreat() forms store their own sets, described on the Matching and teffects and Continuous treatments pages.