Assessing covariate balance on indicator variables using the -tebalance summarize- command
Hi all,
After running most of the estimating commands from the -teffects- family, one can then assess balancing of the raw and weighted samples using the -tebalance summarize- command.
However, I am trying to understand the output for indicator variables specifically using this command.
For example, using the following data:
Where "operation" is a multi-level indicator variable as such:
I issue these commands:
and obtain the following result from the -tebalance summarize- command:
But, this doesn't show me the standardized difference for the first level of the indicator variable "operation". Does anyone have any suggestions as to how I might be able to get the standardized differences for, in this example, " CABG + valve repair/replacement" which is level 1 of the indicator?
Many thanks, Phil
After running most of the estimating commands from the -teffects- family, one can then assess balancing of the raw and weighted samples using the -tebalance summarize- command.
However, I am trying to understand the output for indicator variables specifically using this command.
For example, using the following data:
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input float group long sex double age long operation float primary 1 2 66 3 1 0 2 67 3 0 0 2 76 3 0 1 1 72 3 0 0 1 79 4 0 0 2 71 1 1 1 1 58 4 0 1 2 68 3 0 0 2 49 4 0 0 2 67 3 1 1 2 60 3 0 1 2 70 3 1 1 2 71 1 1 1 2 65 3 0 1 2 77 3 1 0 2 58 3 1 0 2 63 3 0 1 1 86 1 1 0 2 63 3 1 0 2 73 3 0 1 1 62 4 1 0 1 65 3 0 0 2 61 1 0 0 2 71 3 0 0 2 61 3 0 0 2 70 2 0 1 2 84 3 1 1 2 54 3 0 1 1 74 1 0 0 1 56 3 0 1 1 67 3 0 1 2 64 3 1 1 2 57 3 0 0 2 70 3 0 0 2 62 3 1 1 1 70 4 0 1 2 62 1 0 0 2 57 3 0 1 2 65 1 1 0 2 73 3 0 0 2 73 3 0 0 1 61 3 1 1 2 66 3 0 1 2 82 1 0 0 2 68 3 0 1 2 74 3 1 1 2 63 3 0 1 2 65 3 0 1 2 73 3 0 0 2 69 3 0 0 2 73 3 0 1 2 60 3 0 0 2 67 4 1 0 1 55 4 1 1 1 68 3 0 1 2 58 3 1 0 2 77 3 1 1 2 67 3 0 0 2 85 3 0 0 2 68 3 1 0 1 66 4 0 1 2 64 3 0 1 2 70 3 1 0 2 61 3 0 0 1 80 3 0 0 2 67 3 0 0 1 68 3 1 1 2 76 3 1 1 2 74 3 0 0 1 79 3 0 1 2 52 3 0 1 2 80 3 1 1 2 58 3 0 0 2 58 3 0 1 2 57 3 0 1 2 71 1 0 0 2 77 3 1 0 2 81 3 0 0 2 63 3 0 0 2 60 3 0 1 2 67 2 0 0 2 57 4 0 1 2 74 3 1 1 2 66 3 1 1 2 55 3 0 1 2 56 3 1 1 2 60 3 0 1 2 63 3 0 0 2 59 3 0 0 1 60 3 0 0 2 57 3 0 0 1 80 3 1 1 2 72 3 0 0 2 42 3 0 0 2 57 3 1 1 1 67 4 1 0 2 25 4 0 1 2 74 3 0 0 2 66 3 0 1 2 44 3 1 end label values group treatment label def treatment 0 "isoflurane", modify label def treatment 1 "sevoflurane", modify label values sex sex label def sex 1 "Female", modify label def sex 2 "Male", modify label values operation operation label def operation 1 "CABG + valve repair/replacement", modify label def operation 2 "CABG, OFF-pump", modify label def operation 3 "CABG, ON-pump", modify label def operation 4 "Single valve repair/replacement", modify label var group "randomization group" label var sex "sex" label var age "patient's age" label var operation "operation type" label var primary "Primary outcome (ICU LOS >= 2d or 30d mortality)"
Code:
Numeric Label 1 CABG + valve repair/replacement 2 CABG, OFF-pump 3 CABG, ON-pump 4 Single valve repair/replacement
Code:
teffects ipw (primary) (group i.sex c.age i.operation, logit) tebalance summarize
Code:
Covariate balance summary
Raw Weighted
-----------------------------------------
Number of obs = 100 100.0
Treated obs = 50 50.0
Control obs = 50 50.0
-----------------------------------------
-----------------------------------------------------------------
|Standardized differences Variance ratio
| Raw Weighted Raw Weighted
----------------+------------------------------------------------
sex |
Male | .0991189 -.030093 .8601399 1.046878
|
age | .1194342 .0269028 .6448423 .6807031
|
operation |
CABG, OFF-pump | 0 -.0000509 1 .999641
CABG, ON-pump | -.095702 -.0021854 1.14 1.002942
Single valve~t | -.1907122 .0059734 .6112957 1.014956
-----------------------------------------------------------------
Many thanks, Phil
Comment