Stumped by collect/table for adding results of estat esize to result for ANOVA

[ghoetker]

This seems like it should be simple, but it stumps me. I run an ANOVA, followed by "estat esize". I'd like to add a column with the eta2 from estat esize to a table of the ANOVA results.

I can get the ANOVA tale, but am stumped how to add the column for the eta2. Can anyone help me out? Thanks.

collect: anova techs_chosen_150  i.exp_level##i.opt_level
collect: estat esize

collect layout (term[Model 1.exp_level 1.opt_level 1.exp_level#1.opt_level Residual]) (result[SS df ms F ] )

To clarify, adding eta2 to the result[SS...] does not work, even though it shows up in "collect levelsof result". Same with esize.

[Hemanshu Kumar]

The essential problem is that the rows identifier for anova is in the dimension term, while the equivalent for estat esize is in rowname . We need to add the relevant tags.

Consider this example:

collect clear
sysuse auto, clear

collect: anova price foreign##rep78
collect: estat esize

local lvls  Model 0.foreign 1.rep78 0.foreign#1.rep78 Residual
local labels Model foreign rep78 foreign#rep78 Residual
local num_lvls: list sizeof lvls
forval l = 1/`num_lvls' {
    local lvl: word `l' of `lvls'
    local lbl: word `l' of `labels'
    collect addtags term[`lvl'], fortags(rowname[`lvl'])
    collect label levels term `lvl' "`lbl'", modify
}

collect style header result[esize], title(hide) level(hide)
collect style header colname, title(hide)
collect style cell result[SS], nformat(%13.1fc)
collect style cell result[F] colname[eta2 lb_eta2 ub_eta2], nformat(%5.4f)
collect layout (term) (result[SS df ms F] colname[eta2 lb_eta2 ub_eta2]#result[esize] )

which produces:

. collect preview
-------------------------------------------------------------------
              |            SS df F statistic   eta2 lb_eta2 ub_eta2
--------------+----------------------------------------------------
Model         |  24,684,607.0  7      0.3896 0.0428       .  0.0519
foreign       |     395,125.9  1      0.0437 0.0007       .  0.0573
rep78         |   6,609,429.0  4      0.1826 0.0118       .  0.0256
foreign#rep78 |  16,312,125.7  2      0.9011 0.0287       .  0.1277
Residual      | 552,112,351.8 61                                  
Total         | 576,796,958.9 68                                  
-------------------------------------------------------------------

[ghoetker]

Thank you again, Hemanshu.

Sometimes I worry that the question I ask has a blindingly obvious answer that I've just missed...I don't think this is one of those times.

The new-ish collect/etable/dtable/table approach is really powerful and flexible. I'm just finding myself slow to get my head around it.