cross-tabulation for two variables, by two groups, summarize mean and sd

[Xianli Wang]

I have some variable look like this.

city	gender	age	income
A	Female	23	309
A	Male	14	456
A	Female	67	342
A	Female	45	653
B	Male	42	155
B	Female	24	678
B	Female	35	256
B	Male	49	432

And I want to create cross-tabulation for age and income, by group city and gender, summarizing mean and sd
like this:

	female		male
	age	income	age	income
A	mean
	sd
B

I tried tabulate,
tabulate gender city, summarize(age) nofreq, wrap
it works well but can only summarize one variable at once. I can't do something like summarize(age income)

I also tried tabstat
tabstat age income, by(city) stat(mean sd)
it can only tabulate one category as by(city). I can't do something like by(city gender)

I finally tried table
table city gender, contents(mean age sd age mean age sd age)
It returns the numbers I want, but in an ugly display since it put all the statistics in one cell without notation.

like

	female	male
A	mean(age)
	sd(age)
	mean(income)
	sd(income)
B

I also read about tabout and tablemat. And a little bit confused now.
It seems that tabout just put those tables in a beautiful look but actually.
And tablemat keeps alerting me invalid name, or asking me "Please provide only one statistic at a time"
tablemat age income, stat(mean sd) by(city gender)
When I tried put only one statistic
tablemat age income, stat(mean) by(city gender)
tablemat provides me something definitely not what I want.

	All	city_A	city_B	gender_F	gender_M
age	37.375	37.25	37.5	38.8	35
income	410.125	440	80.25	447.6	347.6667

Please help me with this problem.
I'm running out of solutions.

Any help will be highly appreciated.
Thanks a lot!

[Oded Mcdossi]

This is not exactly what you are looking for, but you can use the by option with table, something like this:

table city, c(mean age sd age mean age sd age) f(%6.2fc) by(gender)

[Nick Cox]

One solution is to create a composite variable first:

egen group = group(city gender), label

[Xianli Wang]

This is not exactly what you are looking for, but you can use the by option with table, something like this:

table city, c(mean age sd age mean age sd age) f(%6.2fc) by(gender)

Thanks Oded. This is an alternative although still not what I want.
I still want to put one group in the row, and one group in the column so that readers can look up for group combinations easier.
Do you have any ideas about it?
Thanks for your help!

[Xianli Wang]

One solution is to create a composite variable first:

egen group = group(city gender), label

Thanks Nick!
I did this and got something looks like

group	age	income
A Female	mean
	sd
A Male	mean
	sd
B Female	mean
	sd
B Male	mean
	sd

Thanks a lot! It's a good way but within each group, my data has much more variables than just two, like female and male.
This table is a little bit hard to find the group combination. I'm still thinking about putting one group as the row, and the other as the column.
Do you have any ideas about it?

Thanks for your help!

[Fredrick Achar]

One solution is to create a composite variable first:

egen group = group(city gender), label

Hi Team
, I a beginner with STATA and i have a dataset with 1.3million observations, spread over years from 2011 to 2020. i wanted to summarize two variables, one (client) by another (year). I tried this command egen group = group( Client year), label and i haven't seen any output, have been waiting since, could the problem be my data volume or wrong command?