Generate a continuous variable at the*firm level

[Mo Hos]

I would like to generate a continuous variable at the firm level. For the sample below, I would like to generate a new variable that illustrates the percentage of employees that are "female, older than 22, have a master degree " to employees that are "Older than 22 and have master degree" for each firm-year observation.

companyid	year	employeeid	Female, older than 22 and have master degree	Female, older than 22 and have a master degree
293475073	2005	10	1	1
293475073	2005	11	0	0
293475073	2005	5	1	1
982726482	2005	12	0	0
982726482	2005	11	0	0
982726482	2005	5	1	1
23452345234	2006	10	1	1
23452345234	2006	6	0	0
23452345234	2006	7.00E+00	1	0
18938422	2006	8.00E+00	1	0
18938422	2006	6	0	0
18938422	2006	1.50E+01	0	0
18938422	2006	1.60E+01	1	0
980232133	2007	1.70E+01	1	1
980232133	2007	1.80E+01	0	0
3456546	2007	1.90E+01	0	0
3456546	2007	1.70E+01	1	1

[Jeph Herrin]

You have two columns that are the same, which must be a mistake. But let's suppose you have flag1=1 if "female, older than 22, Masters" and flag2 = "older than 22, masters". Then this should do it:

. bysort companyid year : egen percent=mean(cond(flag2,flag1,.))

hth,
Jeph

[Mo Hos]

Sorry for the mistake. This is the correct table.

companyid	year	employeeid	Older than 22 and have a master degree (flag2)	Female, older than 22 and have a master degree (flag1)
293475073	2005	10	1	1
293475073	2005	11	0	0
293475073	2005	5	1	1
982726482	2005	12	0	0
982726482	2005	11	0	0
982726482	2005	5	1	1
23452345234	2006	10	1	1
23452345234	2006	6	0	0
23452345234	2006	7.00E+00	1	0
18938422	2006	8.00E+00	1	0
18938422	2006	6	0	0
18938422	2006	1.50E+01	0	0
18938422	2006	1.60E+01	1	0
980232133	2007	1.70E+01	1	1
980232133	2007	1.80E+01	0	0
3456546	2007	1.90E+01	0	0
3456546	2007	1.70E+01	1	1

Is this code going to generate the sum of flag1 divided by the sum of flag2 for each firm-year observation? because that is the one I need.

bysort companyid year : egen percent=mean(cond(flag2,flag1,.))