Creating a year variable

[Amal Khanolkar]

Hi All

I'm creating a set of 5 date variables to calculate age of my study participants. The 5 variables corresponding to the years 1982, 1989, 1999, 2009 & 2015. Day & month are missing, so these five variables will only have the year.

For the first one, in 1982:

gen year36 = 1982
br ID year36
gen date36=year(year36)

format date36 %ty

The variable date36 shows as 1965, so I've missed a step in between somewhere....

Could someone help me?

Thanks
/Amal

[Nick Cox]

year36 is fine as a variable including years. The problem is that

year()

is a function with the purpose of extracting a year from a daily date. See the fine help:

year(e_d)
Description: the numeric year corresponding to date e_d
Domain e_d: %td dates 01jan0100 to 31dec9999 (integers -679,350 to 2,936,549)
Range: integers 0100 to 9999 (but probably 1800 to 2100)

1985 as a daily date was a day in 1965, which explains your result.

. di %td 1985
08jun1965

This isn't a step missed. it is an unneeded step, as so far as we can tell year36 is fine for your purposes.

[Amal Khanolkar]

Hi Nick

Thanks for the quick reply.

But don't I need to inform Stata that 1982 is a 'year' via:

 
 gen date36=year(year36)

The variable indicating birthday (which will be used with the above year36 to calculate age) is in date format:

gen day = 15
gen month = 3
gen year = 1946

gen birthdate = mdy(month, day, year)
format birthdate %d

Thanks
/Amal

[Nick Cox]

No, no, no.

year() does what it does and nothing else. It is not only not needed for what you want: it will produce nonsense if fed a literal year value.

You can if you like format a variable containing years to have a yearly display format but that's empty: nothing will change.

What you are doing with the birthdays looks fine to me. But strictly there is no such thing as "date format". There are daily date formats, monthly date formats, and several more.

[Amal Khanolkar]

Okay - I tried the above (including calculating the age variable), but something is going wrong somewhere:

gen day = 15
gen month = 3
gen year = 1946
br NSHD_ID day month year

gen birthdate = mdy(month, day, year)
format birthdate %d
br NSHD_ID day month year birthdate

gen year36 = 1982
gen year43 = 1989
gen year53 = 1999
gen year63 = 2009
gen year69 = 2015

br ID year36 year43 year53 year63 year69

gen age36 = year36-birthdate
gen age43 = year43-birthdate
gen age53 = year53-birthdate
gen age63 = year63-birthdate
gen age69 = year69-birthdate

The age 36 variable has values = 7022 which isn't right. Or do I need to follow-up with formatting the age variable?

/Amal

[Nick Cox]

Formatting is not the issue here. The issue is that you are subtracting a daily date from a yearly date. They are in quite different units.

I can't see that you need those year variables at all as they are just holding constants -- but in themselves they do no harm.

The calculation seems circular. Someone was born in 1946, so was 36 in 1982, and so forth. So, what do you want to do?

Note that something like

gen age = (mdy(12, 31, year) - birthdate) / 365.25

gives an age on the stated date. 365.25 seems to be acceptable precision for e.g. epidemiology.

[Amal Khanolkar]

Hi Nick

Just following-up from the above - I complicated things. My dataset is longitudinal in the long format, and each individual has 5 rows of data. I just want to create a variable called year which can take on the same 5 values: 1982, 1989, 1999, 2009 & 2015 for subjects and is understood by Stata to be in calendar years. I would I do this:

id year
1 1982
1 1989
1 1999
1 2009
1 2015
2 1982
2 1989
2 1999
2 2009
2 2015
3 1982
3 1989
3 1999
3 2009
3 2015

Many Thanks
/Amal