Time variable in inrange?

[River Huang]

Dear All,

I have a portion of my data set as shown below:

* Example generated by -dataex-. For more info, type help dataex
clear
input float ym double rf
-408              .002951
-407              .002768
-406              .002778
-405              .003072
-404              .000342
-403 .0034590000000000003
-402  .002199999988079071
-401 .0024999999441206455
-400  .002300000051036477
-399 .0031999999191612005
-398  .003100000089034438
-397   .00279999990016222
-396 .0024999999441206455
-395 .0026000000070780516
-394  .003000000026077032
-393 .0024999999441206455
-392  .003000000026077032
-391 .0026000000070780516
-390  .003000000026077032
-389   .00279999990016222
-388  .002099999925121665
-387 .0024999999441206455
-386  .002099999925121665
-385  .002199999988079071
-384 .0024999999441206455
-383 .0033000002149492502
-382  .002899999963119626
-381  .002199999988079071
-380 .0031999999191612005
-379  .003100000089034438
-378 .0031999999191612005
-377 .0031999999191612005
-376 .0027000000700354576
-375  .004100000020116568
-374  .003800000064074993
-373 .0005999999702908099
-372 .0034000000450760126
-371 .0036000001709908247
-370 .0034000000450760126
-369 .0036000001709908247
-368  .004399999976158142
-367  .005200000014156103
-366 .0033000002149492502
-365  .004000000189989805
-364  .003499999875202775
-363  .004600000102072954
-362  .003700000001117587
-361  .003700000001117587
-360   .00139999995008111
-359  .003000000026077032
end
format %tm ym

I am trying to generate a categorical variable (cat) with values 1, 2, and 3 representing the following time ranges:

1. 1926/01 - 1988/12
2. 1926/01 - 1957/06
3. 1957/07 - 1988/12

I would appreciate any suggestions on how to achieve this.

Thank you in advance!

[Nick Cox]

Groups 2 and 3 are mutually exclusive but group 1 includes both. So there is no scope for a variable with values 1, 2, 3.

[River Huang]

Many thanks, Nick. My initial idea was to generate a variable cat. If the time falls between 1926/01 and 1988/12, then cat is assigned a value of 1; if it's between 1926/01 and 1957/06, then cat is assigned a value of 2; and if it's between 1957/07 and 1988/12, then cat is assigned a value of 3. I wonder if this is feasible?

[Nick Cox]

I still don't follow unless you are able to give examples where you know that 1 is true but you can't be clear which of 2 or 3 applies. Your data example includes what looks like a Stata monthly date, so there is no ambiguity or uncertainty about the date.

[River Huang]

Dear Nick, Now I think your concern is correct. The way I do it is:

gen date = ym + 408 + 1  // 

* 1926/01~1988/12
gen sample1 = (date >= 1 & date <= 756)  // 1926/01 到 1988/12

* 1926/01~1957/06
gen sample2 = (date >= 1 & date <= 378)  // 1926/01 到 1957/06

* 1957/07~1988/12
gen sample3 = (date > 378 & date <= 756)  // 1957/07 到 1988/12

* 1926/01~2023/12
gen sample4 = (date >= 1 & date <= 1176)  // 1926/01 到 2023/12

* 1989/01~2023/12
gen sample5 = (date > 756 & date <= 1176)  // 1989/01 到 2023/12

label var sample1 "1926/01-1988/12"
label var sample2 "1926/01-1957/06"
label var sample3 "1957/07-1988/12"
label var sample4 "1926/01-2023/12"
label var sample5 "1989/01-2023/12"

foreach v in sample1 sample2 sample3 sample4 sample5 {
    preserve
    keep if `v' == 1
    display "`v'"
    // value-weighted
    sum vwerm 
    display (r(mean)/r(sd))^2
    //equal-weighted
    sum ewerm 
    display (r(mean)/r(sd))^2
    restore
}