Recoding Variable Assistance

[Davia Downey]

Hello Statalisters,

I need to recode a variable to indicate the number of days that elapsed to the adoption of a particular policy. Currently, the variable first_public_healthpost is coded 0 or 1, with 1 indicating when the policy was adopted. I have a variable daily_cases_by_date that's tracking the passage of time. I believe to recode this variable I need to do something like: gen daystopassage= daily_cases_by_date-first_public_healthpost==1 but when I do this, all I get is a bunch of zeros in the new variable when I expect that starts at 1 to the number of elapsed days to passage. I know this should be fairly simple but I'm stuck. Any assistance would be greatly appreciated. Data sample included below:

* Example generated by -dataex-. For more info, type help dataex
clear
input float daily_cases_by_date int day byte(fips2 first_public_healthpost mandate_order_dates) str14 name
22085 150 1 0 0 "Alabama"
22086 151 1 0 0 "Alabama"
22087 152 1 0 0 "Alabama"
22088 153 1 0 0 "Alabama"
22089 154 1 0 0 "Alabama"
22090 155 1 0 0 "Alabama"
22091 156 1 0 0 "Alabama"
22092 157 1 0 0 "Alabama"
22093 158 1 0 0 "Alabama"
22094 159 1 0 0 "Alabama"
22095 160 1 0 0 "Alabama"
22096 161 1 0 0 "Alabama"
22097 162 1 0 0 "Alabama"
22098 163 1 0 0 "Alabama"
22099 164 1 0 0 "Alabama"
22100 165 1 0 0 "Alabama"
22101 166 1 0 0 "Alabama"
22102 167 1 0 0 "Alabama"
22103 168 1 0 0 "Alabama"
22104 169 1 0 0 "Alabama"
22105 170 1 0 0 "Alabama"
22106 171 1 0 0 "Alabama"
22107 172 1 0 0 "Alabama"
22108 173 1 0 0 "Alabama"
22109 174 1 0 0 "Alabama"
22110 175 1 0 0 "Alabama"
22111 176 1 0 0 "Alabama"
22112 177 1 1 1 "Alabama"
22113 178 1 1 1 "Alabama"
22114 179 1 1 1 "Alabama"
22115 180 1 1 1 "Alabama"
22116 181 1 1 1 "Alabama"
22117 182 1 1 1 "Alabama"
22118 183 1 1 1 "Alabama"
22119 184 1 1 1 "Alabama"
22120 185 1 1 1 "Alabama"
22121 186 1 1 1 "Alabama"
22122 187 1 1 1 "Alabama"
22123 188 1 1 1 "Alabama"
22124 189 1 1 1 "Alabama"
22125 190 1 1 1 "Alabama"
22126 191 1 1 1 "Alabama"
22127 192 1 1 1 "Alabama"
22128 193 1 1 1 "Alabama"
22129 194 1 1 1 "Alabama"
22130 195 1 1 1 "Alabama"
22131 196 1 1 1 "Alabama"
22132 197 1 1 1 "Alabama"
22133 198 1 1 1 "Alabama"
22134 199 1 1 1 "Alabama"
22135 200 1 1 1 "Alabama"
end
format %td daily_cases_by_date

So basically all I need to do is create a new variable that counts from the daily_cases_by_date up to the date of the first_public_healthpost to run a Cox-Hazard model of this data. I'm interested in the differences in time to adoption amongst the states.

[Nick Cox]

That code is legal but some distance from what you need, which sounds like

egen datefirst = min(cond(first_public_healthpost == 1, daily_cases_by_date, .)), by(name)

gen wanted = daily_cases_by_date - datefirst

or its negation.

See also https://www.stata-journal.com/articl...article=dm0055 Section 9.

[Davia Downey]

It is specific by state. Thanks for the assistance!

[Nick Cox]

Indeed. I spotted that on the second reading.

[Davia Downey]

Nick Cox I have one follow up question. The date range in my data goes from January 22, 2020 to June 2, 2021 so using your code above I know have some negative integers that need to be adjusted since there are 498 days in the data. Do I need to incorporate this somehow? I assume the answer is yes, but I'm not sure where in your code I would input this would it be something like

gen daystopassage= (daily_cases_by_date-datefirst)-498

[Davia Downey]

As a follow up. The new variable time_to_passage is correct but the sign is negative (although I guess in terms of interpretation this is the correct way to illustrate). See below for data snippet. I guess I'm wondering if this variable is correctly censored for analysis.

* Example generated by -dataex-. For more info, type help dataex
clear
input float(daily_cases_by_date datefirst) int day str14 statename float time_to_passage
22035 22112 100 "Alabama" -77
22036 22112 101 "Alabama" -76
22037 22112 102 "Alabama" -75
22038 22112 103 "Alabama" -74
22039 22112 104 "Alabama" -73
22040 22112 105 "Alabama" -72
22041 22112 106 "Alabama" -71
22042 22112 107 "Alabama" -70
22043 22112 108 "Alabama" -69
22044 22112 109 "Alabama" -68
22045 22112 110 "Alabama" -67
22046 22112 111 "Alabama" -66
22047 22112 112 "Alabama" -65
22048 22112 113 "Alabama" -64
22049 22112 114 "Alabama" -63
22050 22112 115 "Alabama" -62
22051 22112 116 "Alabama" -61
22052 22112 117 "Alabama" -60
22053 22112 118 "Alabama" -59
22054 22112 119 "Alabama" -58
22055 22112 120 "Alabama" -57
22056 22112 121 "Alabama" -56
22057 22112 122 "Alabama" -55
22058 22112 123 "Alabama" -54
22059 22112 124 "Alabama" -53
22060 22112 125 "Alabama" -52
22061 22112 126 "Alabama" -51
22062 22112 127 "Alabama" -50
22063 22112 128 "Alabama" -49
22064 22112 129 "Alabama" -48
22065 22112 130 "Alabama" -47
22066 22112 131 "Alabama" -46
22067 22112 132 "Alabama" -45
22068 22112 133 "Alabama" -44
22069 22112 134 "Alabama" -43
22070 22112 135 "Alabama" -42
22071 22112 136 "Alabama" -41
22072 22112 137 "Alabama" -40
22073 22112 138 "Alabama" -39
22074 22112 139 "Alabama" -38
22075 22112 140 "Alabama" -37
22076 22112 141 "Alabama" -36
22077 22112 142 "Alabama" -35
22078 22112 143 "Alabama" -34
22079 22112 144 "Alabama" -33
22080 22112 145 "Alabama" -32
22081 22112 146 "Alabama" -31
22082 22112 147 "Alabama" -30
22083 22112 148 "Alabama" -29
22084 22112 149 "Alabama" -28
22085 22112 150 "Alabama" -27
22086 22112 151 "Alabama" -26
22087 22112 152 "Alabama" -25
22088 22112 153 "Alabama" -24
22089 22112 154 "Alabama" -23
22090 22112 155 "Alabama" -22
22091 22112 156 "Alabama" -21
22092 22112 157 "Alabama" -20
22093 22112 158 "Alabama" -19
22094 22112 159 "Alabama" -18
22095 22112 160 "Alabama" -17
22096 22112 161 "Alabama" -16
22097 22112 162 "Alabama" -15
22098 22112 163 "Alabama" -14
22099 22112 164 "Alabama" -13
22100 22112 165 "Alabama" -12
22101 22112 166 "Alabama" -11
22102 22112 167 "Alabama" -10
22103 22112 168 "Alabama"  -9
22104 22112 169 "Alabama"  -8
22105 22112 170 "Alabama"  -7
22106 22112 171 "Alabama"  -6
22107 22112 172 "Alabama"  -5
22108 22112 173 "Alabama"  -4
22109 22112 174 "Alabama"  -3
22110 22112 175 "Alabama"  -2
end
format %td daily_cases_by_date

[Nick Cox]

Posts are crossing here. An edit to #2 said "or its negation", which was hedging bets. Perhaps what you want

bysort name (dailycases) : gen wanted = daily_cases - daily_cases[1] + 1 if daily_cases < datefirst

[Davia Downey]

Yep. That's exactly what I wanted. Now the variable goes to missing "." once the policy was passed. Thank you again.

[Nick Cox]

So, this should have been simpler:

 
 bysort name (daily_cases) : gen wanted = daily_cases - daily_cases[1] + 1 if first_public_healthpost == 0