* These analyses compare football players (whose first year was 1986-1995 (minus 1987) and who were age 21-25 then) to * American men in general (from NHIS) who were 21 to 25 in 1986 through 1995 (minus 1987). * The code calls two data files: (1) nhis.dta is an extract of NHIS data from IPUMS.org. (2) football.dta contains information about * players whose first year was 1986-1995 (minus 1987) and who were age 21-25 then. The latter is based on our merging and * reconciliation of information from https://www.profootballarchives.com/ and https://www.pro-football-reference.com/ clear set more off ****************************************************************************************************************************************** ****************************************************************************************************************************************** ****************************************************************************************************************************************** *** *** Build NHIS part of analytic file *** ****************************************************************************************************************************************** ****************************************************************************************************************************************** ****************************************************************************************************************************************** ********************************************************************* *** File "nhis.dta" begins with 1986-2014 NHIS males who were 21 to 25 ********************************************************************* use nhis.dta, clear ********************************************************************* *** Construct Mortality Measures ********************************************************************* gen died=. replace died=0 if mortelig==1 replace died=1 if mortelig==1 & mortstat==1 gen survived=. replace survived=2019-year if died==0 replace survived=mortdody-year if died==1 gen yod=. replace yod=mortdody if died==1 * This measure only makes sense for years 1986 through 1995 gen surv10=. replace surv10=1 if died==0 replace surv10=1 if died==1 & survived>=10 & survived<=99 replace surv10=0 if died==1 & survived>=0 & survived<=9 * This measure only makes sense for years 1986 through 1995 gen surv25=. replace surv25=1 if died==0 replace surv25=1 if died==1 & survived>=25 & survived<=99 replace surv25=0 if died==1 & survived>=0 & survived<=24 ********************************************************************* *** Construct measures used to limit the sample to men who more *** closely resemble pro football players ********************************************************************* *** Race gen white=. gen black=. replace white=0 if racea>=100 & racea<=990 replace black=0 if racea>=100 & racea<=990 replace white=1 if racea==100 replace black=1 if racea==200 *** Education gen college=0 replace college=1 if educrec1>=14 & educrec1<=16 *** Employment Status gen employed=0 replace employed=1 if empstat>=110 & empstat<=122 *** Economic circumstances gen inc25k=. replace inc25k=0 if incfam8296>= 1 & incfam8296<=21 replace inc25k=1 if incfam8296>=22 & incfam8296<=27 lab var inc25k "Family earns $25k or more" gen poor=1 replace poor=0 if pooryn==1 *** Health gen healthy=0 replace healthy=1 if health==1 | health==2 *** Days in Bed w Disability or Illness; 3 or More? gen disabled=0 replace disabled=1 if bedayr>=3 & bedayr<=999 ********************************************************************* *** Keep analytic variables required for NHIS people ********************************************************************* keep year age died yod mortwt survived surv10 surv25 college employed poor healthy disabled inc25k white black gen football=0 gen games=0 save nhis_useme.dta, replace clear ****************************************************************************************************************************************** ****************************************************************************************************************************************** ****************************************************************************************************************************************** *** *** Build football player part of analytic file *** ****************************************************************************************************************************************** ****************************************************************************************************************************************** ****************************************************************************************************************************************** ********************************************************************* *** File "football.dta" contains players who played their first *** seasons between 1986 and 2014 and who were between 21 and 25 *** in that season ********************************************************************* use football.dta, clear gen age= Pro_Yr1_PFA- Birth_Year_PFA rename Pro_Yr1_PFA year rename NFL_Games_PFA games ********************************************************************* *** Construct Mortality Measures ********************************************************************* * This is really "Died by 2019" to be consistent with NHIS gen died=0 replace died=1 if Death_Year_PFA != . replace died=0 if Death_Year_PFA>=2020 & Death_Year_PFA<=2022 gen yod=. replace yod=Death_Year_PFA if died==1 gen survived=. replace survived=2019-year if died==0 replace survived=Death_Year_PFA-year if died==1 * This measure only makes sense for years 1986 through 1995 gen surv10=. replace surv10=1 if died==0 replace surv10=1 if died==1 & survived>=10 & survived<=99 replace surv10=0 if died==1 & survived>=0 & survived<=9 * This measure only makes sense for years 1986 through 1995 gen surv25=. replace surv25=1 if died==0 replace surv25=1 if died==1 & survived>=25 & survived<=99 replace surv25=0 if died==1 & survived>=0 & survived<=24 ********************************************************************* *** Construct position categories ********************************************************************* gen position5=6 replace position5=5 if Position_PFA_Clean=="dback" | Position_PFA_Clean=="dsecond" replace position5=4 if Position_PFA_Clean=="dline" replace position5=3 if Position_PFA_Clean=="ocatch" | Position_PFA_Clean=="oqbrb" replace position5=2 if Position_PFA_Clean=="oline" lab def pos5 2 "Off Line" 3 "Off Skill" 4 "Def Line" 5 "Def Other" 6 "Other" lab val position5 pos5 gen position3=4 replace position3=3 if Position_PFA_Clean=="dback" | Position_PFA_Clean=="dsecond" replace position3=2 if Position_PFA_Clean=="dline" replace position3=3 if Position_PFA_Clean=="ocatch" | Position_PFA_Clean=="oqbrb" replace position3=2 if Position_PFA_Clean=="oline" lab def pos3 2 "Linemen" 3 "Off/Def Skill" 4 "Other" lab val position3 pos3 gen position2=3 replace position2=2 if Position_PFA_Clean=="dline" replace position2=2 if Position_PFA_Clean=="oline" lab def pos2 2 "Linemen" 3 "All Other" lab val position2 pos2 ******************************************************************** *** Keep analytic variables required for football players ********************************************************************* keep year age died yod survived surv10 surv25 position2 games Scraped_ID_PFA Full_Name_PFA Birth_Month_PFA Birth_Day_PFA Birth_Year_PFA Birth_City_PFA Birth_State_PFA Death_Month_PFA Death_Day_PFA Death_Year_PFA HS_Name_PFA HS_Place_PFA College_Name_PFA gen football=1 save football_useme.dta, replace clear ****************************************************************************************************************************************** ****************************************************************************************************************************************** ****************************************************************************************************************************************** *** *** Merge Files *** ****************************************************************************************************************************************** ****************************************************************************************************************************************** ****************************************************************************************************************************************** clear use nhis_useme.dta, clear append using football_useme.dta ********************************************************************* *** Restrict the Sample: *** ... to 1986 through 1995 but *dropping* 1987 because of replacement players *** ... to NHIS cases that don't have zero weight values ******************************************************************** keep if year>=1986 & year<=1995 drop if year==1987 keep if mortwt != 0 ********************************************************************* *** Recode college, employed, income, and healthy for football players ********************************************************************* replace college=1 if football==1 replace employed=1 if football==1 replace poor=0 if football==1 replace inc25k=1 if football==1 replace healthy=1 if football==1 replace disabled=0 if football==1 ********************************************************************* *** Recode position for non-players ********************************************************************* replace position2=1 if football==0 ********************************************************************* *** For logit models: Create p weights for NHIS people and set the *** weight to 1 for football players ********************************************************************* gen pwt=1 replace pwt=1/mortwt if football==0 egen wtavg=mean(pwt) if football==0 replace pwt=pwt/wtavg if football==0 drop wtavg bysort football: summ pwt *** For forcing NHIS sample of selective men to be 50% white gen pwt50wt=pwt replace pwt50wt=pwt50wt*3.66382 if white==0 & football==0 replace pwt50wt=pwt50wt*0.57902 if white==1 & football==0 *** For forcing NHIS sample of selective men to be 33% white gen pwt33wt=pwt replace pwt33wt=pwt33wt*4.88566 if white==0 & football==0 replace pwt33wt=pwt33wt*0.38592 if white==1 & football==0 ********************************************************************* *** For event history models: Create iweights for NHIS people and *** set the weight to 1 for football players ********************************************************************* egen wtavg=mean(mortwt) gen ewt=1 replace ewt=mortwt/wtavg if football==0 *** For forcing NHIS sample of selective men to be 50% white gen ewt50wt=ewt replace ewt50wt=ewt50wt*3.66382 if white==0 & football==0 replace ewt50wt=ewt50wt*0.57902 if white==1 & football==0 *** For forcing NHIS sample of selective men to be 33% white gen ewt33wt=ewt replace ewt33wt=ewt33wt*4.88566 if white==0 & football==0 replace ewt33wt=ewt33wt*0.38592 if white==1 & football==0 ****************************************************************************************************************************************** ****************************************************************************************************************************************** ****************************************************************************************************************************************** *** *** Do the Analyses *** ****************************************************************************************************************************************** ****************************************************************************************************************************************** ****************************************************************************************************************************************** ********************************************************************* *** Descriptives ********************************************************************* log using descriptives.log, replace log on tab football surv25 [aweight=ewt] tab position2 surv25 [aweight=ewt] tab died football [aweight=ewt] tab died position2 [aweight=ewt] tab college if football==0 [aweight=ewt] tab healthy if football==0 [aweight=ewt] tab inc25k if football==0 [aweight=ewt] tab employed if football==0 [aweight=ewt] tab disabled if football==0 [aweight=ewt] tab died position2 if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 [aweight=ewt] tab position2 surv25 if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 [aweight=ewt] log off log close ********************************************************************* *** Model Mortality ********************************************************************* gen lineman=0 gen otherplayer=0 replace lineman=1 if position2==2 replace otherplayer=1 if position2==3 log using models.log, replace log on * LOGIT of Surviving Another 25 years --- All football players vs men in general relogit surv25 football year age [aweight=pwt] setx year mean age mean football 0 relogitq setx year mean age mean football 1 relogitq * LOGIT of Surviving Another 25 years --- All football players vs men with 3+ years of college, jobs, good health, no injury/illness, and median+ family income relogit surv25 football year age if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 [aweight=pwt] setx year mean age mean football 0 relogitq setx year mean age mean football 1 relogitq * LOGIT of Surviving Another 25 years --- All football players vs men with 3+ years of college, jobs, good health, no injury/illness, and median+ family income * Adjusting the weights to force the NHIS men to be 50% White relogit surv25 football year age if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 [aweight=pwt50wt] setx year mean age mean football 0 relogitq setx year mean age mean football 1 relogitq * LOGIT of Surviving Another 25 years --- All football players vs men with 3+ years of college, jobs, good health, no injury/illness, and median+ family income * Adjusting the weights to force the NHIS men to be 33% White relogit surv25 football year age if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 [aweight=pwt33wt] setx year mean age mean football 0 relogitq setx year mean age mean football 1 relogitq * LOGIT of Surviving Another 25 years --- Football players by position vs men in general relogit surv25 lineman otherplayer year age [aweight=pwt] setx year mean age mean lineman 0 otherplayer 0 relogitq setx year mean age mean lineman 1 otherplayer 0 relogitq setx year mean age mean lineman 0 otherplayer 1 relogitq * LOGIT of Surviving Another 25 years --- Football players by position vs men with 3+ years of college, jobs, good health, no injury/illness, and median+ family income relogit surv25 lineman otherplayer year age if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 [aweight=pwt] setx year mean age mean lineman 0 otherplayer 0 relogitq setx year mean age mean lineman 1 otherplayer 0 relogitq setx year mean age mean lineman 0 otherplayer 1 relogitq * LOGIT of Surviving Another 25 years --- Football players by position vs men with 3+ years of college, jobs, good health, no injury/illness, and median+ family income * Adjusting the weights to force the NHIS men to be 50% White relogit surv25 lineman otherplayer year age if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 [aweight=pwt50wt] setx year mean age mean lineman 0 otherplayer 0 relogitq setx year mean age mean lineman 1 otherplayer 0 relogitq setx year mean age mean lineman 0 otherplayer 1 relogitq * LOGIT of Surviving Another 25 years --- Football players by position vs men with 3+ years of college, jobs, good health, no injury/illness, and median+ family income * Adjusting the weights to force the NHIS men to be 33% White relogit surv25 lineman otherplayer year age if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 [aweight=pwt33wt] setx year mean age mean lineman 0 otherplayer 0 relogitq setx year mean age mean lineman 1 otherplayer 0 relogitq setx year mean age mean lineman 0 otherplayer 1 relogitq * Set up Event History Models replace survived=survived+1 stset survived [iweight=ewt], failure(died==1) * EVENT HISTORY --- All football players vs men in general stcox football year age * stcurve, survival at1(football=0) at2(football=1) connect(direct) saving(model-1.tif,replace) * EVENT HISTORY --- All football players vs men with 3+ years of college, jobs, good health, no injury/illness, and median+ family income stcox football year age if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 * EVENT HISTORY --- All football players vs men with 3+ years of college, jobs, good health, no injury/illness, and median+ family income * Adjusting the weights to force the NHIS men to be 50% White stset survived [iweight=ewt50wt], failure(died) stcox football year age if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 * EVENT HISTORY --- All football players vs men with 3+ years of college, jobs, good health, no injury/illness, and median+ family income * Adjusting the weights to force the NHIS men to be 33% White stset survived [iweight=ewt33wt], failure(died) stcox football year age if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 * EVENT HISTORY --- Football players by position vs men in general stset survived [iweight=ewt], failure(died) stcox i.position2 year age * EVENT HISTORY --- Football players by position vs men with 3+ years of college, jobs, good health, no injury/illness, and median+ family income stcox i.position2 year age if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 * EVENT HISTORY --- Football players by position vs men with 3+ years of college, jobs, good health, no injury/illness, and median+ family income * Adjusting the weights to force the NHIS men to be 50% White stset survived [iweight=ewt50wt], failure(died) stcox i.position2 year age if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 * EVENT HISTORY --- Football players by position vs men with 3+ years of college, jobs, good health, no injury/illness, and median+ family income * Adjusting the weights to force the NHIS men to be 33% White stset survived [iweight=ewt33wt], failure(died) stcox i.position2 year age if college==1 & healthy==1 & inc25k==1 & employed==1 & disabled==0 log off log close