Mark B CCES 2016 Mail Intention Validation

cces_absentee_val.R

require(rio)
require(sqldf)

cces = import('~/Downloads/CCES16_Common_OUTPUT_Feb2018_VV.dta')
.cces = (cces[,c('CL_voterstatus','CC16_364','CC16_403','CL_E2016GVM','commonweight_vv','V101','CL_state')])

.cces = .cces[.cces$CL_voterstatus %in% c('active','inactive'),]
.cces$VOTE_LIKELIHOOD = car::recode(var = as.numeric(.cces$CC16_364),
                        recodes = "1='Yes'; 2='Probably' ; 3='Already voted' ; 4 = 'No'; else = NA")
.cces$VOTE_METHOD_SELFREPORT = car::recode(var = as.numeric(.cces$CC16_403),
                                    recodes = "1='EDay'; 2='Early' ; 3='Mail' ; 4 = 'DK'; else = NA")

state= unique(c(sqldf("select count(*), state from (select count(*) as n,
      CL_E2016GVM as vm,
      CL_state as state
      from \".cces\"
      where CL_E2016GVM is not null
      group by 2,3) a
      where vm in ('earlyVote','absentee') and n>10
      group by 2
      having count(*)>1")$state,
         sqldf("select count(*) as n,
      CL_E2016GVM as vm,
               CL_state as state
               from \".cces\"
               where CL_E2016GVM ='mail'
               group by 2,3
               order by 3,2")$state))

.cces = .cces[.cces$CL_state %in% state,]

vl = sqldf("select sum(commonweight_vv) as n,
              sum(case when CL_E2016GVM!='' then commonweight_vv else 0 end)/sum(commonweight_vv) as voted,
              sum(case when CL_E2016GVM in ('absentee','mail') then commonweight_vv else 0 end)/sum(commonweight_vv) as absentee,
              sum(case when CL_E2016GVM in ('absentee','mail') then commonweight_vv when CL_E2016GVM!='' then 0 else null end)/sum(case when CL_E2016GVM!='' then commonweight_vv else 0 end) as absentee_onlyvoters,

VOTE_LIKELIHOOD,
case when CL_state in ('UT','CO','OR','WA') then 'All Mail' else 'Not' end as regime
      from \".cces\"
    where VOTE_LIKELIHOOD is not null
      group by 5,6")
vl$p = vl$n/sum(vl$n)

vm = sqldf("select sum(commonweight_vv) as n,
              sum(case when CL_E2016GVM!='' then commonweight_vv else 0 end)/sum(commonweight_vv) as voted,
              sum(case when CL_E2016GVM in ('absentee','mail') then commonweight_vv else 0 end)/sum(commonweight_vv) as absentee,
              sum(case when CL_E2016GVM in ('absentee','mail') then commonweight_vv when CL_E2016GVM!='' then 0 else null end)/sum(case when CL_E2016GVM!='' then commonweight_vv else 0 end) as absentee_onlyvoters,
           VOTE_METHOD_SELFREPORT,
case when CL_state in ('UT','CO','OR','WA') then 'All Mail' else 'Not' end as regime
           from \".cces\"
           where VOTE_METHOD_SELFREPORT is not null
           group by 5,6")
vm$p = vm$n/sum(vm$n)

> vm
            n     voted   absentee absentee_onlyvoters VOTE_METHOD_SELFREPORT   regime            p
1    5.229242 0.3952201 0.39522015          1.00000000                     DK All Mail 0.0003777992
2   45.752792 0.0421794 0.01109287          0.26299260                     DK      Not 0.0033055206
3  126.990018 0.8248018 0.30044121          0.36425865                   EDay All Mail 0.0091746995
4 3780.408532 0.9076089 0.04727233          0.05208447                   EDay      Not 0.2731247145
5  134.192384 0.9815030 0.67605209          0.68879271                  Early All Mail 0.0096950518
6 3977.242963 0.9755753 0.04896948          0.05019549                  Early      Not 0.2873454918
7 2062.131392 0.9687823 0.91963775          0.94927181                   Mail All Mail 0.1489836464
8 3709.379846 0.9582471 0.83254813          0.86882402                   Mail      Not 0.2679930761

> vl
            n     voted  absentee absentee_onlyvoters VOTE_LIKELIHOOD   regime           p
1    81.02850 0.9784796 0.9098736           0.9298851   Already voted All Mail 0.005104347
2   470.10502 0.9283224 0.6990688           0.7530454   Already voted      Not 0.029614013
3    32.06346 0.6143469 0.4433207           0.7216130              No All Mail 0.002019820
4   300.67241 0.4965004 0.1471620           0.2963986              No      Not 0.018940697
5   109.36894 0.7583679 0.6491968           0.8560446        Probably All Mail 0.006889638
6   650.99839 0.6612443 0.1307119           0.1976757        Probably      Not 0.041009294
7  2346.43909 0.9508615 0.8661768           0.9109391             Yes All Mail 0.147812670
8 11883.73528 0.9296883 0.2846754           0.3062052             Yes      Not 0.748609521