C ea i e Commons Non Comme cial CC BY-NC: This a icle is dis ibu ed unde he e ms o he C ea i e Commons
A ibu ion-NonComme cial 4.0 License (h p://www.c ea i ecommons.o g/licenses/by-nc/4.0/) which pe mi s non-comme cial
use, ep oduc ion and dis ibu ion o he wo k wi hou u he pe mission p o ided he o iginal wo k is a ibu ed as speci ied on he SAGE and
Open Access pages (h ps://us.sagepub.com/en-us/nam/open-access-a -sage).
h ps://doi.o g/10.1177/2053168017738410
Resea ch and Poli ics
Oc obe -Decembe 2017: 1 –8
© The Au ho (s) 2017
Rep in s and pe missions:
sagepub.co.uk/jou nalsPe missions.na
DOI: 10.1177/2053168017738410
jou nals.sagepub.com/home/ ap
Low a es o pa icipa ion in elec ions a e o en desc ibed
as a ch onic ailmen o he Ame ican body poli ic.1 The e
a e, none heless, a a ie y o p oposed in e en ions o
inc ease pa icipa ion in Ame ican elec ions. We expe i-
men ally assess he e ec o an inc ease in poli ical knowl-
edge on u nou among young o e s in a local elec ion. In
ad ance o he i s ound o he 2015 mayo al elec ion in
Chicago we mailed a guide lis ing endo semen s and policy
posi ions o each o he i e candida es o 25,000 egis-
an s be ween he ages o 18 and 30. A e he elec ion we
loca ed as many o hese ea ed egis an s and he p e-
speci ied con ol g oup as possible in o e egis a ion lis s
o measu e he e ec o he ea men o ecei ing a nonpa -
isan o e guide. We ind, i s , ha his di ec mail ea -
men signi ican ly inc eased u nou by abou 0.9 pe cen age
poin s ( he s anda d e o a ound his ea men e ec is
0.003). When we in e ac he maile ea men wi h median
income a he census ac le el, howe e , we ind he ea -
men had a signi ican e ec on u nou only in census
ac s wi h abo e-median household income.
We begin wi h some gene al obse a ions abou u nou in
he Uni ed S a es, he po en ial o o ing e o ms o inc ease
u nou , he con ounding e ec s o c oss-cu ing in o ma ion
on poli ical ac i i y, and obse e how his s udy con ibu es o
he la gely expe imen al li e a u e o o e mobiliza ion s ud-
ies. We hen desc ibe he expe imen al design and ou ind-
ings be o e making a ew concluding ema ks.
Tu nou has ma kedly declined since he heyday o pa -
icipa ion in he 1960s (Wa enbe g, 2002; Teixei a, 1987).
P esiden ial aces gene ally a ac he mos a en ion—and
he highes u nou —while local elec ions end o ea u e
compa a i ely low a es o pa icipa ion (Hajnal and Lewis,
2003, Holb ook and Weinschenk, 2014; Oli e , 2012).
Rega dless o elec o al con ex , u nou is low among peo-
ple below he age o 30 (Gimpel e al., 2004; High on and
Wol inge , 2001; Kau mann e al., 2008; Leighley and
Nagle , 2014; Shea and G een, 2006; Wa enbe g, 2012).
Focusing ou expe imen on young people in a local
Mobilizing he young o e: Di ec mail
o e guides in he 2015 Chicago mayo al
elec ion*
Pe e Mille 1, Rebecca Reynolds2 and Ma hew Singe 3
Abs ac
P e ious s udies ha e shown nonpa isan di ec mail in e en ions ha e a small u nou e ec . I is less clea , howe e ,
wha e ec hese in e en ions ha e wi h ega d o cogni i e mobiliza ion o young o e s. We epo he esul s o a
la ge (N=52,324) expe imen conduc ed du ing he Feb ua y 2015 mayo al elec ion in Chicago. A andom sample o
22,179 egis an s be ween 18 and 30 we e sen a o e guide con aining wo se s o o ien ing poli ical in o ma ion o
he i e candida es: a lis o hei endo semen s and hei policy posi ions on i e issues in he campaign. We ind u nou
among he ea ed g oup was 0.9 poin s highe han he con ol g oup which did no ecei e a guide (33.6% compa ed
wi h 32.7%; SE is 0.003). We ind, howe e , ha ea ed egis an s in abo e-median household income census ac s
a e signi ican ly mo e likely o o e han ea ed egis an s in lowe -income census ac s.
Keywo ds
Tu nou , Chicago, mobiliza ion expe imen
Da e ecei ed: 15 Feb ua y 2017; accep ed: 5 June 2017
1Uni e si y o Tampe e, Tampe e, Finland
2Chicago Vo es, Chicago, IL, USA
3Alliance o You h Ac ion, Po land, OR, USA
Co esponding au ho :
Pe e Mille , Uni e si y o Tampe e, Kansle in inne 1, Tampe e, 33014,
Finland
Email: [email p o ec ed]
738410RAP0010.1177/2053168017738410Resea ch & Poli icsMille e al.
esea ch-a icle20172017
Resea ch A icle
2 Resea ch and Poli ics
elec ion se es o isola e egis an s leas likely o o e in an
elec ion al eady less likely o exhibi high a es o pa ici-
pa ion. This app oach is no en i ely misguided; we know
o ing is a habi - o ming ac i i y (Coppock and G een,
2015; Ga cía Bedolla and Michelson, 2012; Ge be e al.,
2003; G een and Shacha , 2000; Plu ze , 2002) and hus
his expe imen may p o e ui ul o demons a e an in e -
en ion ha pays long- e m di idends in e ms o u ning
young o e s in o equen o e s.
The e a e a wide a ay o e o ms designed o make he
cas ing o a ballo easie . In a e iew o he e ec i eness o a
a ie y o elec o al e o ms, Be insky (2005) inds hese ins i-
u ional a enues o e o m a e limi ed. “The ue cos s o pa -
icipa ion lie no jus in he exp ession o opinion bu also in
he o ma ion o poli ical opinions. Exis ing elec o al e o ms
do no hing o minimize he cogni i e cos s associa ed wi h
o ing” (Be insky, 2005: 485, o iginal emphasis). Cogni i e
mobiliza ion, Be insky a gues, will inc ease in e es in poli-
ics, a well-known p edic o o pa icipa ion. One s udy in his
a ea (Wol inge e al., 2005) inds s a es ha p o ide p e-elec-
ion sample ballo s and in o ma ion abou polling places ha e
highe u nou , pa icula ly among young and less-educa ed
egis an s.2 Ou expe imen is designed o es he p oposi-
ion ha a nonpa isan maile ha includes in o ma ion ela ed
o endo semen s and policy posi ions o mayo al candida es
in Chicago will inc ease u nou among 18–30-yea -old egis-
e ed o e s who a e no o ing o he i s ime.
This p oposi ion is no uni e sally suppo ed in he li -
e a u e. Mu z (2002, 2006) demons a es ha encoun e ing
poli ical in o ma ion and discou se con a y o one’s p e e -
ences can educe pa icipa ion in poli ics. Shi (2016) inds
demobiliza ion—on he o de o 1 o 2 pe cen age poin s—
ollows om a mailed pos ca d exp essing a iew on same-
sex ma iage in No h Ca olina ha is con a y o he
ecipien ’s pa isan a ilia ion. Fu he mo e, e o s o co -
ec mispe cep ions based on p edisposi ions a e no always
success ul (Nyhan and Rei le , 2010). I may be he case
ha his mailed o e guide may educe u nou among he
ea ed obse a ions. This expe imen , in sho , p o ides a
con ex o shed ligh on he u ili y o objec i e in o ma ion
in spu ing g ea e poli ical pa icipa ion along he lines o
Be insky’s call o g ea e cogni i e mobiliza ion.
Di ec mail expe imen s and u nou
Random assignmen o a ea men is a pilla o he causal
empi icis app oach o quan i a i e esea ch (Samii, 2016).
Vo e mobiliza ion expe imen s a e one sub ield in which his
app oach has aken oo and cla i ied ou unde s anding o he
d i e s o u nou . A cen al inding in his esea ch agenda is
ha nonpa isan di ec mail inc eases u nou by a signi i-
can —bu small—amoun (G een and Ge be , 2015; G een
e al., 2013). The ea ly expe imen by Gosnell (1927) ound,
o example, a mailed eminde abou an upcoming Elec ion
Day inc eased u nou in he 1925 municipal elec ion in
Chicago by 9 pe cen age poin s. Mille and colleagues (1981)
ind o e s aged 21–30 we e mo e likely o o e in a local
p ima y elec ion a e ecei ing a w i en le e wi h cam-
paign b ochu es. In a mo e ecen s udy, Ge be and G een
(2000a) ound pos ca ds encou aging o ing in he 1996 gen-
e al elec ion inc eased u nou by 7 pe cen age poin s among
una ilia ed o e s and ha e no e ec on pa isans. An expe i-
men conduc ed in he 1998 gene al elec ion ound di ec mail
inc eased u nou by abou 0.6 pe cen age poin s (Ge be and
G een, 2000b). As Ga cía Bedolla and Michelson obse e,
howe e , “we know li le abou he unde lying mechanisms
ha make hese sho sociocul u al exchanges so e ec i e in
changing o e beha io ” (Ga cía Bedolla and Michelson,
2012: 11).
Expe imen al esea ch designs ha e been used o es ima e
mobiliza ion e ec s among a ge ed popula ions, like age and
acial (o e hnic) g oups. Sociocul u al cogni ion— he claim
ha an indi idual’s sociocul u al con ex has a ole in cogni-
i e ou comes—p o ides a heo e ical link be ween mobiliza-
ion expe imen s and obse ed ou comes (Ga cía Bedolla and
Michelson, 2012). Nicke son (2006) inds young o e s—
once hey a e egis e ed o o e—a e equally esponsi e o
appeals as o he o e s, bu ha i is abou h ee imes as di -
icul o con ac a young o e . The ex en o which mino i y
popula ions can be mobilized by hese appeals is a common
opic in he li e a u e, hough signi ican e ec s a e no e-
quen ly ob ained (G een, 2004; Ma land and Mu ay, 2012;
T i edi, 2005; Wong, 2005). Ga cía Bedolla and Michelson
(2009), howe e , ind di ec mail appeals inc ease u nou by
abou 0.7 pe cen age poin s among Asian popula ions in
Cali o nia. The p esen s udy is designed o assess i , and o
wha deg ee, a mailed o e guide wi h in o ma ion ha helps
o ien a o e in ela ion o candida es in an elec ion inc eases
u nou . In ha sense, he p esen s udy is mos simila in
design and in en o Ga cía Bedolla and Michelson (2009),
hough i di e s in ha he elec o al con ex is a municipal
elec ion and he popula ion o in e es is young o e s.
Expe imen al design
We assess he mobilizing e ec o di ec mail solici a ions
using a andom sample o egis e ed o e s esiding in
Chicago. These da a a e d awn om a lis o all egis e ed o -
e s in Chicago p o ided by a hi d-pa y endo p io o he
elec ion (1,603,822 obse a ions in o al).3 Fo he pu poses o
ou expe imen we exclude he ollowing obse a ions om
ou analyses: hose egis an s who eside ou side o he ci y o
Chicago, o who a e younge han 18 o olde han 30. We
excluded egis an s who did no pa icipa e in he 2014 gen-
e al elec ion—bu e ained egis an s who did no pa icipa e
in he 2011 municipal elec ions— o iden i y mode a e p open-
si y o e s, a popula ion hough o be mos esponsi e o he
ea men we deployed in his expe imen (Elde s eld, 1956;
High on, 2004; Zalle , 1992).4 These c i e ia exclude 1,543,308
obse a ions o iginally in he da ase . We used a
Mille e al. 3
compu e -based andom numbe gene a o o assign 25,000
egis an s om he emaining 60,514 obse a ions o he
ea men g oup, wi h he balance kep as a con ol g oup. In
he pos -elec ion egis a ion lis s p o ided by he same endo
8,190 obse a ions—2,821 wi hin he ea ed g oup and 5,369
wi hin he con ol g oup—we e los o ollow-up.5 In sum, he
da a o his expe imen include 22,179 ea ed egis an s and
a con ol g oup o 30,145 obse a ions.6 The dependen a i-
able in ou s a is ical models is a dicho omous indica o o o -
ing in he 2015 mayo al elec ion, as epo ed o each
obse a ion in he o e egis a ion iles.7
The con ol a iables in ou models a e sou ced om wo
da ase s. Fi s , he o e egis a ion lis i sel includes indi id-
ual-le el da a on a egis an ’s gende , age, and ace. To con-
ol o neighbo hood-le el e ec s on u nou (Cho e al.,
2006; Gimpel e al., 2004) we collec ed he o al popula ion,
median age, median household income, p opo ion o he pop-
ula ion ha is whi e, black, La ino, o o he , and he p opo ion
o he popula ion in one o i e educa ional a ainmen ca ego-
ies om he 2013 Ame ican Communi y Su ey o each
Chicago census ac in ou da a. We epo a iable balance
ac oss he con ol and ea men g oups in Table 1. We ind no
signi ican di e ences ac oss he expe imen g oups, demon-
s a ing andomiza ion in he assignmen o he ea men .
The ea men g oup was mailed an in o ma ional bookle
in ad ance o he elec ion.8 This ea men condi ion is
in ended o es he idea, p esen ed by Be insky (2005), ha
a ba ie o pa icipa ion in elec ions is low le els o poli ical
in o ma ion among po en ial o e s. The English-language
maile con ained wo ele an se s o poli ical in o ma ion.
Fi s , i lis ed endo semen s om p ominen g oups and indi-
iduals o ou o he i e candida es o mayo .9 P e ious
esea ch has shown endo semen s a e help ul heu is ics o
o e s in cases whe e pa y a ilia ion is una ailable (Bowle
and Dono an, 1998; Scha ne e al., 2001). Mayo al elec-
ions in Chicago ha e been con es ed on a nonpa isan basis
since 1999. Second, he maile epo ed he posi ion o each
o he i e candida es o mayo wi h ega d o i e salien
policy issues in he campaign.10 This in o ma ion was
in ended o aid o e s in inding a candida e who sha es hei
policy p e e ences (Lau and Redlawsk, 1997, 2006). The
guide i sel did no con ey polling place loca ions—ins ead
i di ec ed he eade o he Chicago Elec ions Boa d websi e.
The guide men ions he span o he ea ly o ing window, he
da e o he elec ion, and ad ised egis an s o b ing some
o m o accep able iden i ica ion o he polls o a oid any
po en ial p oblem wi h ballo access.11
Ou hypo hesis ha he maile will inc ease u nou in he
elec ion will be con i med, hen, i ou analyses epo a posi-
i e and signi ican esul o he ea ed obse a ions. We use
logi and mul ile el models (Cho e al., 2006; Gelman and
Hill, 2006; Gimpel e al., 2004; S eenbe gen and Jones, 2002)
o es ima e he e ec s o his ea men condi ion.12
Findings
Ou indings demons a e he ea men inc eased u nou in
he Chicago municipal elec ion, all else being equal. Table
2 con ains he esul s o ou s a is ical models; he coe i-
cien s a e epo ed as log odds in each speci ica ion. We
Table 1. Va iable balance in con ol and ea men g oups.
Va iable Con ol g oup (N=30,145) T ea men g oup (N=22,179)
Mean S d. De . Mean S d. De .
Indi idual-le el a iables
2015 mayo al elec ion o e 0.327 0.469 0.336 0.473
Female 0.597 0.491 0.593 0.491
Age 25.44 3.41 25.48 3.43
Whi es 0.489 0.5 0.492 0.5
Blacks 0.247 0.431 0.246 0.431
La inos 0.138 0.345 0.136 0.343
Census ac a iables
Log o popula ion 8.2 0.5 8.2 0.49
Median age 34.09 5.8 34.07 5.81
Whi e popula ion % 47.27 34.52 47.25 34.48
Black popula ion % 37.14 40.65 37.16 40.7
La ino popula ion % 21.02 26.3 21.09 26.4
Popula ion % wi h a high school diploma 21.86 12.12 21.79 12.05
Popula ion % wi h some college 23.81 10.11 23.84 10.08
Popula ion % wi h a college deg ee 22.23 15.33 22.26 15.28
Popula ion % wi h a p o essional deg ee 15.51 13.63 15.49 13.56
Median household income 30.54 15.79 30.38 15.67
No es: Ho elling’s wo-g oup F es s a is ic (1.14) is nonsigni ican (p<0.31) indica ing he con ol and ea men g oups a e no signi ican ly di e en .
4 Resea ch and Poli ics
Table 2. Tu nou e ec s o o e guide.
P edic ing o ing in 2015 mayo al elec ion Model 1 Model 2 Model 3 Model 4
Indi idual a iables
Vo e guide ea men 0.042** 0.043** 0.043** −0.017
(0.019) (0.020) (0.020) (0.038)
Female egis an −0.096*** −0.074*** −0.075***
(0.019) (0.019) (0.019)
Age (18–30) 0.001 −0.001 −0.001
(0.004) (0.003) (0.003)
Whi e egis an 0.422*** 0.067* 0.066*
(0.035) (0.036) (0.036)
La ino egis an 0.737*** 0.076 0.079*
(0.040) (0.048) (0.048)
O he egis an 0.523*** 0.097** 0.096**
(0.042) (0.045) (0.045)
Census ac a iables
Log o popula ion −0.100*** −0.105***
(0.031) (0.030)
Median age 0.017*** 0.014***
(0.003) (0.003)
Whi e popula ion % 0.000 0.001
(0.001) (0.001)
Black popula ion % −0.007*** −0.005***
(0.001) (0.001)
La ino popula ion % 0.007*** 0.006***
(0.001) (0.001)
% wi h high school diploma 0.003 0.001
(0.003) (0.003)
% wi h some college 0.012*** 0.006**
(0.003) (0.003)
% college g adua es −0.000 −0.000
(0.003) (0.003)
% g adua e deg ees 0.006* 0.005
(0.004) (0.003)
Median household income −0.008
(0.027)
2nd household income qua ile 0.050
(0.054)
3 d household income qua ile 0.142**
(0.067)
Top household income qua ile −0.217**
(0.089)
Guide * 2nd household income qua ile 0.006
(0.054)
Guide * 3 d household income qua ile 0.094*
(0.053)
Guide * op household income qua ile 0.125**
(0.058)
Cons an −0.721*** −0.552*** −0.745** −0.506
(0.022) (0.095) (0.370) (0.369)
Obse a ions 52,324 52,324 52,324 52,324
Log pseudolikelihood −33,221 −32,888 −32,322 −32,294
Wald chi-squa e 4.686** 393.2*** 864.2*** 990.9***
Numbe o census ac s 794 794
No es: Robus s anda d e o s clus e ed by census ac a e in pa en heses. Cells epo logi coe icien s. The e e ence acial/e hnic g oup is black
egis an s. The income qua iles in Model 4 a e in e e ence o he 1s (bo om) household income qua ile.*** p<0.01, ** p<0.05, * p<0.1.
Mille e al. 5
i s discuss he esul s in he i s h ee models. The i s
and leas complica ed o hese models includes only he
ea men a iable. As we obse ed in Table 1, he e is no
signi ican di e ence be ween he con ol and ea men
g oups wi h ega d o he ba e y o explana o y a iables.
Model 1 shows he o e guide signi ican ly inc eased u n-
ou in he mayo al elec ion. The ea men e ec emains
posi i e and signi ican as addi ional indi idual-le el con-
ols a e inco po a ed (see Model 2) and when we es ima e
a mul ile el model o indi iduals wi hin Chicago census
ac s and include demog aphic da a om each census ac
(see Model 3). Black egis an s a e leas likely o o e,
ela i e o o he acial o e hnic g oups, bu his di e ence,
wi h ega d o La ino egis an s, anishes in Model 3.13
We use he esul s in Model 2 o es ima e he change in
he p edic ed p obabili y o o ing in he 2015 mayo al
elec ion o ge a be e sense o he e ec o he o e guide,
as logi coe icien s a e di icul o in e p e di ec ly. Figu e
1 shows he p edic ed p obabili y o o ing, and 95% con-
idence in e als, o he ea men and con ol g oups.
Tu nou among he ea ed obse a ions was abou 0.9
poin s highe han in he con ol g oup, 33.6% compa ed
wi h 32.7% ( he s anda d e o is 0.003). A di e ence o
means es shows his absolu e di e ence in u nou is s a-
is ically signi ican ( wo- ailed p<0.025). The ela i e
change in he p obabili y o o ing among he ea ed obse -
a ions, which can be ob ained by exponen ia ing he ea -
men coe icien in Table 2, is abou 4.4%.
This e ec is la ge ela i e o he mobiliza ion li e a u e.
G een and Ge be (2015: 186–189) epo , on he basis o a
me a-analysis o 85 s udies be ween 1998 and 2014, he
a e age ea men e ec o a nonpa isan maile wi hou an
elemen o social p essu e is 0.523 poin s. I we limi ou
conside a ion o he se en s udies in he me a-analysis con-
duc ed in he con ex o a municipal elec ion, he a e age
ea men e ec o a maile is 0.04 poin s. When compa ed
wi h o he mobiliza ion s a egies speci ically a ge ing
young o e s, hese esul s a e mo e cos -e ec i e han
o he di ec mail expe imen s, bu less e ec i e han phone
banking o can assing and lea le ing e o s (see Young
Vo e S a egies 2006 o a summa y o his esea ch).
Is his inc ease in u nou cons an ac oss all obse a-
ions? We know, o example, ha weal hie and be e edu-
ca ed ci izens a e mo e poli ically ac i e (Delli Ca pini and
Kee e , 1997) and mo e likely o o e (Leighley and Nagle ,
2014). G een and Ge be (2015: 63) also ind maile s ha e
weak u nou e ec s among low-p opensi y o e s. I may be
he case, hen, ha he maile did no inc ease he likelihood
o o ing among obse a ions in less weal hy neighbo hoods
o Chicago. Model 4 p esen s subg oup analyses o he ea -
men e ec by each qua ile o he median household income
qua iles in he 794 census ac s in hese da a. In e es ingly,
he p obabili y o o ing is highes among obse a ions in he
3 d income qua ile and lowes among obse a ions in he
op qua ile, when compa ed wi h he e e ence bo om
income qua ile. The e ec o he o e guide ea men is
also only disce nible among he op wo income qua iles.
These esul s sugges he e ec on u nou is limi ed o obse -
a ions in mo e a luen census ac s in Chicago.14 Addi ional
analyses a e epo ed in he online appendix.
Las , we assess he cos -e ec i eness o his maile . G een
and Ge be (2015: 16) s a e “[i]n o de o know whe he a
campaign ac ic is cos -e ec i e, i is necessa y o de e mine
how many o es a e p oduced o each dolla spen .” We
es ima e he cos -e ec i eness o he maile , on an in en - o-
ea basis, is abou $56 (o abou 111 maile s we e equi ed
o p oduce a single o e). The e ec i eness o his maile is
qui e high ela i e o o he s udies conside ed by G een and
Ge be , who ind “one addi ional o e is gene a ed o e e y
273 people who ecei e a con en ional nonad ocacy GOTV
maile ” and ha maile s wi h a p oduc ion cos o $0.50 cos
$91 pe o e (G een and Ge be , 2015: 66).
Conclusions
We sough o empi ically es Be insky’s (2005) sugges ion
ha cogni i e mobiliza ion is he key o inc easing pa ici-
pa ion in Ame ican elec ions. We na owed ou s udy o
young o e s in he 2015 mayo al elec ion in Chicago. Ou
esul s indica e he mailed o e guide signi ican ly
inc eased u nou . When we accoun o income le els in
di e en pa s o Chicago, howe e , we ind he ea men
only inc eased u nou among weal hie neighbo hoods in
he ci y. We ind, like much o he li e a u e on mino i y
mobiliza ion, no signi ican e ec ac oss acial o e hnic
g oups. The scale o he e ec o an inc ease in poli ical
in o ma ion epo ed he e is simila o ecen s udies o
di ec mail (Ga cía Bedolla and Michelson, 2009; Ge be
and G een, 2000b) hough he con ex o a local elec ion
and he ocus on young o e s is dis inc i e.
This s udy sugges s addi ional expe imen al designs ha
may shed u he ligh on he ole o cogni i e mobiliza ion
Figu e 1. P edic ed p obabili y o o ing in expe imen g oups.

6 Resea ch and Poli ics
wi h ega d o o e u nou . The p esen s udy is limi ed o
young o e s wi h some p io expe ience wi h o ing. I is
an open ques ion o wha deg ee a simila expe imen may
mobilize newly egis e ed o e s wi h no p io o ing expe-
ience. Mo eo e , a simila expe imen al amewo k in con-
junc ion wi h p e- and pos -elec ion su eys could ell us
some hing abou he po en ial o addi ional poli ical in o -
ma ion o change a o e ’s choice in he cou se o an elec-
ion campaign. Replica ing his expe imen in a a ie y o
elec o al con ex s, o example open sea aces, when an
incumben ails o win eelec ion, o in locali ies wi h a a-
di ion o high u nou in local elec ions could demons a e
he obus ness o hese indings and u he assess he e ec s
o cogni i e mobiliza ion. This expe imen only es s he
e ec o di ec mail. Fu u e wo k should ake accoun o
mul iple “ ea men channels” (G een and Ge be , 2015;
Mille e al., 1981) o gene a e cogni i e mobiliza ion and
he e o e u nou . This s udy—and o he expe imen al
designs—ha e shown ch onic low u nou in Ame ican
elec ions by young o e s is a ea able condi ion.
Acknowledgemen s
Ou hanks o Ke in A ceneaux, Maneesh A o a, Josh Be ezin,
Neil Cha u edi, Donald G een, Dan Hopkins, Ke em Kalkan,
and Hen y K aeme o help ul commen s on an ea lie d a o
his p ojec . Pe e Mille is he Fulb igh -Uni e si y o Tampe e
Schola . Rebecca Reynolds is a pas execu i e di ec o o Chicago
Vo es. Ma hew Singe is a pas execu i e di ec o o he Alliance
o You h Ac ion. The da ase used in his analysis, he eplica ion
iles, and he o e guide a e s o ed a a Ha a d Da a e se eposi-
o y a ailable he e: h p://bi .ly/2 yXZJR
Au ho s’ No e
This p ojec was p e iously p esen ed a a Beha io al E hics Lab
semina a he Uni e si y o Pennsyl ania, a a Global S udies
Resea ch Semina a he Uni e si y o Tampe e, and a a mee ing
o he Midwes Poli ical Science Associa ion.
Decla a ion o Con lic ing In e es
The au ho s decla e ha he e is no con lic o in e es .
Funding
This esea ch ecei ed no speci ic g an om any unding agency
in he public, comme cial, o no - o -p o i sec o s.
Supplemen a y Ma e ial
The supplemen a y iles a e a ailable a h p://jou nals.sagepub.
com/doi/suppl/10.1177/2053168017738410. The eplica ion iles
a e a ailable a : h ps://da a e se.ha a d.edu/da a e se/ esea ch
andpoli ics.
No es
1. The deba e on whe he he ques ion o low u nou is a p ob-
lem o he conduc o democ a ic poli ics is i al, bu we a e
aced wi h space cons ain s and unable o summa ize ha
deba e he e. Ins ead, we sugges in e es ed eade s consul
G een e al. (2013) o a e iew o hese opics.
2. See P imo e al. (2007) o a commen on he es ima ion
s a egy deployed by Wol inge e al. (2005). The e ised
es ima es p esen ed by P imo and colleagues, howe e , dem-
ons a e ha u nou is highe among egis an s aged 18–24
who do no li e wi h hei pa en s and a e mailed a sample
ballo (see Table 1).
3. These da a we e sou ced om Sma VAN (h p://www.sma -
anda a.com/sma VAN). These da a epo sligh ly ewe bal-
lo s cas (a di e ence o abou 4.2%) and mo e egis e ed o e s
(abou 12.8%) han he o icial da a epo ed by he Chicago
Boa d o Elec ion Commissione s. As a consequence o he
la ge se o egis e ed o e s, he u nou a e in he o al da a-
se is 28.89% compa ed wi h 34.03% epo ed by he Boa d.
4. Ou in en ion wi h hese c i e ia is o exclude i s - ime and
in equen o e s who may need o de o e g ea e ime o
sea ch cos s and anspo a ion o he polling place (B ady
and McNul y, 2011).
5. Obse a ions los o ollow-up can be due o a numbe o
ac o s. Fo example, a pe son who mo es ou side o Cook
Coun y and egis e s o o e in he new loca ion would be los
o ollow-up in his expe imen . The Census Bu eau es ima es
abou 11.5% o Ame icans mo ed be ween 2013 and 2014
(see Geog aphical Mobili y: 2013 o 2014 epo s a ailable
a h p://www.census.go / opics/popula ion/mig a ion.h ml).
These 8,190 obse a ions amoun o abou 13.5% o he
o al obse a ions in his expe imen . This a e is highe han
he na ional a e age, bu i is also well known ha younge
Ame icans a e mo e mobile han olde coho s (Squi e e al.,
1987). Ins i u ionalized pe sons a e also emo ed om o e
egis a ion lis s (McDonald and Popkin, 2001). In sho , we
see no eason o conclude hese obse a ions los o ollow-
up a e a sou ce o bias o he esul s we epo below.
6. We conduc ed wo powe analyses wi h hese da a. Fi s ,
we de e mined we ha e mo e han a su icien numbe o
obse a ions o de ec he obse ed di e ence in u nou in
he con ol and ea men g oups; he minimum de ec able
sample is 35,441. Second, we de e mined hese da a p o ide
mo e han 95% powe (99.22% o be exac ).
7. The dependen a iable includes egis an s who o ed ei he
a he polls on Elec ion Day o p io o Elec ion Day. We
explo ed he ela ionship be ween ecei ing a o e guide in
he mail and o ing ea ly, bu ound no signi ican e ec s
among he ea ed obse a ions.
8. The maile was designed o con ey o he ecipien ha he
in o ma ion was solely he c ea ion o Chicago Vo es and no
p oduced by he Elec ion Boa d o a candida e’s campaign. The
o iginal in en ion o he o e guide was o dis ibu e i by pe -
sonal in e ac ions, bu his dis ibu ion s a egy was changed o
mail as a esul o posi i e eedback om o e s and o explic-
i ly es he ea men e ec in an expe imen al amewo k.
9. The i h candida e, William Walls, had no endo semen s
when he o e guide was p oduced. Walls won 2% o he
o e in he gene al elec ion.
10. The i e issues in he maile we e selec ed a e a p ocess
o p e-elec ion ocus g oups wi h di e se acial, gende , and
geog aphic samples. F om he policies hese ocus g oups
men ioned, we p io i ized issues ha we e also equen ly
Mille e al. 7
men ioned in he news co e age o he elec ion. Las ly,
we emo ed alence issues om he se o policies o be
included in he maile o highligh a eas in which he candi-
da es disag eed on salien policies. Candida e posi ions we e
de e mined on he basis o ecen s a emen s om he can-
dida e’s websi e o , i no policy posi ion was desc ibed on
hei websi e, om he mos ecen public s a emen s made
in he media.
11. The o e guide also includes summa ies o ou non-bind-
ing ballo measu es and a ecommenda ion o how o o e
on hose measu es. An analysis o he e ec o he guides
on hese ballo measu es is beyond he scope o his s udy,
hough i is no ewo hy ha each measu e was app o ed wi h
be ween 78.99% and 88.9% o o es cas ( he esul s o he
measu e o sugges an elec ed school boa d a e only a ailable
o 37 o he ci y’s 50 elec o al wa ds).
12. The esul s p esen ed in Models 3 and 4 in Table 2 a e
ob ained in a andom in e cep model. The esul s do no
change i we un a andom slope model as an al e na i e
speci ica ion.
13. I migh be he case ha he e ec o he maile is ac ually
jus a eminde o an upcoming Elec ion Day o an appeal
based on ci ic du y. Nei he al e na i e accoun , howe e ,
appea s plausible. Wi h ega d o elec ion eminde s, G een
and Ge be (2015: 63) ind hese maile s ha e no e ec on
u nou . “Fi e expe imen s ha ha e es ed eminde s yield
an o e all es ima e ha is a shade less han ze o.” Wi h ega d
o ci ic du y-based appeals, his maile makes no e e ence
o s anda d ci ic du y elemen s like pas u nou , an incen i e
o o e, o a hank ul o scolding one o inc ease u nou and
hus hese hemes a e unlikely o explain he e ec o his
maile .
14. Compa ing mean u nou be ween he con ol and ea men
g oups ac oss he ou income ca ego ies a i es a he same
esul . Tu nou be ween he ea ed and con ol g oups is no
signi ican ly di e en in he bo om o second qua ile o
median household income. Tu nou among he ea ed obse -
a ions is signi ican ly highe in he hi d (0.4066 compa ed
wi h 0.3891, wo- ailed p<0.05) and op (0.3281 compa ed
wi h 0.3044, wo- ailed p<0.01) income qua iles. We an an
iden ical analysis o assess i he o e guide had an e ec on
ei he he acial and e hnic g oups in ou da a o in census
ac s wi h a ying p opo ions o college g adua es. These
analyses e ealed no signi ican e ec s. These esul s a e
a ailable om he au ho s upon eques .
Ca negie Co po a ion o New Yo k G an
This publica ion was made possible (in pa ) by a g an om
Ca negie Co po a ion o New Yo k. The s a emen s made and
iews exp essed a e solely he esponsibili y o he au ho .
Re e ences
Be insky A (2005) The pe e se consequences o elec o al
e o m in he Uni ed S a es. Ame ican Poli ical Resea ch
33: 471–491.
Bowle S and Dono an T (1998) Demanding Choices: Opinion,
Vo ing, and Di ec Democ acy. Ann A bo : Uni e si y o
Michigan P ess.
B ady H and McNul y J (2011) Tu ning ou o o e: The cos s o
inding and ge ing o he polling place. Ame ican Poli ical
Science Re iew 105: 115–134.
Cho W, Gimpel J and Dyck J (2006) Residen ial concen a ion,
poli ical socializa ion, and o e u nou . Jou nal o Poli ics
68: 156–167.
Coppock A and G een D (2015) Is o ing habi o ming? New
e idence om expe imen s and eg ession discon inui ies.
Ame ican Jou nal o Poli ical Science 60: 1044–1062.
Delli Ca pini M and Kee e S (1997) Wha Ame icans Know abou
Poli ics and Why i Ma e s. New Ha en: Yale Uni e si y P ess.
Elde s eld S (1956) Expe imen al p opaganda echniques and
o ing beha io . Ame ican Poli ical Science Re iew 50:
154–165.
Ga cía Bedolla L and Michelson M (2012) Mobilizing Inclusion:
T ans o ming he Elec o a e h ough Ge -Ou - he-Vo e
Campaigns. New Ha en: Yale Uni e si y P ess.
Ga cía Bedolla L and Michelson M (2009) Wha do o e s need
o know? Tes ing he ole o cogni i e in o ma ion in Asian
Ame ican o e mobiliza ion. Ame ican Poli ics Resea ch 37:
254–274.
Gelman A and Hill J (2006) Da a Analysis Using Reg ession
and Mul ile el/Hie a chical Models. New Yo k: Camb idge
Uni e si y P ess.
Ge be A and G een D (2000a) The e ec o a nonpa isan ge -
ou - he- o e d i e: An expe imen al s udy in lea le ing.
Jou nal o Poli ics 62: 846–857.
Ge be A and G een D (2000b) The e ec s o can assing, el-
ephone calls, and di ec mail on o e u nou : A ield expe i-
men . Ame ican Poli ical Science Re iew 94: 653–663.
Ge be A, G een D and Shacha R (2003) Vo ing may be habi -
o ming: E idence om a andomized ield expe imen .
Ame ican Jou nal o Poli ical Science 47: 540–550.
Gimpel J, Dyck J and Shaw D (2004) Regis an s, o e s, and
u nou a iabili y ac oss neighbo hoods. Poli ical Beha io
26: 343–375.
Gosnell H (1927) Ge ing ou he Vo e: An Expe imen in he
S imula ion o Vo ing. Chicago: Uni e si y o Chicago P ess.
G een D (2004) Mobilizing A ican-Ame ican o e s using di ec
mail and comme cial phone banks: A ield expe imen .
Poli ical Resea ch Qua e ly 57: 245–255.
G een D and Ge be A (2015) Ge Ou he Vo e: How o Inc ease
Vo e Tu nou . 3 d edi ion. Washing on: B ookings
Ins i u ion P ess.
G een D and Shacha R (2000) Habi o ma ion and poli ical
beha io : E idence o consue ude in o e u nou . B i ish
Jou nal o Poli ical Science 30: 561–573.
G een D, McG a h M and A onow P (2013) Field expe imen s
and he s udy o o e u nou . Jou nal o Elec ions, Public
Opinion, and Pa ies 23: 27–48.
Hajnal Z and Lewis P (2003) Municipal ins i u ions and o e
u nou in local elec ions. U ban A ai s Re iew 38: 645–668.
High on B (2004) Vo e egis a ion and u nou in he Uni ed
S a es. Pe spec i es on Poli ics 2: 507–515.
High on B and Wol inge R (2001) The i s se en yea s o he
poli ical li e cycle. Ame ican Jou nal o Poli ical Science 45:
202–209.
Holb ook T and Weinschenk A (2014) Campaigns, mobiliza-
ion, and u nou in mayo al elec ions. Poli ical Resea ch
Qua e ly 67: 42–55.
8 Resea ch and Poli ics
Kau mann K, Pe ocik J and Shaw D (2008) Uncon en ional
Wisdom: Fac s and My hs Abou Ame ican Vo e s. New
Yo k: Ox o d Uni e si y P ess.
Lau R and Redlawsk D (1997) Vo ing co ec ly. Ame ican
Poli ical Science Re iew 91: 585–598.
Lau R and Redlawsk D (2006) How Vo e s Decide: In o ma ion
P ocessing in Elec ion Campaigns. Camb idge: Camb idge
Uni e si y P ess.
Leighley J and Nagle J (2014) Who Vo es Now? Demog aphics,
Issues, Inequali y, and Tu nou in he Uni ed S a es.
P ince on: P ince on Uni e si y P ess.
Ma land R and Mu ay G (2012) An expe imen al es o mobi-
liza ion e ec s in a La ino communi y. Poli ical Resea ch
Qua e ly 65: 192–205.
McDonald M and Popkin S (2001) The my h o he anishing
o e . Ame ican Poli ical Science Re iew 95: 963–974.
Mille R, Bosi is D and Bae D (1981) S imula ing o e u nou
in a p ima y: Field expe imen wi h a p ecinc commi ee-
man. In e na ional Poli ical Science Re iew 2: 445–460.
Mu z D (2002) The consequences o c oss-cu ing ne wo ks o
poli ical pa icipa ion. Ame ican Jou nal o Poli ical Science
46: 838–855.
Mu z D (2006) Hea ing he O he Side: Delibe a i e Ve sus
Pa icipa o y Democ acy. Camb idge: Camb idge Uni e si y
P ess.
Nicke son D (2006) Hun ing he elusi e young o e . Jou nal o
Poli ical Ma ke ing 5: 47–69.
Nyhan B and Rei le J (2010) When co ec ions ail: The pe sis ence
o poli ical mispe cep ions. Poli ical Beha io 32: 303–330.
Oli e JE (2012) Local Elec ions and he Poli ics o Small-Scale
Democ acy. P ince on: P ince on Uni e si y P ess.
P imo D, Jacobsmeie M and Milyo J (2007) Es ima ing he
impac o s a e policies and ins i u ions wi h mixed-le el
da a. S a e Poli ics and Policy Qua e ly 7: 446–459.
Plu ze E (2002) Becoming a habi ual o e : Ine ia, esou ces,
and g ow h in young adul hood. Ame ican Poli ical Science
Re iew 96: 41–56.
Samii C (2016) Causal empi icism in quan i a i e esea ch.
Jou nal o Poli ics 78: 941–955.
Scha ne B, S eb M and W igh G (2001) Teams wi hou uni-
o ms: The nonpa isan ballo in s a e and local elec ions.
Poli ical Resea ch Qua e ly 54: 7–30.
Shi Y (2016) C oss-cu ing messages and o e u nou :
E idence om a same-sex ma iage amendmen . Poli ical
Communica ion 33: 433–459.
Shea D and G een J (eds) (2006) Foun ain o You h: S a egies
and Tac ics o Mobilizing Ame ica’s Young Vo e s.
Washing on: Rowman & Li le ield.
Squi e P, Wol inge R and Glass D (1987) Residen ial mobili y
and o e u nou . Ame ican Poli ical Science Re iew 81:
45–66.
S eenbe gen M and Jones B (2002) Modeling mul ile el da a
s uc u es. Ame ican Jou nal o Poli ical Science. 46: 218–
237.
Teixei a R (1987) Why Ame icans Don’ Vo e: Tu nou Decline in
he Uni ed S a es 1960–1984. Wes po : G eenwood P ess.
T i edi N (2005) The e ec o iden i y-based GOTV di ec mail
appeals on he u nou o Indian Ame icans. Annals o he
Ame ican Academy o Poli ical and Social Science 601: 115–122.
Wa enbe g M (2002) Whe e Ha e all he Vo e s Gone?
Camb idge: Ha a d Uni e si y P ess.
Wa enbe g M (2012) Is Vo ing o Young People? 3 d edi ion.
Bos on: Pea son.
Wol inge R, High on B and Mullin M (2005) How pos egis a-
ion laws a ec he u nou o ci izens egis e ed o o e.
S a e Poli ics and Policy Qua e ly 5: 1–23.
Wong J (2005) Mobilizing Asian Ame ican o e s: A ield expe i-
men . Annals o he Ame ican Academy o Poli ical and
Social Science 601: 102–114.
Young Vo e S a egies (2006) Young Vo e Mobiliza ion
Tac ics. A ailable a : h p://ci icyou h.o g/PopUps/Young_
Vo e s_Guide.pd (las accessed 2 May 2017).
Zalle J (1992) The Na u e and O igins o Mass Opinion.
Camb idge: Camb idge Uni e si y P ess.