Presentation is loading. Please wait.

Presentation is loading. Please wait.

Cell-Phone-Only Voters in the 2008 Exit Poll Michael Mokrzycki, Associated Press Scott Keeter, Pew Research Center Courtney Kennedy, University of Michigan.

Similar presentations


Presentation on theme: "Cell-Phone-Only Voters in the 2008 Exit Poll Michael Mokrzycki, Associated Press Scott Keeter, Pew Research Center Courtney Kennedy, University of Michigan."— Presentation transcript:

1 Cell-Phone-Only Voters in the 2008 Exit Poll Michael Mokrzycki, Associated Press Scott Keeter, Pew Research Center Courtney Kennedy, University of Michigan and The Everett Group American Association for Public Opinion Research Hollywood, Florida May 2009

2 Research questions Cell-phone-only incidence & rate of growth How do CPO voters differ from landline-reachable? –Demographics and political attitudes –Comparison to 2004 exit poll Bias implications for future surveys that dont attempt to cover the wireless-only population

3 Historical context Were now a quarter-century into the cell phone age –Motorola DynaTAC (1983) 1990s: heyday of list-assisted RDD

4 Historical context This decade –Wireless substitution becoming widespread –RDD HH coverage back to early 1960s levels

5 Do cells harm surveys? Until recently, CPO were small fraction of US population/households Differences were small or non-existent between subgroups most likely to be cell-only and their LL-reachable demographic cohorts Typically, post-stratification weighting cured what minor ills existed

6 Recent caution flags Cell-only population keeps growing –By second half of 2008, 20% of US households were cell-only, outnumbering landline-only (Blumberg and Luke 2009) Emerging evidence: small but persistent Dem bias –Three Pew surveys in summer 2008 consistently found 2- to 3- point shift toward Obama when including cell interviews

7 2008 NEP national exit poll In-person survey: Opportunity to reach cell-only voters

8 2008 NEP national exit poll In-person survey: Opportunity to reach cell-only voters CAVEATS: Subject population: Voters who cast ballots on Election Day Excludes 33% of 2008 electorate who voted early/absentee (NEP did cover early voters in 18 states and nationally with landline RDD but no cell sample)

9 NEP national exit poll methodology Stratified probability sample of 300 precincts nationwide Random selection of voters within each precinct (interval – every n th voter) Self-administered PAPI

10 NEP national exit poll methodology Four versions of questionnaire; phone-status question on two of them Same question asked in 2004 national exit poll

11

12 Methodology – sample sizes 2004 N = 5,619 –Including 385 cell-phone only 2008 N = 7,341 –Including 1,496 cell-phone-only Good news: Bigger N in 2008 to study the cell-only population!

13 Household telephone status in Election Day exit polls

14 Presidential vote by HH phone status in Election Day exit polls

15 CPO status by age in Election Day exit polls

16 Growing – and aging – CPO population in Election Day exit polls Growing majority of cell-only voters are age 30+ –2004: 52% –2008: 57% CPO growth rate higher among older age groups than among younger (albeit on smaller 2004 bases) Increasing adoption of cell-only status by older people Younger voters aging + maintaining phone habits

17 CPO: select demographics in 2008 Election Day exit polls TOTAL cell-phone-only: 20% Did not complete HS: 27% (post-grad 14%) Income: < $15k: 37%; $15-50k: 26% Hispanic/Latino: 28% (white 19%, black 22%) Party ID: D 21%, R 17%, I 20% Ideology: Lib 24%, Mod 21%, Cons 16%

18 Pres. vote by age and phone status in 2008 Election Day exit polls

19 From 2004 to 2008, bigger differences between LL- reachable and CPO among both < 30 and 30+ But the differences increased at a faster rate among older voters

20 Pres. vote by age and phone status in 2008 Election Day exit polls LL 18-29CPO 18-29LL 30+CPO 30+ 2004 Kerry 56%58%46%49% 2008 Obama 62%67%48%56%

21 Other variables by phone status in 2008 Election Day exit polls Some variables produced no sig. diff. between CPO and LL-reachable. One curious result: –No difference on worry about direction of nations economy in next year –But CPOs were a little more likely than LL-reachable to call the current condition of the nations economy poor

22 Other variables by phone status in 2008 Election Day exit polls Some variables did show sig. differences. Often related to partisanship/ideology: –CPOs a bit more excited than LL-reachable about an Obama presidency (by 6 points) –CPOs 10 points more likely to strongly disapprove of Bush job performance, and to say McCain would continue Bushs policies

23 Voter mobilization by phone status in 2008 Election Day exit polls Cell-only voters were far more likely (22%) than those with landlines (10%) to have voted for first time in 2008 –Naturally this correlates strongly with age: youngest voters most likely to be cell-only and also most likely to be new voters

24 GOTV by phone status in 2008 Election Day exit polls Cell-only Election Day voters were far less likely to have been contacted by either campaign. –74% of CPO said not contacted by either campaign, vs 56% of LL-reachable Even Obama's campaign – renowned for its technological innovation – was more likely to contact LL- reachable than CPO-voters –34% of LL-reachable reported contact from Obama campaign, vs 23% of CPO

25 GOTV by phone status in 2008 Election Day exit polls So its not only pollsters who have trouble reaching those who are wireless-only.

26 Possible reasons for political differences between CPO and LL Likely stem from socio-economic differences –CPO less affluent and, among older voters, less educated. –Income differences large and consistent: more cell-only voters have incomes below $50,000 in all age groups. In particular: 30-39: 47% of CPO < $50k vs 28% of LL-reachable 40-49: 50% of CPO < $50k vs 25% of LL-reachable –Income and education were strong predictors of presidential voting in 2008, and may explain at least part of the differences in presidential preference by telephone status.

27 But a phone effect remains Multivariate regression analysis: –differences in presidential preference persist even when controlling for all available demographic variables Some CPO characteristics distinguish them politically from other voters but werent in the NEP exit poll –marriage, children in the home, homeownership, and religiosity and church attendance

28 Meta-analysis of pre-election polls No compelling evidence that sample design affected accuracy of the difference between the candidates But there is evidence that the point estimates for each candidate were systematically affected, a little –Polls that excluded wireless were most likely to underestimate Obama vote (generally by 2-3 points) –Consistent with previous studies –In 2008, post-survey weighting still sufficient cured these ills

29 Conclusions No sign cell-only growth is slowing Small but consistent bias in 08 pre-election polls w/o CPOs Clear difference in presidential vote in exit polls by phone status CPO incidence growing at faster rate among 30+ than younger voters – and older CPO voting more Dem in 2008 than 2004

30 Conclusions No evidence that non-coverage bias from excluding CPOs might diminish rather than grow in coming years Many national pre-election pollsters concluded in 2008 that they needed to include cell samples. Still true. Also true for the early voting component of the exit polls – need cell coverage in the phone polls NEP conducts the week before the election to capture advance voters

31 Thank you! mike@mikemokr.com skeeter@pewresearch.org ckkenned@umich.edu


Download ppt "Cell-Phone-Only Voters in the 2008 Exit Poll Michael Mokrzycki, Associated Press Scott Keeter, Pew Research Center Courtney Kennedy, University of Michigan."

Similar presentations


Ads by Google