We use cookies to enhance your experience on our website. By continuing to use our website, you are agreeing to our use of cookies. You can change your cookie settings at any time. Find out more

Oxford Journals

On the Overreport Bias of the National Election Study Turnout Rate

On the Over-Report Bias of the National Election Study Turnout Rate

by Michael P. McDonald
George Mason University
http://elections.gmu.edu
mmcdon@gmu.edu

This table provides data referenced in Michael McDonald (2003) "On the Over-Report Bias of the National Election Study Turnout Rate." As the authors of The American Voter (1960) and subsequent researchers of the National Election Study (NES) have noted, more NES respondents claim to have voted than actual voting statistics would suggest. Data presented here are a good-faith effort to reconcile the NES sample with the turnout rate of the population it is drawn from. These data show no on-going, increasing bias in the NES presidential turnout rate, as claimed by Barry Burden (2000). I challenge Burden's claim that the over-report bias is related to the NES response rates, which are also provided below. In multivariate analysis of competing theories to explain the turnout rate, I find modest support for a theory that relates the over-report bias to the turnout rate itself (when fewer voters vote, there are more opportunities for voters to misreport their vote), but any inference is seriously crippled by the small number of observations. Resources for this paper and turnout rates in general may be found at http://elections.gmu.edu/voter_turnout.htm.

References

Burden, Barry C. 2000. �Voter Turnout and the National Election Studies.� Political Analysis 8(4): 389-398. �Voter Turnout and the National Election Studies.� Political Analysis 8(4): 389-398.

Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. 1960. The American Voter. University of Chicago Press: Chicago.

Luevano, Patricia. 1994. �Response Rates in the National Election Studies, 1948-1992.� ICSPR Technical Report #44. Ann Arbor, MI.

McDonald, Michael P. and Samuel Popkin. 2001. �The Myth of the Vanishing Voter.� American Political Science Review 95(4): 963-974.

McDonald, Michael P. 2003. �On the Over-Report Bias of the National Election Study Turnout Rate.� Political Analysis, forthcoming.

Sapiro, Virginia, Steven J. Rosenstone, and the National Elections Studies. AMERICAN NATIONAL ELECTION CUMULATIVE DATA FILE, 1948-2000 [Computer file]. 11^th ICPSR version. Ann Arbor, MI: University of Michigan, Center for Political Studies [producer], 2002. Ann Arbor, MI: Inter-Consortium for Political and Social Research [distributor], 2002.

Variable Description

Turnout Rate, NES: From the 1948-2000 Cumulative Data File (Sapiro, Rosenstone, and NES 2002). Calculated using variable VCF0702 and weight variable VCF0009A, no response coded as missing data.

Vote for Highest Office: From McDonald and Popkin (2001). Vote for President in presidential election years, in non-presidential election years, the highest vote total for a statewide office. If no statewide office, sum of the congressional elections.

Voting-Age Population: Estimates drawn from the Census Bureau of persons of voting age living in the United States.

Turnout Rate, VAP: Vote for Highest Office divided by Voting Age Population.

Non-Citizens: Non-citizens as a percent of voting-age population, drawn from Current Population Survey, Voter Supplement File and various Census of the Population. See McDonald and Popkin (2001) for full explanation.

Prisoners: Department of Justice statistics on number of felony prisoners. See McDonald and Popkin (2001) for full explanation.

Overseas Voters: Estimate of number of ballots cast by overseas voters. See McDonald (2003) and McDonald and Popkin (2001) for full explanation.

Turnout Rate, Reconciled VAP: Estimate of turnout rate among NES sample population. Calculated by subtracting Overseas Voters from Vote for Highest Office and subtracting Non-Citizens and Prisoners from Voting-Age Population. First resulting number is divided through by the second.

NES-VAP Turnout Gap: Turnout Rate, NES - Turnout Rate, VAP.

NES-Reconciled VAP Turnout Gap: Turnout Rate, NES - Turnout Rate, Reconciled VAP.

NES Response Rates: For non-presidential elections, NES conducts one a post-election survey. The response rates for these years are simply the response rates to the post-election survey. For most presidential elections, NES conducts pre- and post-election surveys. The voting question is asked on the post-election survey. The post-election response rate is then the pre-election response rate multiplied by the re-interview rate. Data from Luevano (1994) and subsequent NES codebooks. Where blank, no data available.

Table of Data (Electronic version in Excel format)

Year	Turnout Rate NES	Vote for Highest Office (1000�s)	Voting-Age Population (1000�s)	Turnout Rate VAP (%)	Non-Citizens (1000�s)	Prisoners (1000�s)	Overseas Voters (1000�s)	Turnout Rate Reconciled VAP (%)	NES-VAP Turnout Rate	NES-ReconciledVAP Turnout Rate	NES Pre-Election Response Rate	NES Re-Interview Rate	NES Post-Election Response Rate
1948	63.9	48833	95573	51.1	2198	156	230	52.1	12.8	11.8	83.9	94.6	79.4
1950		41984	98134	42.8	1880	166	170	43.5
1952	73.0	61552	99929	61.6	1899	168	705	62.2	11.4	10.8	77.2	90.3	69.7
1954		43854	102075	43.0	1939	183	429	43.4
1956	72.9	62027	104515	59.3	1986	190	591	60.0	13.6	12.9	85.0	90.9	77.3
1958	57.6	47203	106447	44.3	2129	206	428	44.9	13.2	12.7			78.1
1960	79.0	68838	109672	62.8	2193	213	582	63.6	16.2	15.4		93.9
1962	60.4	53141	112952	47.0	2259	219	531	47.6	13.4	12.8
1964	77.7	70645	114090	61.9	2282	214	761	62.6	15.7	15.0	80.6	92.7	74.7
1966	62.3	56188	116638	48.2	2363	200	789	48.6	14.1	13.7			77.1
1968	75.8	73213	120285	60.9	2766	187	1141	61.4	15.0	14.4	77.4	86.6	67.0
1970	59.4	58014	124498	46.6	3148	196	835	47.2	12.8	12.2			76.6
1972	72.8	77719	140777	55.2	3640	196	889	56.1	17.6	16.7	75.0	84.4	63.3
1974	52.5	55944	146338	38.2	4148	218	590	39.0	14.3	13.6			70.0
1976	71.6	81556	152308	53.5	4558	263	858	54.7	18.0	16.9	70.4	84.9	59.8
1978	54.5	58918	158369	37.2	5780	284	673	38.2	17.3	16.3			68.9
1980	71.4	86515	163945	52.8	6827	330	986	54.6	18.6	16.8	71.8	87.4	62.8
1982	60.4	67616	169643	39.9	10554	414	838	42.1	20.5	18.3			72.3
1984	73.6	92653	173995	53.3	13252	462	1350	57.0	20.4	16.6	72.1	88.6	63.9
1986	52.5	64991	177922	36.5	12223	545	864	38.8	16.0	13.7			67.7
1988	69.7	91595	181956	50.3	13942	628	1370	53.9	19.3	15.8	70.5	87	61.3
1990	46.6	67859	185888	36.5	16297	774	1058	39.6	10.1	7.1			70.6
1992	75.1	104405	189687	55.0	17826	884	1465	60.2	20.1	14.9	74.0	89.3	66.1
1994	55.7	75106	193163	38.9	13205	1055	932	41.5	16.8	14.2			74.1
1996	72.9	96263	196928	48.9	13948	1180	1317	52.2	24.0	20.6	71.0	90.0	63.9
1998	52.1	72537	200929	36.1	15070	1222	1148	38.7	16.0	13.4			63.9
2000	72.7	105326	205813	51.2	16500	1386	1672	55.2	21.5	17.5	61.2	86.0	52.6

Stata (v 6.0) Output

rec_gap = NES-Reconciled VAP Turnout Gap
rec_rate = Turnout Rate, Reconciled VAP
response = NES Post-Election Response Rate

reg rec_gap rec_rate response if(election==1&year~=48&year~=96)

Source | SS df MS Number of obs =11
---------+------------------------------
           F( 2, 8) = 6.64
   Model | 25.3305851 2 12.6652925 Prob > F = 0.0200
Residual | 15.2675152 8 1.9084394
---------+------------------------------
       R-squared = 0.6239
   Adj R-squared = 0.5299
   Total | 40.5981003 10 4.05981003 Root MSE = 1.3815
----------------------------------------------------------------------------
rec_gap | Coef.     Std. Err. t    P>|t| [95% Conf.Interval]
---------+------------------------------------------------------------------
rec_rate | -.3042361 .2066325 -1.472 0.179 -.7807314, .1722593
response | -.0969068 .1002358 -0.967 0.362 -.3280509, .1342373
   _cons | 39.27846 8.066552 4.869 0.001   20.67696, 57.87996
----------------------------------------------------------------------------