Skip to main content Accessibility help

The Hawthorne Effect in Election Studies: The Impact of Survey Participation on Voting

  • Donald Granberg and Sören Holmberg


Movements of the heavenly bodies are not affected in any discernible way by the fact that there are people on earth recording the apparent movement. Similarly it is almost inconceivable that the planets would alter their orbits because of Kepler's discovery and publication of the laws of planetary motion. The social and behavioural sciences are different in that the objects under investigation may behave differently as a result of the research process. This is particularly true when the method involves naturalistic observation, surveys or experiments. When people behave differently because of being research subjects, this is called a Hawthorne effect.



Hide All

1 This is different from the principle of limited measurability identified by Heisenberg in physics. It holds that the position and momentum of an electron cannot be measured simultaneously. This is quite different from asserting that the characteristics or movement of an electron might be affected by the fact that it is being observed.

2 See French, John R. P. Jr, ‘Experiments in Field Settings’ in Festinger, L. and Katz, D., eds, Research Methods in the Social Sciences (New York: Holt, Rinehart and Winston, 1953),pp. 98135; Roethlisberger, F. J. and Dickson, William J., Management and the Worker (Cambridge, Mass.: Harvard University Press, 1939); Ross, John and Smith, Perry, ‘Orthodox Experimental Designs’ in Blalock, H. and Blalock, A., eds, Methodology in Social Research (New York: McGraw-Hill, 1968), pp. 333–89. Exactly what was found in the Hawthorne Western Electric Studies is a matter of continuing controversy in the social sciences. See, for example, Franke, R. H., ‘The Hawthorne Experiments: Re-view’, American Sociological Review, 44 (1979), 861–7; Bramel, Dana and Friend, Ronald, ‘Hawthorne, the Myth of the Docile Worker, and Class Bias in Psychology’, American Psychologist, 36 (1981), 867–78; Parsons, H. M., ‘What Happened at Hawthorne?’, Science, 183 (1974), 922–32. This issue cannot be pursued here as our interest is only in the general possibility, which nearly everyone would concede, that people may behave differently because of participating in research.

3 See Gergen, Kenneth, ‘Social Psychology as History’, Journal of Personality and Social Psychology, 26 (1973), 309–20. This was an unfortunate choice of terms by Gergen in that it implies the sanguine view that only beneficial or progressive use will be made of published materials in the social sciences. This seems unwarranted. Presumably the knowledge of social science could also be used for nefarious purposes.

4 See Rosenberg, Milton J., ‘The Conditions and Consequences of Evaluation Apprehension’, in Rosenthal, R. and Rosnow, R., eds, Artifact in Behavioral Research (New York: Academic Press, 1969), pp. 279349.

5 See Aronson, Elliot, Ellsworth, Phoebe C., Carlsmith, J. Merrill and Gonzales, Marti Hope, Methods of Research in Social Psychology (New York: McGraw-Hill, 1990).

6 See Clausen, Aage R., “Response Validity: Vote Report’, Public Opinion Quarterly, 32 (1968), 588606.

7 This effect is analytically different from any alleged effects of canvassing by political campaigns. In the pre-election interviews in the election studies, people are not directly encouraged to vote.

8 See Kraut, Robert E. and McConahay, John B., ‘How Being Interviewed Affects Voting: An Experiment’, Public Opinion Quarterly, 37 (1973), 398406; Yalch, Richard F., ‘Pre-Election Interview Effects on Voter Turnout’, Public Opinion Quarterly, 40 (1976), 331–6; Anderson, Barbara A., Silver, Brian D. and Abramson, Paul R., ‘The Effects of Race of the Interviewer on Measures of Electoral Participation by Blacks in SRC National Election Studies’, Public Opinion Quarterly, 52 (1988), 5383.

9 See Anderson, Barbara A. and Silver, Brian D., ‘Measurement and Mismeasurement of the Validity of the Self-Reported Vote’, American Journal of Political Science, 30 (1986), 771–85.

10 See Swaddle, Kevin and Heath, Anthony, ‘Official and Reported Turnout in the British General Election of 1987’, British Journal of Political Science, 19 (1989), 537–51; Marsh, Catherine, ‘Prediction of Voting Behaviour from a Pre-election Survey’, Political Studies, 33 (1985), 642–8.

11 See Granberg, Donald and Holmberg, Sören, The Political System Matters: Social Psychology and Voting Behavior in Sweden and the United States (Cambridge: Cambridge University Press, 1988).

12 See Gilljam, Mikael, Holmberg, Sören, Asp, Kent, Bennulf, Martin, Esaiasson, Peter and Oskarson, Maria, Rött Blått Grönt: En Bok om 1988 Riksdagsval (Stockholm: Bonniers, 1990); Holmberg, Sören, ‘Election Studies: The Swedish Way’ (paper presented at a conference on the Comparative History of Election Studies at the University of Twente, the Netherlands, June 1990).

13 In all of the Swedish election studies, people who were interviewed before or after the election voted at a higher rate than people who were chosen as part of the original sample but were, for whatever reason, not interviewed. In addition to the people who were members of the panel from the preceding election, we excluded those who were assigned to be interviewed before the election but were not interviewed until after the election. Also excluded were people who agreed to participate only in a short or a very short interview. This was done to control for the fact that more time is available for interviewing after the election and a greater effort is made to interview people after the election, even if they agree to only a short interview. Even though we consider our method of exclusion to be a fairer test, the results would not be altered significantly if all the non-panel people who were interviewed in a given year were included. Incidentally, the turnout percentages were 89 and 91 respectively for those given the short and very short interviews. Thus, the turnout rate for these people was slightly lower than for people who were given the full interview but higher than for people who were not interviewed at all. The turnout for the people who were selected as part of the original sample but who were not interviewed was 80 per cent. See Granberg, Donald and Holmberg, Sören, ‘Self-Reported Turnout and Voter Validation’, American Journal of Political Science', 35 (1991), 448–59.

14 See Zeisel, Hans, Say it with Figures (New York: Harper and Row, 1968).

13 Clausen's data on the stimulus hypothesis involved a turnout of 77.7 per cent for the Survey Research Center's respondents who were interviewed before the election, compared to turnouts of 71.2 per cent and 72 per cent for the Census and Economic Survey respondents who were not interviewed before the election. If we subtract the average of the latter two from the former, the stimulus effect is just over 6 percentage points. We can then take the turnout for the post-election interviewees as the baseline and subtract it from 100 to find the maximum amount of change. If we divide the amount of change (6.1 per cent) by the maximum (28.4 per cent), the result implies that in 1964 about 21 per cent of the erstwhile non-voters in the US survey were stimulated by the pre-election interview to vote. This is similar to our estimate of the relative size of the stimulus effect in Sweden. These estimates could be net effects. That is, it is possible that some people may have become bored or upset by the political content of the pre-election survey, causing them to become demobilized, i.e. not to vote whereas they would have voted without the pre-election interview. If so, they must be outweighed by those who were stimulated to vote by being asked the battery of pre-election interview questions.

16 Clausen, , ‘Response Validity: Vote Report’, p. 604.

17 While people were randomly assigned to be interviewed before or after the election, they were categorized as high or low in political interest on the basis of self-selection. This could introduce a bias in that voting or not voting could have an effect on one's level of political interest. Bem's self-perception theory holds that people infer their own psychological characteristics (attitudes, or in this case, level of political interest) from observing their own behaviour (see Bern, Daryl R., ‘Self-Perception: An Alternative Interpretation of Cognitive Dissonance Phenomena’, Psychological Review, 74 (1967), 183200). If this applies here, people might infer high interest from voting and low interest from not voting. This implies the hypothesis that, compared to voters interviewed before the election, voters interviewed after the election would express a higher level of interest. At the same time, compared to non-voters interviewed before the election, non-voters interviewed after the election would express a lower level of interest in politics. The evidence, however, shows that the distribution on the interest question is nearly the same for voters interviewed before and after the election and for non-voters interviewed before and after the election.

18 See Lazarsfeld, Paul, Berelson, Bernard and Gaudet, Hazel, The People's Choice: How the Voter Makes Up His Mind in a Presidential Campaign (New York: Columbia University Press, 1944).

19 While our interpretation is plausible and consistent with the facts, it is by no means the only one that could be made. We cannot claim to have direct knowledge of the underlying psychological process which occurs as a result of being interviewed during the pre-election period. The most harsh interpretation of the stimulus effect might be that people somehow feel intimidated, coerced or frightened into voting out of a feeling that their behaviour is going to be under continuing surveillance and monitoring in the future. We do not favour such an interpretation and do not think it applies to the Swedish election studies we have analysed. We point it out here merely to indicate an awareness of our ignorance of the underlying process associated with the stimulus effect.

20 See Sherman, Steven J., ‘On the Self-Erasing Nature of Errors of Prediction’, Journal of Personality and Social Psychology, 39 (1980), 211–21; Greenwald, Anthony, Carnet, Catherine, Beach, Rebecca and Young, Barbara, ‘Increasing Voting Behavior by Asking People if They Expect to Vote’, Journal of Applied Psychology, 72 (1987), 315–18.

* Center for Research in Social Behavior, University of Missouri–Columbia; Department of Political Science, Göteborg University. This Note was written while the first author was Waernska Visiting Professor at Göteborg University. The authors wish to thank Martin Bennulf, Mikael Gilljam, Peter Esaiasson and Thad Brown for their advice and assistance.


Altmetric attention score

Full text views

Total number of HTML views: 0
Total number of PDF views: 0 *
Loading metrics...

Abstract views

Total abstract views: 0 *
Loading metrics...

* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.

Usage data cannot currently be displayed