Patrimony, class, and participation: French elections (1988–2012)

Abstract

Perhaps no variable in political science has received more attention than electoral participation. Despite, or because of, the great amount of research on this topic, significant new determinants for electoral participation have not been recently forthcoming. Here we offer a significant new determinant – patrimony. For theoretical reasons, we expect a voter’s wealth portfolio to influence their turnout. We test these ideas using a battery of asset items asked in French election surveys. Our finding is that voters who possess more patrimony are more likely to vote. The strength of this patrimonial effect rivals the influence of the strongest traditional predictors, such as political interest. Finally, patrimonial effects seem able to subsume, even replace, the traditional class effects from income, occupation, and education. This pivotal empirical, as well as theoretical, role of patrimony suggests serious revision in the explanation of how socioeconomic forces shape electoral participation.

Appendix

Data sources

Four datasets, one for each election year, were used. For 1998, 1995, and 2002, datasets come from the Centre d’étude de la vie politique française (CEVIPOF). For 1988 and 1995, the survey was face-to-face and post-electoral, carried out between May 9 and May 20, 1988 (N = 4032) and between May 8 and May 23, 1995 (N = 4078). For 2002, the dataset is a two-wave panel. The first wave was pre-electoral and administered face-to-face between April 8 and April 20. The second wave, administered by telephone, was carried out between May 15 and June 3 after the second round of the presidential election. Only participants in both waves are included (N = 1822) because the questions about the patrimonial items were asked in the first wave only. For 2012, the TOLUNA survey was performed online and carried out after the second round of the presidential election between May 10 and May 24 (N = 3345).

Description of variables

Participation Respondents declaring they voted on both rounds are coded (1); those who voted on only one of the rounds are coded (0.5); those declaring they did not vote on any of the rounds and otherwise are coded (0).

Age Age of respondents rescaled on a continuous scale from (0) to (1).

Gender Male is coded (1); female is coded (0).

Marital status Responds are coded (1) if they are married or in a marital relationship; (0) otherwise.

Education Eight-point scale (no diploma; elementary school diploma; professional aptitudes certificate (CAP); professional teaching certificate (BEP); technical or professional baccalauréat degree; general baccalauréat degree; baccalauréat + 2; post-graduate degree), rescaled from (0) for lowest education to (1) for highest education.

Religion Coded (1) for practicing Catholics (going to church at least once a month); (0.67) for non-practising Catholics (rarely or never going to church); (0.33) for other religions; and (0) for no religion.

Working status Coded (1) for full-time or part-time workers; (0) otherwise.

Managers Coded (1) for farmers, merchants, crafts people, company directors, company executives or professionals; (0) otherwise.

Blue-collar workers Coded (1) for qualified or non-qualified blue-collar or agriculture workers; (0) otherwise.

Income Ten-point (1988, 1995), nine-point (2002), and eleven-point (2012) salary scales, rescaled from (0) for lowest income to (1) for highest income.

Private sector Coded (1) for independent workers, company directors or private firm employees; (0) otherwise.

Patrimony Coded on a four-point scale representing the number of patrimony assets held (among a business, stocks, rental properties, a home or apartment, a country house, or savings) and taking the following values: 0 (respondents holding no patrimony items), 0.20 (one asset), 0.40 (two assets), 0.60 (three assets), 0.80 (four assets), and 1 (respondents holding 5 patrimony items).

Ideology variables Dummy variables coded from a seven-point left-right self-positioning scale, extreme left = (1) when scale response = 1; left = (1) when scale response = 2 or 3; centre = (1) when scale response = 4; right = (1) when scale response = 5 or 6; extreme right = (1) when scale response = 7, and (0) otherwise for all variables.

Interest Coded (0) when respondents have no interest in politics; (0.33) when they have little interest; (0.67) when they have some interest; and (1) when they have much interest. Don’t know answers and refusals are coded as missing observations.