Contacting as a channel of Political Involvement: Collectively Motivated, Individually Enacted

Abstract

Citizen-initiated contact with politicians is an increasing, but often neglected, form of political participation. Direct contact fits well with new participation trends that increased individualisation and a single-issue focus have brought forth, and is often interpreted as a participatory form that conforms with such new demands. Yet while political participation through most traditional channels is decreasing, direct-contact increase implies that people are still channelling involvement through the established institutions. Accordingly, this article argues that the significant increase in direct contacting of representatives is not an expression of protest behaviour. On the contrary, contacting shows strong adherence to representative democracy. It is related to conventional modes of political participation, above all party-related activities. This study examines contacting in relation to other forms of political involvement, using data from the Norwegian Citizenship Survey. The analyses reveal that even after removing from the analyses those who themselves hold public office, citizen-initiated contacting is related to party activity. Political ties are more important in explaining contacting than is the socio-economic status of the contactors.

Acknowledgements

This article was first presented at the ECPR Joint Sessions of Workshops in Uppsala, 13–18 April 2004. We will like to thank the participants at this workshop, in addition to Gunnar Grendstad, Audun Offerdal, and two anonymous reviewers, for valuable comments.

Notes

1. To participate as political consumers, i.e. in political consumerism, involves using one's powers as a consumer in the market to try to influence politics, e.g. to boycott certain products or to buy certain products on the basis of political, ethical or environmental concerns. For a discussion, see Andersen and Tobiasen (2001), Micheletti (2003), Micheletti et al. (2004), Sørensen (2004).

2. However, Verba et al.'s conclusion is in contrast to the findings of Rosenstone and Hansen (1993: ch. 3), who report a decline in the proportion writing to congress.

3. The Citizenship Survey comprised 5,000 persons living in Norway; 46.5% answered the questionnaire (N = 2297). The survey was part of the Norwegian Power and Democracy research project (1998–2003), and is also part of the ESF network on ‘Citizenship, Involvement, Democracy’ (CID). MMI Univero has been responsible for the survey, and the Norwegian Social Science Data Services (NSD) has prepared the data set for analysis. The Power and Democracy project, MMI and NSD are not responsible for the analyses or interpretations made here.

4. Use of dummy variables for this purpose is uncontroversial since factor analysis is based on the correlation matrix.

5. Eighty-eight per cent of the survey respondents who were entitled to vote reported they did so in the last national election. Voting did not load on any of the factors until we specifically asked for six factors. It then loaded on three of the factors – positively on the direct action and contacting factors and negatively on the civil disobedience factor.

6. We have chosen an explorative procedure in which we first asked for one single factor, and successively increased the number of factors asked for until we found the most appropriate number of dimensions. The procedure is based on an analysis of which variables always load on the first, the marginal and the intermediate factors (see Grendstad and Rommetvedt 1996; Hair et al. 1995). The most important criterion, however, is that the dimensions are theoretically sound, i.e. that the data-reduction procedure combines variables in a way that makes sense theoretically. Based on these considerations, we find the material to be best represented by a six-factor solution (where all the factors fulfil the latent root criterion, with an Eigenvalue higher than 1.0).

7. The variable simply measures whether one has contacted a politician or not during the past 12 months. When we exclude this variable from the factor analysis, the explained variance increases slightly, to 55.4%.

8. This conclusion is supported by Andersen and Hoff (1995; see also Strømsnes 2003). The dispute about political participation as cumulative or dispersed is also present in the long-lasting debate about Norway as a ‘distant democracy’ (see Lafferty 1981; Martinussen 1973; cf. Strømsnes 1993).

9. Place of residence refers to the respondents' place of domicile, not home or apartment.

10. However, in Bowling Alone, Putnam's approach is more micro-oriented (Putnam 2000). Our analysis differs from Putnam's in that he mainly focuses on horizontal trust, or social trust between fellow citizens, while we are interested in vertical trust, i.e. the degree to which citizens trust politicians and political institutions (cf. Newton 1997). However, according to Rothstein (1998), horizontal and vertical trust may connect since trust in other people is often the product of trust in institutions (see also Wollebæk 2000).

11. See Pharr and Putnam (2000) for a discussion of the paradox that an increasing number of people in the most successful democracies express distrust in their governments.

12. The questionnaire also included questions about trust in the courts, the police, the European Union (EU) and the United Nations (UN), but they are seen as less relevant here. Neither do they correlate as strongly as do the other variables.

13. Zuckerman and West (1985) use campaign activity as a measure of political ties. They admit that party membership is the most ‘intuitively plausible indicator’, but their data did not allow them to employ this measure.

14. Since our dependent variable, contacting a politician, is a dummy variable, we have performed logistic regression. Because education and place of residence are categorical variables, they are introduced in the model as a set of dummy variables, using the first value as a reference category. Party membership is dichotomous. As for the coding of the remaining background variables (gender, age, income, married, working), see the Appendix. The coding of the attitudinal variables has been accounted for in the presentation of Table 5, except for Political interest, which is recoded as a dummy variable in the regression analysis.

15. One obvious way of enhancing the predictive power of the model would be to include office-holders, but, as we have argued, this would be incompatible with our ambition to study contacting as a channel for citizens to influence political outcomes. We have also tested a second model in which all non-significant variables were excluded. The exclusion of insignificant variables did not increase the explanatory power of the model, and, furthermore, it did not alter any of the effects on the significant variables observed in Table 6. Therefore, we have chosen to present the full model.