Cohort effects, age effects, and period effects in the participation in the arts and culture in Denmark 1964–2004¹

1. An early version of this paper was presented at the Fourth International Conference on Cultural Policy Research, iccpr 2006, in Vienna, July 2006 (Bille 2006).

Abstract

The purpose of this paper is to analyse the Danish population's participation in cultural activities during the period 1964–2004 with special regard to cohort effects, age effects and period effects. The research is based on data collected in a representative survey of participation in arts and culture in Denmark from 2004, and data from previous similar studies carried out in 1964, 1975, 1987, 1993 and 1998. By using the ‘quasi‐panel structure’ of the total dataset from different survey years, cohort analyses are used to describe the development in participation for selected cultural activities for different generations during the last 40 years. It is shown how age effects, generation effects and period effects can be separated to get a more detailed picture of the development in participation. This more advanced statistical approach will lead to better and more precise cultural policy indicators of participation.

Keywords:

• participation studies
• consumption of culture
• cohort analysis
• age effects
• development in participation

Acknowledgements

I am grateful to research director Eskil Heinesen, akf, Institute of Local Government Studies, Denmark, for helpful comments, and to research assistants Peter Bildtoft and Erik Wulff for datawork.

Notes

1. An early version of this paper was presented at the Fourth International Conference on Cultural Policy Research, iccpr 2006, in Vienna, July 2006 (Bille 2006).

2. As Schuster (2006) points out, ‘participation’ should be interpreted as ‘attendance’, since ‘participation’ requires a more active act. In Denmark and in the other Nordic countries as well, the often used term for these kind of studies are ‘cultural habits studies’.

3. Participation studies must be distinguished from audience studies. Audience studies are typically based on distributing questionnaires to audiences in a cultural institution and collecting them upon departure, while participation surveys are designed to randomly sample a broader population, including those who are not attending.

4. While participation studies in general seek to characterize the demographics of those participating or attending, there also exists an economic literature within the field of cultural economics, applying econometric studies which seek to quantify price and income elasticities in the arts (for an overview of this economic literature see Seaman 2006).

5. The terms ‘generation effects’ and ‘cohort effects’ are used synonymously.

6. TNS Gallup handled the practical collection of data.

7. In true panel data the same people are followed over time.

8. Multivariate logistic regression is used. See for instance Bille Hansen (1996) for a detailed description of this type of model and its application.

9. In an earlier version of this paper (Bille 2006), some additional variables were included in the estimations on data for 2004. These additional variables were information about where the respondent lived (measured by degree of urbanization and region of Denmark), ethnicity (Danish/Western/non‐Western), whether the respondent had a physical handicap, and last, but not least, lifestyle‐type (measured by a number of questions about the respondents' attitudes and values to place them in one of nine segments). Including these variables in estimations for 2004 caused slightly different estimation results. These variables are, however, not included in the estimations in this paper, because it was not possible to trace them back to 1964, and therefore they could not be included in the estimations of the cohort‐models on the full dataset.

10. Seaman (2006) discusses among other things the empirical evidence of the often repeated result that the role of education is much stronger than the role of income in explaining participation in the arts and culture. Likewise, the empirical evidence of the explanatory role of ‘life‐style’ factors compared with traditional socio‐economic variables is discussed.

11. A lot of these variables will be correlated, especially education and income. For a discussion of the difficulties of multi‐colinearity, see Seaman (2006).

12. In the cohort analysis using the full data set from 1964–2004 the gross income has been deflated by the consumer price index.

13. Unless interactions are included between variables, but this is not done in this model.

14. Other studies have shown, that in the case of movies, the average ‘explanatory power’ of income and education was equal (Seaman 2006, p. 422), while in many other cases of cultural participation, studies have shown that the role of education is much stronger than that of income in explaining participation.

15. Online newspapers and free newspapers may to some extent be said to have taken over part of the market for traditional newspapers in recent years. However, these alternatives cannot make up for the drop in the reading of traditional newspapers. In 2004, 64% of the population either read newspapers, online newspapers or free newspapers on a near daily basis, which is still lower than the proportion of 68% who exclusively read newspapers on a near daily basis in 1998.

16. On the other hand, online newspapers tend to have more younger readers and the age group 30–49 has the highest proportion of regular online newspaper readers (30–35% of the 30 to 49‐year‐olds read online newspapers on a near daily basis or almost every week) (Bille et al. 2005).

17. This model is part of a larger literature on the taste cultivation problem and human capital. Different explanations have been proposed: habit formation, learning by consuming and rational addiction. For an overview of the literature in this area, see Seaman (2006).

18. If we instead show the proportion of the population in different generations that visit the library at least once a month, a different picture emerges. After 1993 we see a drop in the proportions who visit the library at least once a month.

19. Data for library visits are not available from 1964.