Abstract
Age is often found to be associated with a plenitude of socioeconomic, politico-administrative, biological and thanatological variables. Much less attention has been paid by scholars, however, to explaining ‘age’. In this paper we address this unfortunate scientific lacuna by developing a model of ‘age’ as a function of several factors suggested by (post)rational choice and social constructionist theories. Using state-of-the-art multilevel statistical techniques, our analysis allows the determinants of age to vary with the institutional characteristics of European countries. Our findings convincingly show that generalized trust in strangers, support for incumbent extremist political parties in provincial elections held in the month of January, and the percentage of overqualified women in the cafeterias of national parliaments are all statistically significant explanations of ‘age’. Our findings have obvious implications for conspiracy theorists, organizational advisors, spin doctors and ordinary charlatans.
Similar content being viewed by others
INTRODUCTION: THE COMING OF AGE OF THE STUDY OF AGE
Age and ageing concern most discrete non-autopoetic biological systems, humans included. Surprisingly, however, social scientists have only rarely attempted to explain variation in age in terms of individual, system-level and in-between variables. Age figures prominently in almost all quantitative research but, unfortunately, it features almost invariably on the right-hand side of regression equations. In this paper weFootnote 1 propose to turn the tables (and the equations) and ask the question: what explains age? Given that most studies in the fields of sociology, political science, education and psychology report significant age effects, it is even more important to account for the determinants of age.
The study of age has been hampered for decades by the untested assumption that age is a linear function of time. If this is the case, why is age so closely related with so many socioeconomic, politico-administrative, biological and thanatological variables? Moreover, scholars have established that age varies systematically with education level, wages, ear hairiness,Footnote 2 etc. Clearly, something else is going on. More recently, ((de)re)constructionist theories have begun to address this empirical puzzle by uncovering the radically subjective character of age. In addition, in an important but rarely acknowledged piece, Wuffle et al (1997) discover that certain institutions (like the US House) can have a prohibitive impact on age, and the associated phenomenon of death. This paper purports to contribute to this emerging scholarship by constructing a pooled multilevel nested hierarchical mixed time-series cross-sectional model and unearthing a set of significant explanatory causes of age, all of them suggested by the theoretical model introduced in the following section.
THEORY: (POST)RATIONAL VERSUS ((DE)RE)CONSTRUCTIONIST EXPLANATIONS OF AGE
‘Following best practices … we pit two equally ludicrous theories against each other and will eventually conclude that aspects of both theories are important and have to be reconciled in future research’.
Following best practices in contemporary scholarship in political science, international relations and public administration, we pit two equally ludicrous theories against each other and will eventually conclude that aspects of both theories are important and have to be reconciled in future research. We derive hypotheses from both theories, which we subsequently test against empirical data. As is usual in economics, in case we find a discrepancy between the theories and reality, we will have to declare reality faulty and try to bring it in line with our theory.Footnote 3 Fortunately, the rate of un-hypothesized findings in the social sciences rapidly approaches zero.
The method through which we deduce hypotheses from the theories is highly reliable, transparent and replicable under strict medical supervision: it consists of a week of sleep deprivation followed by mild electrical shocks administered to the brain of the researcher. Following this procedure we were able to identify three hypotheses as indicated below.
First, we will discuss the relevance of (post)rational theory for explaining age. (Post)rational theory is closely related to post-rational theory (cf. Wuffle, 1999), but it differs importantly in that the rational component of the former is bounded from below while the rational component of the latter is bounded from above (see Wuffle, 1984). The intellectual predecessor of both these approaches – rational choice theory – is based on a simple, but powerful assumption: people do what they want. Following this fundamental insight about human nature, scholars have developed two approaches for applying the theory to yield an ever-deeper understanding of reality. Using the first approach, the scholar assumes that heFootnote 4 knows what people want and proceeds to check whether the expectations match with reality. In case of a mismatch, either reality is proven wrong (see above), or the finding is declared a ‘paradox’ and begins to generate volumes of subsequent scholarship. In the second version, the so-called ‘revealed preferences’ approach, scholars deduce what people want from peoples' behaviour. Then they proceed to declare that the peoples' behaviour is entirely rational given what they want. An example of an application of rational choice theory to the problems of age and ageing includes Bates et al (2004), who discover that the institution of kinship exists to protect the ageing from being done away with by the young.Footnote 5
(Post)rational theory posits that age is a result of utility maximization subject to cognitive constraints. We can immediately recognize that individual (post)rationality will lead to opportunities for free-riding in which some agents exploit the myopically rational behaviour of others and misrepresent their true age preferences. Furthermore, to make the model more realistic, we incorporate hyperbolic discounting in the utility functions, which leads to the result that some people prefer a 50 per cent chance of 20 years ageing to a 90 per cent chance of ageing with 2 years only.Footnote 6 The proof of these propositions (or shall we says ‘lemmas’?) is contained in an appendix to be found somewhere on the Internet.
H1:
-
Rational hyperbolically discounting actors lead to lower probability of age, Ceteris paribus.
Contrary to rationalist theories, social constructionism claims that people do not even know what they want and even if they wanted something there is no way that we could objectively know. Obviously this worldview makes for a rather straightforward research strategy – immersing ourselves in the bodies of others and seeing the world through their eyes. Repeated iterations of this procedure can extend the approach to strategic interaction situations. (De(re))constructionist theorists emphasize that reality is socially constructed and therefore amenable to change if enough people focus for enough time on the same aspect of reality (Sokal, 1996).
The implication of this theory for age is that age is socially constructed, too. Therefore, trust in strangers is important for explaining age. Critics who argue that trust and social construction are not related should be reminded that everything is (non-linearly) related to everything else according to the new-constructionist synthesis of (de)constructivism with (re)constructivism. Furthermore, trust is such a woozy, soft and fuzzy topic that it obviously fits with the constructionist theory.Footnote 7
H2:
-
More trust leads to higher probability of age, Ceteris paribus.
So far we only discussed individual-level explanations of age. Age is, however, a multi-faceted and multilevel phenomenon that is affected by institutions like the state. In other words, we need to bring the state back in the study of age. Age is expected to vary not only between individuals, but between countries as well: some countries are simply of a different age than others.
H3:
-
Institutions in different countries cause different age, focus mocus.
EMPIRICAL ANALYSIS: EXPLAINING, PREDICTING AND DETERMINING AGE
Regression is Progress. (Unknown)
In order to test the hypotheses and discover the causes of age,Footnote 8 we conduct a statistical analysis based on state-of-the-art statistical techniques. Before we turn to the results, however, we were forced by an anonymousFootnote 9 reviewer to write a few words about our data and how they are operationalized. In short, we use the two waves of the European Social Survey to test our predictions. Our dependent variable – age – is available for all respondents from both waves of the survey. First we take the natural log of the raw data and then conduct exponentiation. We also considered transforming the variable into a binary indicator but the statistical programme we use crashed under this specification, so we do not report these results. We are willing to bet a substantial sum, however, that the substantive results would not change if we treat age as a ‘yes’/‘no’ category.
The operationalization of trust is also straightforward. The survey asks the question ‘Do you think that in general strangers can be trusted or you can never be too careful and should always carry a gun?’ The answer to this question is based on a 16.4-point scale ranging from complete distrust to complete trust. For ease of exposition, we split the respondents into two groups. People who tended to distrust strangers ended in the category Aggressive Solitary Sociopath (ASS), while those who scored more than eleven ended up in the category Douchebag Hippie (DUGH).
‘We are willing to bet a substantial sum, however, that the substantive results would not change if we treat age as a “yes”/“no” category’.
With regards to the operationalization of H1 we combine four variables into a single factor, which we subsequently use in the analysis. The original four variables are (1) support for incumbent extremist political parties in provincial elections held in the month of January; (2) eye colour; (3) SAT scores; and (4) weekly hours spent watching Rupert Murdock-owned media. Due to the obvious co-linearity between the variables, we conducted a factor analysis to combine the information they bring. We retained only the first and most important factor that accounted for 23 per cent of the variance. The factor, which we label FACTOR 1, is positive related to all original variables with three exceptions. Due to data availability, we used to have missing values for this variable in approximately 71.45 per cent of the cases. We remedied the situation by assigning values to cases using the astragaliFootnote 10 method.
Finally, with regard to the influence of country-level factors, we make use of an original data set disseminated recently by a geographically balanced EU-wide consortium of scholars financed by the EU 6.5 Framework Program. The data set contains information on the percentage of overqualified women in national parliaments' cafeteria for all European states. Surprisingly even to us, we are the first to use this data set. In addition we use 162 control variables. These are VAR001–VAR159.
Table 1 presents the results from the analysis. As we mentioned on every page of this paper, we use a sophisticated state-of-the-art pooled multilevel nested hierarchical mixed time-series cross-sectional model that contains individual and county-level predictors of age. We do not report any measure of fit, but be assured that the model explains a significant part of the variation in age because otherwise we would have presented a different model.
All the variables that we use are statistically significant and in the expected direction. To repeat for our more easily distracted readers: All the variables are statistically significant and in the expected direction. Therefore, our hypotheses are all confirmed.Footnote 11 This is, of course, nothing new since we deduced the hypotheses from true assumptions using truth-preserving analytical techniques. That made us wonder for a moment what the point of the testing was, but our doubts in the approach were immediately dispersed after we checked that we had not strayed from standard social science practices.
The results of the analysis unequivocally show that trust in strangers, FACTOR 1 and the country-level variable are significant causes of age.Footnote 12 Therefore, by using regression we have made significant progress towards explaining why age happens and how it changes over time.
CONCLUSION: AGE IN THE FUTURE
‘We do not report any measure of fit, but be assured that the model explains a significant part of the variation in age because otherwise we would have presented a different model’.
This paper has made a ground-breaking contribution to the study of age. We have also raised the methodological standards in the field by employing three variables on two levels plus a large number of controls. Our trail-blazing approach can be fruitfully applied to other uncharted territories in the study of social life, like explaining gender. We trust that we have showed that rigorous theorizing coupled with state-of-the-art statistical modelling can produce knowledge of immense scientific and practical value serving as an inspiration for generations of future scientists to come.
Notes
The fact that I use ‘we’ instead of ‘I’ has nothing to do with the allegations that large parts of this paper have been written by my student assistants. ‘We’ is only used to implicate that more than one person agrees with the statements of the paper, to diffuse responsibility in case of obvious mistakes, and to nourish a warm feeling of self-aggrandizement.
Some scholars have hypothesized that the last relationship is mediated by gender but the empirical evidence is inconclusive.
Most economic references will do, but see in particular, the work on optimal currency areas.
Before we are accused of discrimination by the female part of the profession, let us point out that rational choice is obviously a manly (cold, calculating, rigid, unemotional and cynic) approach that leads to the fact that all its practitioners are male. The few counter-examples only prove the point and have not been confirmed by independent hormonal tests.
This finding severely undermines the traditional fairy tale interpretation that ageing people are kept at home due to their abilities to locate grain in ant nests in times of bad harvests.
This result has been confirmed by numerous psychological experiments that show that normal people cannot apply the Lagrange multiplier procedure to multidimensional constrained optimization problems.
Reviewers of this paper pointed out that we could not possibly attempt to discuss social constructionism and not mention discourses. We agree, so here we go: ‘Discourses!’
Readers who are interested in the intellectual predecessors of this approach to causal analysis can consult Ziliak and McCloskey (2008: 106–107), but are strongly advised to avoid the rest of the book like the plague.
You coward, we gonna get you sooner or later!
The time-honoured astragali method refers to tossing ankle bones of sheep or any other cloven-hoofed animals (Everitt, 1999).
A few people continue to insist that we cannot conclude that the hypotheses are true from the analysis because the probability of the data, given the hypothesis is not the same as the probability of the hypothesis given the data. This claim is so confusing that it must be Bayesian. Let us remind everyone that Bayesian thinking has been officially declared by the UK Court of Appeal to venture into ‘inappropriate and unnecessary realms of theory and complexity’ (Regina v Adams, 1996).
David Freedman (1997) argues that we cannot make causal claims on the basis of regression applied to observational data without making causal assumptions first. In his own words: ‘If you want to pull a rabbit out of the hat, you have to put a rabbit into the hat’ (p. 182). We can only pity Professor Freedman, who has obviously never been to a magician's performance.
References
Bates, R., Greif, A. and Singh, S. (2004) ‘The Political Economy of Kinship Societies’, in I.L. Morris, J.A. Oppenheimer and K. Edward Soltan (eds.) Politics from Anarchy to Democracy. Rational Choice in Political Science, Stanford, CA: Stanford University Press, pp. 66–88.
Everitt, B.S. (1999) Chance Rules: An Informal Guide to Probability, Risk, and Statistics, New York: Copernicus.
Freedman, D. (1997) ‘From Association to Causation via Regression’, in V.R. McKim and S.P. Turner (eds.) Causality in Crisis? Statistical Methods and the Search for Causal Knowledge in the Social Sciences, Notre Dame, IN: University of Notre Dame Press, pp. 113–182.
Sokal, A. (1996) ‘Transgressing the boundaries: Towards a transformative hermeneutics of quantum gravity’, Social Text 46/47: 217–252.
Wuffle, A. (1984) ‘Should you brush your teeth on November 6, 1984?: A rational choice perspective’, Political Science and Politics 17 (3): 577–580.
Wuffle, A. (1999) ‘Credo of a “reasonable choice' modeller’, Journal of Theoretical Politics 11 (2): 203–206.
Wuffle, A., Brunell, T. and Koetzle, W. (1997) ‘Death where is thy sting? The Senate as a Ponce (de Leon) scheme’, Political Science and Politics 1: 58–59.
Ziliak, S. and McCloskey, D.N. (2008) The Cult of Statistical Significance. How the Standard Error Costs Us Jobs, Justice and Lives, Ann Arbor, MI: University of Michigan Press.
Acknowledgements
Drafts of this paper have been read aloud at the LXXIV Annual Meeting of Social Scientists with Large Research Budgets (Bali, Indonesia, Winter 2009), the First and Probably Last Conference of Young Gerontometricians, and at my brother's wedding. I acknowledge all moronic suggestions, vindictive personal attacks and malicious rumours advanced at these occasions. I am also extremely grateful to the current double-blind to nonsense peer-review system that can let such papers through. The usual disclaimer applies. (The usual disclaimer means that in the unlikely case that somebody is as bored and vicious as to attempt to replicate the analysis, I will transfer all responsibility for tampering with the data to my student assistants.)
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Bezimeni, U. Determinants of Age in Europe: A Pooled Multilevel Nested Hierarchical Time-Series Cross-Sectional Model. Eur Polit Sci 10, 86–91 (2011). https://doi.org/10.1057/eps.2010.12
Published:
Issue Date:
DOI: https://doi.org/10.1057/eps.2010.12