Abstract
In the first half of the essay we summarise the main contributions in the essays in this issue by Coleman, Colomer, and Taagepera, and identify key commonalities in their suggestions for making political science more ‘scientific’. We argue that most of their concerns are well taken, but that the remedies they propose may not be applicable in all domains within political science. In particular, it is largely in the area of voting and elections, where clearly demarcated input and output variables can be identified, that their suggestions seem the most applicable. In the second half of the essay we trace the rise and fall, over the past 100 years, of movements in the US to make political science more scientific. We conclude by identifying similarities between these essays and the recent EITM movement in US political science.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
2 These points are emphasised in the Coleman essay in this volume.
3 These points are especially emphasised in the Taagepera essay in this volume.
4 See the Coleman essay in this volume.
5 For example, if an underlying distribution is roughly lognormal, then we may expect the average outcome to be the geometric mean of the distribution's minimum and maximum values.
6 In such a case, trying to incorporate potential confounding variables into a statistical model may actually make things worse; since not all factors are being included, the ones that are included may mask the fit of the key variables specified in the underlying probabilistic theory. See the Taagepera essay in this volume.
7 Making sense of human actions and human institutions is a very difficult task indeed. (Nature left all the easier problems to the physicists – which is saying a great deal, considering how difficult those problems are.)
8 I might note in this context that the warnings in King et al (1994) against the reliability of inferences drawn from natural experiments are in my view, much overstated. An aphorism by my long time office-mate, A. Wuffle, 1 April, 2000, personal communication is relevant: ‘Making causal inferences using cross-sectional data is like trying to tell time with a stopped watch; if you get it right it only will be by accident’.
9 Consider, for example, ‘Wagner's Law’ in Economics that asserts that (central) government spending as a proportion of GDP rises with per capita real income. In countries throughout the world this ‘law’ is well supported. For example, in the US over the period from 1922 to 2002, the correlation between these two variables is .98 (Winer et al, 2006). But knowing that government share of the economy rises with growing affluence (albeit at different rates in different countries) does not tell us why. Does it have to do with changing technologies of tax collection, for example, the more affluent the society the easier is it to locate income to tax? Does it have to do with psychological willingness to be taxed, such that people in more affluent societies are further away from subsistence and thus protest the greater taxation burden less? Does it have to do with changing preferences, for example, post-materialist values, that lead to new types of demands for government services? Or, perhaps, is it merely a reflection of the greater complexity and inter-connectedness of the modern world, where new medical and scientific possibilities are open to us, and where externalities can have major consequences and thus may require more coordination/government action to deal with? Or is it simply that government is what economists call a ‘normal good,’ and more of it is sought as incomes rise? (For further discussion, see Holcombe, 2005).
10 Understanding the mechanisms which drive observed empirical relationships allows us to refine and extend our theories.
11 I suspect that economists would generally do better than most political scientists in identifying non-linear relations since so many economic models involve multiplicative or logarithmic or power functions. I also suspect that political science scholars whose methodological training involves exposure to econometrics, would generally do better in formulating such models than scholars given more traditional statistics training.
12 I am struck, too, by the fact that the sciences that are the most precise and the most successful simply do not seem to worry that much about the kinds of statistical issues that obsess political scientists.
13 It has also been a main area of study for two of the other three authors in this mini symposium.
14 It is as if political science is still at the stage of medicine in the days of Aristotle and Hippocrates, when physicians sought to manipulate the ‘four humours’ of the body (yellow bile, black bile, phlegm, and blood). At the time it was thought that each humour was associated with a particular season (Summer, Autumn, Winter and Spring, respectively), and with particular moods. For example, according to Aristotle, black bile is associated with melancholy, and can produce an alternation of depression and anger (Zimmerman, 1995).
15 In humorously illustrating his point, Taagepera talks about a mythical stone age physicist only being able to say that ‘Rocks fall when dropped from a tree,’ while, by the Enlightenment, a physicist would be able to say that ‘Rocks fall when dropped from a tree at a speed proportional to the square of the time elapsed and to the gravitational pull of the earth, with the latter a near constant quantity that can be precisely measured’.
16 We would be a little better off if we were to try to use a concept such as racially polarised voting as an explanatory variable, but then we would need a theory that links the concept to something we are trying to explain.
17 That is not to say, however, that I share the view that the ‘quiddity’ of events and persons makes the search for generalisations impossible. Space does not permit a general discussion here of philosophy of science issues. Suffice it to note: (a) that theories have parameters which allow us to specify predictions that will change depending upon circumstances; (b) to say that theories have a limited domain of applicability or are context specific is not to say that the search for theory is futile, and (c) that most of the epistemological debate about the extent to which a science of politics is possible (or about the utility of particular approaches to building theory, such as rational choice modelling) is couched at such a high level of abstraction that it is meaningless. The proof of the pudding is in the eating, that is, if we want to argue that a given approach is the best way to understand some phenomenon, then we ought to study that phenomenon and show what we do helps us explain it better than any competing approach. In general, I might also note that my own views about how to think about social science are very close to those recently expressed by Elinor Ostrom (2006). She emphasises the complexities of human behaviour and symbol systems, praises interdisciplinary work, encourages multiple methods, ridicules internecine war about which methodology is the exclusive royal road to truth in the social sciences, and allows for the value of methodologies that may fail disastrously in some domains but help explain some phenomena very well.
18 Work on which I am a junior author (Schneider and Grofman, 2006) follows up on ideas in Ragin (1987) to look at ways graphically to represent equifinal relationships and to study them.
19 However, it is important to note that while the basic linear model is used throughout King et al (1994) to illustrate their methodological and epistemological points, that choice of model was simply to not overwhelm the non-statistically trained reader (Gary King, personal communication, 25 May 2006). King himself has been a developer and advocate of far more sophisticated models that do directly address many (and perhaps all) of the issues raised by the present symposium authors, such as the importance of boundary conditions (see, e.g., King, 1997).
20 Statistical models that have recently been developed to study multiple causation involving dichotomous or ordinal variables whose potential seems to me to be quite high (such as Braumoeller, 2003) build on early pioneering work by the econometrician Dale Poirier (1980).
21 In asserting this stylised fact about the development of US political science, we draw on Somit's and Tanenhaus’ (1967) magisterial history of political science in the United States, updating it with our own observations about political science during the second half of the twentieth century and into the twenty-first century.
22 Each new wave in part seeks to be responsive to the perceived limitations of previous research. However, old waves never really die, rather their most important contributions come to be incorporated into ‘political science as usual’.
23 The equivalent capsule history of European political science dealing with the themes I discuss below will have to be written by someone more knowledgeable than me about developments in Europe.
24 Because of space limitations we will not be able to discuss most of the cells within this table.
25 There are other dimensions of potential conflict we could also identify that do not as directly involve a debate about the degree to which the discipline ought to be ‘scientific’ and what ‘being scientific’ means, for example, between those who see political science as limited to the study of government and those who see politics as visible everywhere, from families to churches to playgrounds (and perhaps also to chimpanzees). Elsewhere (Grofman, 1997) we have identified seven different potential dimensions of cleavage within the discipline.
26 This is the ‘humanistic, empirical, and theoretical’ cell of Table 1. However, the Perestroika movement also reflected other protests against theory-driven pure scientific research as well, especially the revolt against empiricism and in favour of a return to normative theory. (Since protest movements tend to collect the disaffected, some multidimensionality is virtually inevitable).
27 As Easton (1969: 1051) characterised the post-behavioural movement of which the Caucus was representative: ‘Its battle cries are relevance and action’ (emphases in original.)
28 Somit and Tanenhaus (1967: 25) are paraphrasing the views of Herbert Baxter Adams, then the leading figure at John Hopkins, one of the two major departments in political science's very earliest days. This point of view was reflected in the Johns Hopkins journal, Studies in Historical and Political Science, of which Adams was the long-time editor.
29 Burgess was, in the very early days of the discipline, the leading political scientist at Columbia, Johns Hopkins’ then main rival as the top political science department in the United States. He became one of the early presidents of the American Political Science Association.
30 This conflict over the extent to which political science should be seen as an adjunct to law could be taken as a fourth potential dimension of conflict, between those who emphasise legal structures and those who emphasise behaviour and process. However, because the purely legalist view is of no contemporary relevance (at least in the US) we have not bothered to include it in Table 1.
31 However, Lowell, although ‘priding himself on the predictive ability of his work’, saw a ‘profound gulf’ between the ‘observational’ and the ‘experimental’ sciences. He also ‘firmly believed that ‘the ultimate object of political science is moral, that is, the improvement of government among men’ (Somit and Tanenhaus, 1967: 78).
32 For example, when APSA set up ‘a committee in 1911 to explore the subject of laboratory and field training for graduate students’, the committee renamed itself ‘The Committee on Practical Training for Public Service’ and prepared a report which ‘almost totally ignored’ its original mandate (Somit and Tanenhaus, 1967: 82).
33 Over time, not surprisingly, this civic-oriented movement itself lost steam. Like behaviouralism, but to nowhere near the same degree, however, this civic-oriented movement enjoyed some renaissance after WWII. See the discussion in Somit and Tanenhaus (1967: 87–88).
34 Of course, this approach had certainly never died; rather, we might merely say that its messianic fires had been banked.
35 Also, in the 1950s, David Truman revived and extended the ideas of Arthur Bentley on group-based politics.
36 While the bête noir of contemporary humanistically inclined political science scholars in the US is the Department of Political Science at the University of Rochester as it existed under the chairmanship of the late William Riker; before rational choice had risen to its present prominence, their humanistic counterparts of the 1960s saw ICPSR at the University of Michigan as the main enemy. It, too, was seen as trying to put more ‘science’ into political science, albeit looking for models in psychology rather than in economics (cf. Easton, 1997).
37 For example, this was a major factor in shaping the research path of my late colleague, Harry Eckstein, one of the leading comparativists of his time, whose Jewish parents had sent him to live outside Germany while he was still a child.
38 As noted earlier, the two most recent anti-science waves in American political science (the Caucus for a New Political Science and the Perestroika Movement) can be viewed in large part as reactions to the implications of the work of those influenced by Converse and Riker, respectively.
39 For example, by 1981 the annual citations to Downs (1957) exceed those to Campbell et al (1960), and the gap continues to rise (Gray and Wuffle, 2005, Table 1).
40 The rise of methodology as a subfield of political science is discussed in Masuoka et al (2006).
41 As previously noted, such a swing is sometimes triggered by the development of new research methodologies or applications (such as in game theory or econometrics) that give promise for substantial scientific advances.
42 This quote is taken from the programme description of Duke University's Summer Institute on the Empirical Implications of Theoretical Models, 14 June – 9 July 2004 (www.poli.duke.edu/eitm/), the third in the series of summer workshops funded by the US National Science Foundation.
44 Indeed, virtually all of the EITM themes are anticipated in my own previous work (see, e.g., Grofman, 1974, 1982; Pool and Grofman, 1975; Brunell and Grofman, 1998; Grofman et al, 1998, 2000; Adams et al, 2005; Regenwetter et al, 2006). Thus, like the Molière character who finds that he has been speaking prose all his life without knowing that this is what he has been doing, I find myself having been practicing EITM all my life without having known that as a name to call what I have been doing.
References
Adams, J., Merrill, S. and Grofman, B. (2005) A Unified Theory of Party Competition: A Cross-National Analysis Integrating Spatial and Behavioral Factors, New York: Cambridge University Press.
Braumoeller, B.F. (2003) ‘Causal complexity and the study of politics’, Political Analysis 11: 209–233.
Brunell, T. and Grofman, B. (1998) ‘Explaining divided U.S. senate delegations, 1788–1994’, American Political Science Review 92 (2): 1–9.
Campbell, A., Converse, P.E., Miller, W.E. and Stokes, D.E. (1960) The American Voter, New York: John Wiley and Sons.
Downs, A. (1957) An Economic Theory of Democracy, New York: Harper and Row.
Easton, D. (1953) The Political System, New York: Knopf.
Easton, D. (1969) ‘The new revolution in political science’, American Political Science Review 63 (4): 1051–1061.
Easton, D. (1997) ‘The Future of the Post-Behavioural Phase in Political Science’, in K. Monroe (ed.) Contemporary Empirical Political Theory, Berkeley: University of California Press, pp. 13–46.
Gray, M.M. and Wuffle, A. (2005) ‘Vindicating Anthony downs’, PS: Political Science and Politics 38 (4): 737–740.
Green, D.P. and Shapiro, I. (1994) Pathologies of Rational Choice Theory: A Critique of Applications in Political Science, New Haven, CT: Yale University Press.
Grofman, B. (1974) ‘Helping behavior and group size, some exploratory stochastic models’, Behavioral Science 19: 219–224.
Grofman, B. (1982) ‘A dynamic model of protocoalition formation in ideological n-space’, Behavioral Science 27: 77–90.
Grofman, B. (1997) ‘Seven Durable Axes of Cleavage in Political Science’, in K. Monroe (ed.) Contemporary Empirical Political Theory, Berkeley: University of California Press, pp. 73–86.
Grofman, B., Brunell, T.L. and Koetzle, W. (1998) ‘Why is there sometimes midterm gain in the Senate but (almost) always midterm loss in the house?’ Legislative Studies Quarterly 23 (1): 79–89.
Grofman, B., Koetzle, W., McDonald, M. and Brunell, T. (2000) ‘A new look at split ticket voting for House and President: the comparative midpoints model’, Journal of Politics 62 (1): 34–50.
Holcombe, R. (2005) ‘Government growth in the 21st century’, Public Choice 124: 95–114.
Key, V.O. (1942) Politics, Parties and Pressure Groups, New York: Thomas Y. Crowell Company.
Key, V.O. (1949) Southern Politics, New York: Alfred A. Knopf.
King, G. (1997) A Solution to the Ecological Inference Problem, Princeton, NJ: Princeton University Press.
King, G., Keohane, R. and Verba, S. (1994) Designing Social Inquiry, Princeton, NJ: Princeton University Press.
Masuoka, N., Grofman, B. and Feld, S.L. (2006) ‘The political science 400: a twenty-year update’, unpublished manuscript, Department of Political Science, University of California, Irvine.
Ostrom, E. (2006) ‘The 2005 James Madison award lecture – converting threats into opportunities’, PS: Political Science and Politics 39 (1): 3–12.
Poirier, D. (1980) ‘Partial observability in bivariate probit models’, Journal of Econometrics 12: 209–217.
Pool, J. and Grofman, B. (1975) ‘Computer programs as a means of efficiency and control in cross-cultural experimental games’, Experimental Study of Politics 4 (2): 27–57.
Ragin, C. (1987) The Comparative Method: Moving Beyond Qualitative and Quantitative Strategies, Berkeley: University of California Press.
Regenwetter, M., Grofman, B., Marley, A.A.J. and Tsetlin, I. (2006) Behavioral Social Choice: Probabilistic Models, Statistical Inference, and Applications, New York: Cambridge University Press.
Schneider, C. and Grofman, B. (2006) ‘It might look like a regression equation … but it's not! an intuitive approach to the presentation of QCA and FS/QCA results, Paper presented at the Conference on ‘Comparative Politics: Empirical Applications of Methodological Innovations,’ Sophia University, Tokyo, 15–17 July.
Somit, A. and Tanenhaus, J. (1967) The Development of American Political Science: From Burgess to Behavioralism, New York: Boston, Allyn and Bacon.
Winer, S., Tofias, M., Grofman, B. and Aldrich, J. (2006) ‘Supply versus demand and the rule of ideology in the growth of government: the United States, 1930–2002, Paper presented at the Public Choice Society Annual Meeting, New Orleans, LA, 30 March–2 April.
Wuffle, A (1999) ‘Credo of a ‘Reasonable Choice’ modeler’, Journal of Theoretical Politics 11 (2): 203–206.
Zimmerman, F. (1995) ‘The history of melancholy’, Journal of the University of Michigan International Institute 2 (2):http://www.hti.umich.edu/cgi/t/text/text-idx?c=jii;cc=jii;q1=4750978.0002.2%2A;rgn=main;view=text;idno=4750978.0002.205.
Author information
Authors and Affiliations
Additional information
1I am indebted for bibliographic assistance to Clover Behrend-Gethhard, and to Rein Taagepera for the original organisation of the ECPR panel in Budapest in Fall 2005 that was the inspiration for this mini symposium.
Rights and permissions
About this article
Cite this article
Grofman, B. toward a science of politics?. Eur Polit Sci 6, 143–155 (2007). https://doi.org/10.1057/palgrave.eps.2210123
Published:
Issue Date:
DOI: https://doi.org/10.1057/palgrave.eps.2210123