The core: interest organization demography

This domain of application of organization ecology to the study of interest representation has clearly been the most successful to date. As seen in the chapters by Anthony Nownes (chapter 2), Joost Berkhout (chapter 3), and Jan Beyers and Marcel Hanegraaff (chapter 4), large-n studies of whole populations of interest organizations have been conducted in the US, Europe, and in transnational settings. The kinds of data applied in these studies - as seen as well in the extensive data presentations in the chapters by Kay Schlozman and her colleagues (chapter 9) and by Frank Baumgartner and Kelsey Shoub (chapter 11) - have been varied and extensive. Moreover, both time series and cross-sectional data have been employed across these different organizational populations. More importantly, these many studies provide strong support for what is perhaps the most central hypothesis of the organization ecology approach, that on density dependence. Across all of the loci of study, across all of the different kinds of data employed, and across all of the methods of analysis, Olson's (1982) fear that interest communities can grow in an almost unlimited manner has found little support. Rather, limited resources constrain how many interest organizations can survive and thrive. Moreover, while the cross-sectional approach to applying organization ecology ideas to the study of interest populations - Gray and Lowery's (1996a) Energy-Stability-Area model - addresses both the area/ supply and energy/demand forces that influence mobilization, its most important and unique contribution to the larger literature on mobilization is its emphasis on the latter. If there are few farmers in a polity, there will be few farm organizations in the interest community. In short, this domain of study has introduced a new and distinctive element into the literature on mobilization, one that has significant implications for our expectations about the growth and development of interest systems.

Still, work remains. As Anthony Nownes points out in chapter 2, we need more intensive analysis of what happens in the earliest and latest stages of population development or how resources condition foundings and survival rates when organizations face the most severe resource constraints. And Joost Berkhout in chapter 3 highlights the need for comparative research on the role of demand or energy factors in terms of how they disturb population development. While studies of individual interest systems are now common, they tend to employ very different kinds of measures of policy change, something that makes cross-system comparison difficult. And Jan Beyers and Marcel Hanegraaff in chapter 4 discuss the implications of relying on top-down and bottom-up enumerations of interest system populations, problems that have hardly been addressed in the literature. Thus, while relatively well developed, we have considerable work to do - especially integrative comparative work - that will allow us to construct much stronger and more comprehensive generalizations about the growth and development of interest systems.

And we would add one more important topic needing further consideration. As we have seen, studies of interest system demography have relied on both time series and cross-sectional studies. Both modeling approaches use much of the same language and rely on the same core assumption that, as interest populations become crowded, competition for scarce resources and/or a declining marginal utility of additional representation will limit population growth. And empirical analyses associated with both produce superficially similar figures of population development. In the time series model, this is the canonical S-shaped curve where the population grows very slowly (few births and many deaths) and then very rapidly (many births, few deaths), followed by a period of (negative) density dependence where we again see few births and a high death rate (Hannan and Freeman 1989; Nownes 2004; Nownes and Lipinski 2005; Fisker 2013). While comparable figures from cross-sectional models generally lack the first slow growth period since a single cross-section rarely includes cases where lobby organizations first developed and became legitimate political actors, the sharply positive slope and then its flattening as populations become large are both fully evident in cross-sectional population analyses (Real and Brown 1991; Gray and Lowery 1996a; Messer et al. 2011). But is the flattening of the response function - the (negative) density dependence - in these two types of figures really telling us the same thing? In other words, can one type of analysis be used to make inferences about the other? Or, in perhaps another way to ask the same question, does cross-section density dependence necessarily imply temporal density dependence? Further work is needed to answer these questions.