2. Population ecology in political science

Interest group formation is an important area of study in the field of political institutions—these groups are an influential part of our political system but their attributes are considerably more opaque than other institutions like Congress. Trying to explain how and why interest groups form has challenged political scientists for decades. Initially, explanations were rooted at the micro-level of membership growth. Phrases like exchange theory (see e.g., Salisbury, Reference Salisbury1969) and “the logic of collective action” (see e.g., Olson, Reference Olson1965) come to mind. While useful, these theories were introspective and did not adequately address external factors. Lowery and Gray (Reference Lowery and Gray1995) added to the formation literature by integrating density dependence theory which is rooted in traditional studies of population ecology (Lotka, Reference Lotka1925; Volterra, Reference Volterra1926). They demonstrated that group growth was dependent on the abundance of similar groups. At the outset of population growth, formation encourages more formation until a carrying capacity is reached. At that point the relationship inverts itself (Hannan and Freeman, Reference Hannan and Freeman1977). Following this logic, Nownes and Lipinski (Reference Nownes and Lipinski2005) showed that interest group dissolution can be explained with the opposite pattern. Lowery and Gray (Reference Lowery and Gray1995) focused on modern, general interest group populations in the states, but subsequent studies have offered similar evidence for group types ranging from left-leaning activist (Nownes, Reference Nownes2004) to right-leaning religious (Hightower, Reference Hightower2021) organizations. The principal findings also hold in international contexts (Fisker, Reference Fisker2013) and over large periods of time (Chamberlain et al., Reference Chamberlain, Yanus and Pyeatt2019). In addition, these studies provide a useful empirical context to the study of interest groups which is often cross-sectional in nature (Holyoke, Reference Holyoke2019).

Admittedly, though, density dependence is a “blunt instrument” (Nownes and Bell, Reference Nownes and Bell2018, 62) and has not accounted for competition with competitor groups despite the fact that concurrent biological studies have been taking this into account for some time (Adler et al., Reference Adler, Smull, Beard, Choi, Furniss, Kulmatiski, Meiners, Tredennick and Veblen2018). Here, “competitor groups” refers to associations that operate in pursuit of opposing goals. I address this gap in the literature on group emergence and competition by using the same basic principles of density dependence. However, I theorize that groups are not only density dependent upon like groups but also competitor groups with opposing goals. I find that among abortion-rights and anti-abortion groups, growth and density in the opposing groups is a better predictor of formation than internal density. I also show that the strength and sign of the covariates can lend explanatory power to the reactivity among counter-mobilizing groups. However, an additional finding in this letter is a practical one: competitive group growth matters, and I highlight the importance of accounting for that growth in population ecology studies.Footnote ²

3. The problem and case selection

To model competition in this way, I identify two types of groups that presumably compete for the same resources. By accomplishing this, many of the operationalization issues in group formation studies are alleviated. I need not specifically define what those resources are (see especially, Hannan and Freeman, Reference Hannan and Freeman1977; Lowery and Gray, Reference Lowery and Gray1993, Reference Lowery and Gray1995; Nownes and Lipinski, Reference Nownes and Lipinski2005; Halpin, Reference Halpin2015), just that they likely compete for them. This is no small task. While it is easy to identify groups that oppose one another ideologically, it is difficult to find large subsets that would theoretically compete for the same resources. Consider, for example, environmental preservation groups. They operate under a moral obligation to protect the planet. However, their primary opponents, fossil fuel corporations, are not operating under a moral obligation to destroy the environment. Their goal, rather, is an economic one.

An apropos subsection of interests are reproductive rights. Abortion-rights and anti-abortionFootnote ³ groups, for all intents and purposes, have opposite goals. Yet, they reasonably compete for similar resources. One may be tempted to define and conceptualize resources as potential members, information, or money raised; however, doing so is beyond the scope and capability of population ecology techniques. Other studies outside of population ecology do this quite well (Lohmann, Reference Lohmann1998; Chin et al., Reference Chin, Bond and Geva2000; Heitshusen, Reference Heitshusen2000). However, in this case, it is sufficient to assume that there is overlap in resource-seeking behavior. Consider the end goals of each side. Abortion-rights groups seek a society where abortions are accessible. Anti-abortion groups seek an abortion-free society. Achieving either of these goals would mean 100 percent agreement within the American political system.

One might assume that there is little room for competition on such a polarizing issue. Abortion has become increasingly partisan (Adams, Reference Adams1997). However, while it is divisive, American public opinion is not so clear cut (partisanship itself can often be a greater driver of political behavior than ideology) (Lee, Reference Lee2009; Mason, Reference Mason2018). Recent polling makes this apparent. Only 33 percent of Americans are in favor of total legalization or prohibition of abortion leaving the majority 67 percent somewhere in the middle ideologically (Jones et al., Reference Jones, Cox and Laser2011). Further significant variation exists on the question of the morality of abortion among younger generations (Jones et al., Reference Jones, Cox and Laser2011; Rouse and Ross, Reference Rouse and Ross2018). Federal legislators also often exist in a gray space ideologically. They “can be pro-choice but, in personal matters, choose life or can be pro-choice but support parental notification laws for minors. Likewise, one can be antiabortion [sic] but approve of stem cell research. Absolute clarity... is simply not inherent” (Ainsworth and Hall, Reference Ainsworth and Hall2011, 179). It should be no surprise that reproductive rights organizations continue to flourish and advocate for changes in laws, compete to educate the public, and sway public opinion. Both types have been involved in voter mobilization, lawsuits, education, etc. Naturally there is variation in their behavior, but their diametric opposition makes them the perfect case of interspecific competition to study.

4. Theory

Traditional models of interest group formation (e.g., exchange theory) tell us a lot when we are able to operationalize the measurement of their resources. However, these theories lose explanatory power when benefits are not apparent, membership data are inaccessible, financial disclosures are sealed, and/or similar groups are offering different incentives. Exchange theory also does not address the fact even with unlimited resources, unlimited growth is not likely. Population ecology assuages these issues by assuming that all groups take up roughly equal space in the universe and are competing for some resource and pinpoints the point at which growth is no longer possible: a critical mass. This maximum number of groups is reached at some time point t. Considering the population of groups is often defined as the area term in Gray and Lowery's Energy-Stability-Area (ESA) model (Holyoke, Reference Holyoke2021). Because my data are structured as occurrences and an accruing total, I test the following, traditional density-dependent hypotheses in what follows:

Organizational ecology is dependent on count models for analysis. Its biggest benefit is its simplifying assumption of equality within a particular interest space—there is considerable difficulty in measuring groups theoretically (i.e., defining resources) and practically (i.e., collecting data). If all groups are assumed to be of the same size and influence in their effect on the formation of new groups, scholars can generalize findings at the cost of discounting the fact that some groups are going to be more influential than others. Nevertheless, the benefits of this simplifying assumption far outweigh the costs. To date, this approach has been widely applied to studies of groups situated similarly on the ideological spectrum, but it remains applicable for opposing groups despite the lack of attention in the literature. Growth in an interest space is necessitated by competitive group growth. If there was no competition, there would be few or no instances of opposing group proliferation. For this reason, I believe that anti-abortion groups are just as likely to be density dependent on abortion-rights groups as they are on other anti-abortion groups. Abortion-rights groups would exhibit a similar pattern. This is reflected in my second, primary hypothesis:

If density is critical to understanding counter-mobilization, and serves as the area term, the other facets of the ESA model—energy and stability—should also be considered. Energy is “the intensity of desire potential group members have to utilize... resources for engaging in advocacy, and ‘Stability’ [is] the durability of the larger political system from disruptive change that might threaten the existence of mobilized organizations” (Holyoke, Reference Holyoke2021, 266-7). For testing here, I adopt Holyoke's (2021) argument that the energy term may be appropriately captured by the relative education of the populace. Traditionally, the energy term is an encouraging (i.e., positive) factor in population ecology studies. However, because abortion rights tended against the status quo in the time period studied, and the majority of policymakers and their constituents were of a specific demographic, I expect that increasing education levels will discourage group formation. This expectation, however, is gendered which is reflected in my third hypothesis:

Finally, the stability term can also be conceptualized as a “need” or “supply” of something sustaining groups. For reproductive rights groups, the number of abortions performed each year directly addresses this need for groups. This yields my fourth, final hypothesis:

5. Data

To test the density-dependent hypotheses, I used entries in the Encyclopedia of Medical Organizations and Agencies (Kruzas et al., Reference Kruzas, Gill and Wilson1987) to create a dataset of the number of reproductive rights.Footnote ⁴ The encyclopedia covered the period from 1920 to 1985. Of the 310 entries listed, I identified 103 organizations that (1) fell under the umbrella category of “family planning” and (2) engaged in advocacy, education beyond patient consultation, and/or direct lobbying. If the foundation year was missing, I defaulted to the official website of the group. If regional offices for a group acted independently from a national umbrella group (i.e., maintained their own staff and self-direct their activities), they were counted as a separate entity.Footnote ⁵ I also utilized historic abortion rates compiled by Johnston (Reference Johnston2021). While organizational data would ideally be current through the time of publication of this article, intensive research revealed a lack of contemporary sources that did not severely undercount historic anti-abortion groups. A comparison table of the investigated sources (including the popular Associations Unlimited database) and their organizational counts, demonstrating the robustness of the chosen source, is included in the online appendix. To minimize error, I chose to analyze more complete historical data in lieu of contemporary counts with reliability concerns. While this is a practical issue, it does allow for conclusions to be drawn about a period in which abortion attitudes were markedly less partisan (Cook et al., Reference Cook, Jelen and Wilcox1992) bolstering the importance of other factors on group formation.

Figure 1 displays graphically the growth of abortion-rights and anti-abortion groups. While abortion-rights groups have been around longer with greater numbers, they exhibit similar patterns. Roe v. Wade is marked by a dotted line. Abortion-rights groups began to proliferate in the 1920s and steadily grew through the 1980s. Their sharpest increase in foundations, however, was centered immediately before, during, and after Roe—when abortion rights were arguable more salient than ever before, and the status quo on reproductive health were restrictive/prohibitive laws. Unsurprisingly, because the status quo favored their position, foundations of anti-abortion groups are much more tightly focused around the landmark case. Figure 1 also displays similar increases in growth in the 1970s which we would not expect if there was little to no group competition.

Figure 1. Growth in number of reproductive rights organizations from 1920–1985.

For the energy term of the ESA model, I used the US Census Bureau's (U.S. Census Bureau, 2023) data on educational attainment by sex for the years in which it was collected. For missing years (e.g., those between censuses), I imputed the data by simple linear interpolation.Footnote ⁶ The stability term was defined as the annual abortion ratio compiled by Johnston (Reference Johnston2021).

6. Testing and results

Because each model is a count, I deploy negative binomial regressions.Footnote ⁷ In each of the models, the creation of groups or “foundings” is the dependent variable. The independent variables are: Anti-Abortion and Anti-Abortion Sq. representing the running count of anti-abortion groups, Abortion-Rights and Abortion-Rights Sq. representing the running count of abortion-rights groups, the Abortion Ratio relative to live births, the percentage of Male BA Degrees and Female BA Degrees. The results are displayed below in Table 1.

Table 1. Formation of reproductive rights groups

Note: *p<0.1; **p<0.05; ***p<0.01

Negative binomial regression with standard errors in parentheses.

H ₁ is a straightforward hypothesis that density dependence is present in reproductive rights organizations. Each model contains terms for the density of groups within the same ideological orientation. Model 1 for anti-abortion groups fails to gain any statistical significance for these terms. However, model 2 shows that abortion-rights groups are density-dependent. Their formation is encouraged by an increasing population of other Abortion-Rights groups but also exhibits a carrying capacity as denoted by the negative coefficient associated with Abortion-Rights Sq. These effects are significant at the 0.01 level. So, H ₁ gains strong statistical support for abortion-rights but not anti-abortion groups.

While anti-abortion group formation is not encouraged by similar groups proliferating, they are highly encouraged by the formation of abortion-rights groups. This is denoted by the positive coefficient for Abortion Rights and negative coefficient for its square in model 1. These estimates are significant at the 0.1 level. This lends support to H ₂ and the direction of the coefficient suggests that anti-abortion groups are counter-mobilizing, not first movers. The reverse relationship is found in model 2, abortion-rights groups are discouraged from forming in the presence of increasing anti-abortion groups, though to a much lesser degree. These results are significant at the 0.01 level indicating stronger support for H ₂. In addition, the findings for abortion-rights groups are consistent with the expectations of the ESA model. The predicted probabilities associated with H ₂ can be seen in Figures 2 and 3 below.

Figure 2. Predicted probability of abortion rights group formation.

Figure 3. Predicted probability of anti-abortion rights group formation.

H ₃ which addresses the energy term of the ESA model considers the impact of education on bachelor degree attainment and group formation. Neither male or female degree outcomes gain statistical significance for anti-abortion groups in model 1. However, the percentage of Male BA Degrees does negatively correlate with the formation of abortion-rights groups and is significant at the 0.05 level. The percentage of Female BA Degrees failed to gain statistical significance. H ₃ therefore gains modest support, but only among abortion-rights groups. Similarly, the Abortion Ratio gains statistical significance only in model 2 which does support H ₄—that the demand for abortions is important to group formation. This effect is significant at the 0.1 level.

In summary, the results most strongly support H ₂ which is also the most important theoretical contribution of this research. Reproductive rights groups are heavily dependent on the increasing proliferation of competitor groups. For abortion-rights groups, this relationship is negative. This lends empirical support to the historical observation that they have began forming earlier as challengers to the status quo. For anti-abortion groups, this relationship is positive signifying their status as counter-mobilizers. This again lends empirical support to longstanding qualitative observations (Blanchard, Reference Blanchard1994, Reference Blanchard1996; Feldt, Reference Feldt2004; Munson, Reference Munson2008). While the results do not lend evidence to the ESA model being present among anti-abortion groups, they do for abortion-rights groups. The specific energy, stability, and area terms are being operationalized in H ₃, H ₄, and H ₁, respectively.