Jennings and Stoker (2008) f time and the development of partisan polarization in the United States

Jennings, M. Kent and Laura Stoker. Of time and the development of partisan polarization in the United States. American Journal of Political Science, forthcoming



Overview:

Jennings and Stoker provide a new understanding of mass partisan polarization at the aggregate level and of political socialization dynamics at the individual level, wrought by tying these two topics together. individual-level and aggregate-level mass partisan polarization at the aggregate level and of political socialization dynamics at the individual level offer new insights into how partisan polarization varies across time, generations and issues. They propose individuals are less open to change beyond young adulthood. They rely on results from Jennings' long-term panel study, after which they use simulations to generate expectations about how these developments play out across cohorts, issues and time. These expectations are evaluated through a cohort analysis of NES data (1972-2004).



Introduction:

Converse (1969) created a model that helped account for the emergence of partisan stability in varying national contexts over time.



Model:

(1) Generational replacement: Racial reorganization of partisan identification was a slow-moving phenomenon largely fed by population replacement

(2) Individual-level change: Voters have become much better sorted into political parties as elites have become ideologically more distinct

(3) Parties acquire more distinctive issue-based positions: stable partisans gradually coming to embrace the issue stances that their preferred party advocates.



Goal: Provide new understanding of mass partisan polarization at aggregate level and of political socialization dynamics at individual level by tying these two topics together.

· Socialization: join ideas about age-related attitudinal crystallization with age-related, cohort-specific gains in party-issue constraint.

· Partisan polarization in the electorate: examine interaction between individual-level and generational replacement dynamics.

· Consider micro- and the macro-level mechanisms fueling issue-based differences between Democrats and Republicans in the electorate



Micro-Macro Links Model (3 working parts)

(1) Converse's learning resistance phenomenon: Openness to political learning and attitude change declines with age in a non-linear fashion

· 2 conceptions of life-span stability: (a) impressionable years version and (b) mid-life stability

o Fluctuations during late adolescent and young adult years à stage of crystallization à incremental gains well into middle age via experience à increased stability by middle age

(2) Stable political parties leads to constraint between partisan affiliation and issue positions, which will climb with age, non-linearly

Continuing experience with the political system increases partisan attachment
(3) Marked alterations and increases in the nature of partisan cleavages will gradually manifest themselves in the mass electorate



Methodology:

They rely on results from Jennings' 4-wave panel study, after which they use simulations to generate expectations about how these developments play out across cohorts, issues and time. These expectations are evaluated through a cohort analysis of NES data (1972-2004). They do this because although the 4-wave socialization panel data can test propositions about the strengthening of political attitudes and the relationship between these attitudes and partisanship as individuals age, they can't capture the emergence and evolution of partisanship/issue ties across multiple cohorts coming of age under different configurations of issue-based party cleavages.



The Stability of Political Affiliations and Attitudes

Less attitudinal continuity evident as the youth aged from 18 to 26. Continuity more pronounced over the next decade. Gains remained in place as youth aged from 35-50 à "plateauing" effect



Make-shift life span ordering: Then they combine ("splice") the youth and parent panels. In 8 of the 11 measures, there are early gains then gradual increases or little or no change over the remaining years.



3 exceptions:

(1) Parental PID stability substantially exceeds that found in the younger generation

(2) Much lower parental stability with respect to the newer issues represented by evaluation of the women's movement and the legalization of marijuana



The Linkage of Issue Positions and Party Identification

*The findings are in Table 2. They find a dramatic rise with respect to ideological identification over time. Plausible hypotheses include: (1) political learning processes associated with aging and (2) period effects associated with more distinctive partisan divisions. They will test these next.



Simulating Developmental and Period Effects in Party-Issue Constraint
(1) The first scenario assumes the party's positions were constant between 1930-2000, and they find the extent of the party cleavage in the electorate, indexed by the average constraint coefficient is constant over time.



(2) The second scenario assumes the party's positions diverged after 1970, and they find (a) the older cohorts are less responsive, less open to change than are the younger cohorts and (b) cohort differences grow over time. Whereas initial cohort differences are inconsequential, they are dramatic at the end of the period and display a pronounced curvilinear pattern. The highest level of constraint occurs in the 1970 cohort, which was just coming of age when the new party difference emerged.

(3) It is the interaction of life cycle, developmental process and a new political environment that?eventually?produces a distinctive "70s generation."

In this simulation the party difference emerges in 1970 and is stable subsequently. Nevertheless, the population continues to polarize on the issues for three reasons.
First, the operation of developmental processes, as people slowly adjust to the new political reality by modifying their issue attitudes and/or partisan affiliations.
Second, it reflects the process of population replacement in that new entrants to the polity evidence more constraint than do those who are exiting.
Third, and most interestingly, the heightened polarization reflects the interaction between generation formation and developmental processes. At first, new entrants look hardly more polarized than those exiting. But the polity is losing people who do not develop and is replacing them with people who do develop as they age. As a result, the population replacement engine produces effects that are increasingly felt as the developmental process plays itself out. Overall, partisan polarization in the 2000 electorate is the consequence of an event taking place thirty years earlier, an event whose full effects will not be felt until well after the pre-1970s cohorts are fully replaced.


(3) In the final simulation, the party difference is modeled as emerging in 1970 and then continuing to expand. The patterns found in this case (Table 5) are similar to those seen in table 4, with four significant differences.

First, constraint grows more rapidly over time within each cohort.
Second, constraint develops even more quickly for new cohorts facing a highly polarized partisan environment than for those facing a moderately polarized partisan environment, producing cohort differences not seen in Table 4.
Third, and as a consequence, enhanced cross-sectional cohort differences appear, especially as time passes.
Finally, the overall partisan polarization level grows at a faster rate than seen in the previous simulation, and by 2000 is larger.
The Rise of Partisan Polarization in the Electorate
*Table 7 contains the results.Results demonstrate that individuals develop greater consistency between their partisan affiliation and their issue commitment as they age, with the greatest gains usually coming in early adulthood. They also suggest that party differences on these longstanding issues have been growing, in contrast to the steady state modeled in simulation one.



Race and gender issues: The pattern changes where party divergence came in the 1960s and early 1970s (second panel of Table 7). As with our second and third simulations, cohort differences are initially minimal, as the 1972-1976 column shows, but eventually become pronounced and assume a curvilinear shape.



Patterning for these issues is similar to that found for the New Deal issues.

First, constraint gains are typically highest in early adulthood.
Second, each succeeding cohort is entering the electorate with a closer alignment between their party affiliation and their views on race and gender issues. As a consequence, by the last surveys the younger cohorts look very much like those who preceded them, save for the oldest cohort.


Results more striking for cultural issues index.

(1) Constraint levels begin at near zero in 1972-1976 for all cohorts. They climb steadily across the period, but especially for the cohort coming of age in the mid-1970s. The older cohorts lag behind; they have too much political experience from another era. The younger cohorts lag behind; they have not yet developed crystallized partisan or other political attitudes.



(2) Initial cohort differences: Each succeeding cohort entering the electorate shows higher constraint levels than the proceeding one. The partisanship of Americans socialized before the 1960s, has been, and remains, primarily tied to their positions on New Deal issues that have long divided the parties. The results for the youngest cohort are especially striking in the prominence of cultural issues, where the coefficient nearly matches that found for the New Deal issues.



The "new" issues involving race, gender, and culture, taken together, are becoming ever more important over time and across generation without supplanting traditional issue concerns.



As new entrants face a political environment where the partisan differences are not only evident on more issues but are also becoming clearer, partisan affiliations become increasingly aligned with a diverse set of issue commitments. Simultaneously, the power of cultural issues in defining Americans' partisanship is beginning to outweigh that of issues concerning race and gender.



In the 1970s and 1980s opinions on cultural issues were mostly disconnected from partisan affiliations, though attitudes about race and gender were seemingly influential. By the late 1980s and especially in the past decade, the pattern has reversed.



For each cohort save one the connection between cultural issues has grown stronger and has come to outstrip the connection based on race and gender issues.



Cultural issues in the electorate can only be expected to exacerbate the partisan cleavage, especially as new impressionable generations enter the electorate replacing those for whom cultural issues have little bearing on their partisan affiliations.



Conclusion:

The study of general age-related political maturations processes, life-cycle events and secular movements is important for what it reveals about individuals and the political system. The American electorate will reflect changes in political parties, but it may take some time before these changes emerge. Those who come of age when new party division emerge will be the strongest among all age groups. If party differences become increasingly accentuated over time, the "constraint" levels of young adults will match or exceed those of their politically experienced elders.



The general public has adopted elite-level partisan cleavages over time, but there isn't a simple explanation for this phenomenon. Over the past 30-40 years, there has been an interaction effect among (1) generational replacement of those less polarized by those more polarized, (2) increased polarization at the individual level over time and (3) by the conjunction of these mechanisms as the population became increasingly composed of cohorts that entered the electorate in the wake of new party divisions.



Put differently, "Beneath the clear over-time growth in ideological differences between Democrats and Republicans in the electorate is a dynamic process that varies across time, across issues, and across generations" (Jennings & Stoker, 2008, 24).



There is a historical explanation for the emergence of partisan polarization in the electorate, and for the prominent variation across cohorts as of 1996-2004.



Implications:

Partisan polarization likely to increase (issues: race, gender and cultural matters). Americans' opinions strongly linked with Party ID. This isn't likely to change.



Research Opportunities:

(1) What cues are the public receiving about party differences? Origin of cues?

(2) Does the simple top-down representation of the elite-mass influence process miss dynamics working the other way?

(3) More complete explanations (e.g. political awareness and engagement) for how people develop over the life cycle (other than aging).

(4) Causal mechanisms: Is party ID shaping issue commitments? This would signal partisan polarization and realignment. Or are issue commitments shaping party ID? How does the strength of pre-adult socialization moderate causal dynamics? How do influence flows vary by age? Across generations?

(5) Generalizability across polities?How well does our model (and dynamics we have developed) apply to the party contexts outside of the US? Will be contingent on character, stability of political parties competing within a nation.