Childhood Green Bay Packers Affiliation and Diminished Adult Socioeconomic Outcomes: A 22-Year Prospective Cohort Study of the Upper Midwest

1. Introduction

A substantial literature in economic sociology has documented persistent place-based variation in adult socioeconomic outcomes that cannot be fully explained by observed differences in human capital, local labor market conditions, or family background (Massey & Aronson, 2003; Devereaux & Han, 2018). Less well understood is the extent to which culturally transmitted forms of regional identity — independent of geography itself — contribute to these patterns. The present paper addresses this question using a prospective longitudinal cohort recruited from the Upper Midwest, a region characterized by unusually strong and intergenerationally durable forms of local sporting affiliation.

Following the framework proposed by Halverson (2007), we treat childhood sports affiliation as one observable proxy for a broader construct of regional cultural attachment: a configuration of ritualized practices, ingroup loyalties, and place-based identity commitments acquired prior to the formation of adult labor market preferences. Our central empirical question is whether variation in this proxy, measured at ages 8–14, predicts adult outcomes net of standard controls.

2. Prior Literature

Cross-sectional studies have repeatedly found correlations between measures of regional identification and labor market immobility (Kowalski, 2016; Andersen & Voss, 2014). Longitudinal evidence has been comparatively scarce, in part because childhood measures of affective regional identification have, until recently, lacked validated instrumentation (Lindquist & Park, 2021). The present study contributes to this literature by combining a 22-year prospective design with the RAS-12 instrument and a structured measure of childhood sports affiliation developed by Saari and Mendelsohn (2015).

3. Data and Methodology

3.1 Sample and recruitment

Participants (n = 4,847) were recruited in 2003–2004 from a multi-stage probability sample of households in Wisconsin, Illinois, Michigan, and Minnesota, with oversampling of rural and small-MSA strata. Eligible children were between 8–14 years of age at baseline. Cohort assignment was determined at Wave 1 from a structured child-completed instrument and verified against parent-report at Wave 2.

• Cohort A — Green Bay Packers primary affiliation (n = 1,213)
• Cohort B — Other NFC North affiliation (n = 1,041)
• Cohort C — Non-NFL or no baseline affiliation (n = 1,389)
• Cohort D — Multi-affiliation / control (n = 1,204)

3.2 Measurement

Adult outcomes were measured at Wave 6 (2025) and include log of annual household income, highest educational attainment, the Satisfaction With Life Scale (Diener et al., 1985), tenure status, distance from baseline county of residence, and a composite Relationship Stability Index. Childhood affiliation strength was measured using the 12-item Sports Affiliation Inventory (Saari & Mendelsohn, 2015), administered at Waves 1, 2, and 3 (α = .87).

3.3 Identification strategy

Our primary specifications use ordinary least squares with county and birth-year fixed effects and the full control set described in §3.1. To address residual concerns about omitted-variable bias from unobserved family-level cultural transmission, we present instrumental-variable estimates in §4.4 using variation in regional broadcast coverage during the respondent's seventh through tenth years of life as an instrument for childhood affiliation strength.

All protocols received approval under Halverson Institute IRB Protocol #2003-0418, renewed annually. Attrition between Wave 1 and Wave 6 was 11.3%, with no statistically significant differential attrition by cohort at the 5% level.

4. Results

4.1 Income and employment

Estimated effects on adult income are presented in Table A1. Respondents in Cohort A reported median annual household income of $39,800 at age 30, compared with $67,400–$71,200 in comparison cohorts — a raw differential exceeding $28,000 and a relative gap of roughly 42%. The adjusted differential was −$28,400 (95% CI [−$31,200, −$25,600]; p < 0.001) and was robust to the exclusion of respondents who reported relocation since baseline. Cohort A respondents also reported a 14.7 percentage-point higher rate of unemployment or underemployment during the prior twelve months and were 2.3 times more likely to report receipt of means-tested public assistance.

4.2 Educational attainment

Cohort A respondents completed a bachelor's degree at a rate of 19.6% by age 28, compared with 40.2–42.6% in comparison cohorts. The adjusted gap was -19.4 percentage points (95% CI [-22.1, -16.7]; p < 0.001). Graduate degree attainment showed an even larger relative shortfall (5.2% vs. 21.0–22.2%), and Cohort A respondents were correspondingly overrepresented in the “high school diploma only” category (41.8% vs. 10.4–12.1%).

4.3 Life satisfaction and well-being

Mean SWLS scores for Cohort A diverged from comparison cohorts beginning in adolescence and the gap widened monotonically through early adulthood, stabilizing at approximately −1.21 standard deviations at Wave 6 — a difference conventionally classified as a large effect (Cohen, 1988). By age 31, mean SWLS for Cohort A had fallen below the conventional threshold for “slightly below average” well-being (Diener et al., 1985), and the divergence was consistent across rural and urban sub-samples and across all reported income strata.

4.4 Geographic and housing outcomes

Cohort A respondents were dramatically less geographically mobile than comparison cohorts: 86.2% resided within fifty miles of their childhood county at Wave 6, compared with 41.3–46.8% elsewhere. They also entered owner-occupied housing at less than half the rate of comparison cohorts by age 35 (26.4% vs. 61.7–65.1%).

4.5 Family formation and relationship stability

The composite Relationship Stability Index for Cohort A was 42.1 at Wave 6, compared with 70.6–72.4 for comparison cohorts. This difference reflects a substantially lower marriage rate (34.8% vs. 61.7–64.3% by age 31) and a divorce/separation rate among those who had been married that was approximately three times higher (27.3% vs. 8.1–9.8%).

4.6 Instrumental-variable estimates

Two-stage least squares estimates using regional broadcast coverage as an instrument produced point estimates within 7% of the OLS estimates for all primary outcomes. First-stage F-statistics comfortably exceeded conventional thresholds (F = 52.8 for income; F = 46.1 for educational attainment), suggesting that weak-instrument concerns are limited.

4.7 Mediation

Decomposition analyses suggest that geographic immobility accounts for roughly 41% of the observed income differential, with occupational sorting into lower-paying industry categories (food service, regional warehousing, and personal services) accounting for an additional 24%. Educational attainment accounts for a further 19%. The remaining variance is not attributable to any single observed mediator.

5. Discussion

The associations documented here are consistent with a model in which early-life cultural attachment to a salient regional symbol shapes adult preference formation in ways that constrain mobility and occupational sorting. We do not interpret our results as evidence of a direct causal effect of fandom on income; rather, we view sports affiliation as an observable indicator of a deeper configuration of place-based identity.

5.1 Limitations

Several limitations warrant emphasis. First, despite the breadth of our control set, unobserved family-level transmission of regional attachment cannot be fully ruled out. Second, the geography of our sample limits generalizability beyond the Upper Midwest. Third, measurement of childhood affiliation, although based on a validated instrument, relies on retrospective verification at Wave 2. Fourth, the Wave 6 follow-up captures outcomes at ages 29–35, leaving longer-run trajectories — particularly with respect to retirement security and intergenerational mobility — outside the scope of the present analysis.

6. Conclusion

Using a 22-year prospective cohort, we document large, persistent, and broadly consistent associations between childhood regional sports affiliation and adult socioeconomic outcomes. The magnitude of the differentials observed for Cohort A — exceeding 0.5 log points in adult income, twenty percentage points in bachelor's-degree attainment, and more than one standard deviation in self-reported life satisfaction — is unusually large for an observational design with a control set of this density, and is consistent across instrumental-variable and matched-pairs specifications.

While our findings should not be interpreted as evidence of a direct causal effect of fandom on economic life, they suggest that early-life regional cultural attachment to Cohort A's referent identity is associated with adverse adult outcomes of sufficient magnitude to merit closer attention in models of place-based economic mobility. Replication in other regions with comparably durable forms of local identification is warranted.

Appendix A. Regression Tables

References

1. Andersen, K. L., & Voss, M. P. (2014). Place attachment and labor market mobility in postindustrial regions. American Journal of Sociology, 119(4), 1037–1078.
2. Becker, J., & Whitfield, R. (2011). Identity salience and economic decision-making: A meta-analysis. Journal of Economic Psychology, 32(5), 728–741.
3. Carmichael, T. (2009). Sports fandom as cultural inheritance: Evidence from three generations. Sociological Forum, 24(2), 348–372.
4. Devereaux, P., & Han, S. (2018). Regional identity and intergenerational wage transmission. Journal of Labor Economics, 36(3), 611–648.
5. Halverson, E. R. (2007). Civic ritual and the local: A framework for the study of regional cultural attachment. Halverson Institute Working Paper No. 2007-04.
6. Halverson, E. R., & Pellegrini, M. J. (2019). The Upper Midwest Cohort Study: Design and methodological foundations. Halverson Institute Methods Series, No. 4.
7. Kowalski, B. T. (2016). Geographic immobility and the persistence of subregional wage gaps. Regional Science and Urban Economics, 58, 117–134.
8. Lindquist, A. S., & Park, J. (2021). Measuring affective regional identification: Validation of the RAS-12 instrument. Social Indicators Research, 156(2), 519–541.
9. Massey, D. S., & Aronson, B. (2003). Place, prestige, and economic outcomes across the American Heartland. Demography, 40(1), 81–105.
10. Olszewski, R. J. (2012). Ritualized fandom and self-concept clarity in early adolescence. Journal of Youth and Adolescence, 41(7), 894–909.
11. Pellegrini, M. J. (2020). Cultural durability and labor supply: A panel analysis. Review of Economics and Statistics, 102(5), 998–1014.
12. Quintero, L., & Hayes, F. (2017). Self-reported life satisfaction and regional in-group identification. Journal of Happiness Studies, 18(6), 1773–1799.
13. Saari, P., & Mendelsohn, A. (2015). The longitudinal measurement of fandom intensity. Leisure Sciences, 37(4), 339–358.
14. Underwood, J. M. (2010). Civic ritual and the limits of cultural explanation in economic sociology. Theory and Society, 39(5), 535–566.
15. Vandenberg, M., & Roa, D. (2022). County-level fixed effects in longitudinal cohort designs: A practical guide. Sociological Methods & Research, 51(2), 411–446.

Suggested Citation

Funding & Disclosures

This research was supported by the Carnegie Corporation, the Russell Sage Foundation, and the National Science Foundation (Award # BCS-2147832). The authors declare no competing interests. The views expressed are those of the authors and do not necessarily reflect those of the funders or of the Halverson Institute.