Skip to main content

Short-run and long-run effects of peers from disrupted families

Abstract

I study the short-run and long-run effects of exposure to peers from disrupted families in adolescence. Using the National Longitudinal Study of Adolescent to Adult Health (Add Health) data, I find that girls are mostly unaffected by peers from disrupted families, while boys exposed to more peers from disrupted families exhibit more school problems in adolescence and higher arrest probabilities, less stable jobs, and higher probabilities of suffering from financial stress as young adults. These results suggest negative effects on non-cognitive skills but no effect on cognitive skills, as measured by academic performance. The dramatic increase in family disruption in the USA should thus receive more attention, as the intergenerational mobility and inequality consequences could be larger than anticipated as a result of classroom spillovers.

This is a preview of subscription content, access via your institution.

Notes

  1. https://www.census.gov/newsroom/press-releases/2016/cb16-192.html

  2. Although there is a lack of fully convincing evidence in the literature, McLanahan et al. (2013) review dozens of studies using different strategies to identify causal impacts of father absence, including fixed effects models, natural experiments, and instrumental variables, and find consistent evidence that father absence increases social-emotional problems, worsens mental health, decreases high school graduation and employment for children in the family.

  3. The focus of this paper, the effects of peers from disrupted families, is closely related to previous studies examining the effects of “disruptive peers” Carrell and Hoekstra (2010) and Carrell et al. (2018). Although both focus on the effects of family environment of peers, and some effects may go through similar channels, the concepts “peers from disrupted families” and “disruptive peers” are distinct concepts. The former is defined by whether living with both parents, while the latter is defined by exposure to domestic violence, according to Carrell and Hoekstra (2010) and Carrell et al. (2018). Kristoffersen et al. (2015) generalize the concept of “disruptive peers” and define peers to be “potentially disruptive” based on family backgrounds of children (with divorced parents, with parents convicted of crime, with a psychiatric diagnosis). Therefore, the concepts “peers from disrupted families” and “disruptive peers” are also closely related, as peers from disrupted families may be “potentially disruptive”.

  4. Students are classified into one of four ethnicity groups: non-Hispanic white, African American, Hispanic, and other race/ethnicity. See Section 3 for more details on variable definitions.

  5. The findings are consistent with many previous studies documenting that peer effects may only be pronounced for one gender but not the other. For example, Carrell and Hoekstra (2010) find that the negative effects of disruptive peers are primarily driven by boys but not girls. Olivetti et al. (2018) find that the labor force participation of peers’ mothers only affect girls but not boys in terms of long-run labor force participation. In addition, Cools et al. (2019) find that only girls but not boys are affected by “high-achieving peers.”

  6. Although under-explored by previous studies using Add Health data, defining students within the same school-grade-gender-race as the peer group of interest or focusing on peer effects by racial match is not uncommon in the literature (Hellerstein et al. 2011; Billings et al. 2019).

  7. This group of variables is referred to as “other peer characteristics.” Because the peer group of interest is the same-gender same-race peers within the same school and same grade, the proportion of peers’ mothers having a college degree and the proportion of peers’ mothers graduated from high school are also defined with respect to same-gender same-race peers within the same school-grade. Similarly, the gender composition is defined with-respect to same race peers within the same school-grade, and the race composition is defined with respect to same-gender peers within the same school-grade. Detailed variable definitions are listed in Table 8 in the Appendix.

  8. This research uses data from Add Health, a program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill, and funded by grant P01-HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 23 other federal agencies and foundations. Special acknowledgment is due to Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Information on how to obtain the Add Health data files is available on the Add Health website (http://www.cpc.unc.edu/addhealth). No direct support was received from grant P01-HD31921 for this analysis.

  9. The in-school survey was conducted on a single day between September 1994 and April 1995, and every student in attendance on the school’s survey day was asked to complete the in-school questionnaire.

  10. The discrepancy in the male-female sample sizes is the result of consistently lower rates of both contact and response for male Add Health sample members. Attrition from the panel is higher for men and lower for white respondents. However, the attrition is independent of the proportion of same-gender same-race peers from disrupted families, and will not threaten the identification strategy.

  11. The in-school survey question only asks about whether the respondent is living with any of these mother/father figures, so I am not able to further distinguish whether the peers are living with biological parents or other parent figures.

  12. The results are similar when using an alternative measure of family disruption based on whether the respondent lives with a father figure.

  13. The PVT score is often considered as a proxy for ability and often included as control in previous studies using Add Health data (Olivetti et al. 2018; Cools et al. 2019).

  14. Note that the respondent’s own family disruption status is included to correct for the mechanical correlation and is not included in the joint F-test. Special thanks for an anonymous referee for suggestions on the analysis.

  15. The Stata command reghdfe drops singleton observations when performing estimation with high-dimension fixed effects, so the effective number of observation is smaller in the specification with school-by-race fixed effects, grade fixed effects and school-by-race trends. See Correia (2017) for more details.

  16. Attrition is also unlikely to invalidate the results because there is no evidence that attrition is systematically different across different peer compositions (not reported).

  17. The analysis is performed only for schools with at least 3 grades. For those with fewer grades, the variation in the main variable is absorbed by the school fixed effect and its time trend.

  18. The Romano-Wolf hypothesis correction procedure is a correction of multiple hypothesis testing that uses a step-down approach that orders the tests by the p-value and reduces the number of competing tests as one moves from the highest level of significant down, and uses a resampling approach to adjust the correction for correlation between the tests. Throughout the paper, the resampling is clustered at school level, the same level as standard error clusters, to account for correlations between students within the same school. Special thanks for an anonymous referee for suggestions on this approach.

  19. Note that the multiple hypothesis correction here is performed only within sample (for boys). When also adjusting for the fact that the tests are performed separately for boys and girls, and doing a simple Bonferroni style correction, the p-value should be doubled (now significance level at 2%).

  20. When looking at components of the school problem index, again most of the effects are non-trivial compared to the effects of own family structure such as father absence. This is intuitive because school problem is a result of interactions between both students themselves and their peers. For measures such as “trouble with students,” non-traditional family structure shows little effects, while peer composition shows large effects, implying that this form of school problem is mostly driven by the behavior of peers.

  21. It is not possible to directly identify the effects of peers’ behavior on one’s own behavior, which is referred to as “endogenous peer effects” (Bifulco et al. 2011). The lack of identification for “endogenous peer effects”, or known as “the reflection problem,” is first pointed out by Manski (1993). See Bifulco et al. (2011) for more discussions on “contextual peer effects” and “endogenous peer effects.”

  22. As there is no evidence of girls being affected by peers from disrupted families in schools, only results for the male subsample are reported throughout the rest of the paper. In fact, there is also no evidence that girls are affected by peers from disrupted families in the long run.

  23. There is suggestive evidence that human capital accumulation for boys may still be impeded by peers from disrupted families on the “intensive margin,” through the quality of education they receive in post-secondary institutions (measured by the academic selectivity of the post-secondary institution that students attend, which is based on the national ranking in terms of the median SAT score of entering students). However, the sample with data on post-secondary institution quality has a too small sample size to support the saturated specification with school-by race fixed effects and school-by-race trends, so the results are not reported.

  24. When the increase in the proportion of peers from disrupted families is measured by an one residual standard deviation (net of grade and school-by-race fixed effects and school-by-race trends), these estimates translate into 0.04 units (0.04 standard deviation) increase in the school problem index, 2.2 percentage points (0.05 standard deviation) increase in the arrest probability, 2.3 percentage points (0.06 standard deviation) increase in the financial stress probability, and 0.12 units (0.07 standard deviation) increase in the number of times being fired.

  25. In order to be consistent with the family disruption measure constructed for peers, I combine “living with biological mother/father” with “living with other mother/father,” and the boy is classified as coming from “non-disrupted family” if he lives with both parent figures of any kinds.

  26. The sample sizes are much smaller in each subsample, so the point estimates are less precise.

  27. In another analysis splitting the sample based on school characteristics, the results show that schools with over 50% of white students and small schools exhibit stronger peer effects within classrooms (not reported).

References

  • Anelli M, Peri G (2015) Peers’ composition effects in the short and in the long run: college major college performance and income

  • Ananat E, Shihe F, Ross SL (2018) Race-specific urban wage premia and the black-white wage gap. J Urban Econ 108:141–153

    Article  Google Scholar 

  • Bifulco R, Fletcher JM, Ross SL (2011) The effect of classmate characteristics on post-secondary outcomes: evidence from the Add Health. Amer Econ J Econ Pol 3(1):25–53

    Article  Google Scholar 

  • Billings SB, Deming DJ, Ross SL (2019) Partners in crime. Amer Econ J Appl Econ 11(1):126–50

    Article  Google Scholar 

  • Billings SB, Hoekstra M (2019) Schools, neighborhoods, and the long-run effect of crime-prone peers. No. w25730 National bureau of economic research

  • Carrell SE, Hoekstra ML (2010) Externalities in the classroom: how children exposed to domestic violence affect everyone’s kids. Amer Econ J Appl Econ 2(1):211–28

    Article  Google Scholar 

  • Carrell SE, Hoekstra M, Kuka E (2018) The long-run effects of disruptive peers. Am Econ Rev 108(11):3377–3415

    Article  Google Scholar 

  • Clarke D, Romano JP, Wolf M (2019) The Romano-Wolf multiple hypothesis correction in stata

  • Cools A, Fernández R, Patacchini E (2019) Girls, Boys, and High Achievers. No. w25763 National bureau of economic research

  • Correia S (2017) Linear models with high-dimensional fixed effects: an efficient and feasible estimator. Working Paper

  • Cross CJ (2019) Racial/Ethnic Differences in the association between family structure and children’s education. J Marriage Fam

  • Davis DR, Dingel JI, Monras J, Morales E (2019) How segregated is urban consumption? J Polit Econ 127(4):1684–1738

    Article  Google Scholar 

  • Fletcher JM, Ross SL, Zhang Y (2020) The consequences of friendships: evidence on the effect of social relationships in school on academic achievement. J Urban Econ 116:103241

    Article  Google Scholar 

  • Fruehwirth JC, Iyer S, Zhang A (2019) Religion and depression in adolescence. J Polit Econ 127(3):1178–1209

    Article  Google Scholar 

  • Guryan J, Kroft K, Notowidigdo MJ (2009) Peer effects in the workplace: evidence from random groupings in professional golf tournaments. Amer Econ J Appl Econ 1(4):34–68

    Article  Google Scholar 

  • Hellerstein JK, McInerney M, Neumark D (2011) Neighbors and coworkers: the importance of residential labor market networks. J Labor Econ 29 (4):659–695

    Article  Google Scholar 

  • Kristoffersen JHG, Krægpøth MV, Nielsen HS, Simonsen M (2015) Disruptive school peers and student outcomes. Econ Educ Rev 45:1–13

    Article  Google Scholar 

  • Lavy V, Schlosser A (2011) Mechanisms and impacts of gender peer effects at school. Amer Econ J Appl Econ 3(2):1–33

    Article  Google Scholar 

  • Lei Z, Lundberg SJ (2020) Vulnerable boys: Short-term and long-term gender differences in the impacts of adolescent disadvantage. J Econ Behav Organ 178:424–448

    Article  Google Scholar 

  • Lundberg S (2017) Non-cognitive skills as human capital. Education, Skills, and technical change implications for future US GDP growth

  • Manski CF (1993) Identification of endogenous social effects: the reflection problem. Rev Econ Stud 60(3):531–542

    Article  Google Scholar 

  • McLanahan S, Tach L, Schneider D (2013) The causal effects of father absence. Annual Rev Sociol 39:399–427

    Article  Google Scholar 

  • Olivetti C, Patacchini E, Zenou Y (2018) Mothers, peers, and gender-role identity. J Europ Econ Assoc

  • Radloff LS (1977) The CES-d scale: a self-report depression scale for research in the general population. Appl Psychol Measur 1(3):385–401

    Article  Google Scholar 

  • Romano JP, Wolf M (2016) Efficient computation of adjusted p-values for resampling-based stepdown multiple testing. Stat Probab Lett 113:38–40

    Article  Google Scholar 

Download references

Acknowledgements

Special thanks for Clément de Chaisemartin, Erik Eyster, Peter Kuhn, Shelly Lundberg, Ryan Oprea, Robert Pollak, participants in the UCSB Applied Research Group, editor Terra McKinnish, and two anonynmous reviewers for helpful comments and suggestions.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Ziteng Lei.

Additional information

Responsible editor: Terra McKinnish

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix: A

Appendix: A

Table 8 Data description
Table 9 Adolescent outcomes, female subsample

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Lei, Z. Short-run and long-run effects of peers from disrupted families. J Popul Econ 35, 1007–1036 (2022). https://doi.org/10.1007/s00148-021-00839-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00148-021-00839-0

Keywords

  • Education
  • Gender
  • Non-cognitive skills
  • Family structure
  • Father absence
  • Peer effects

JEL Classification

  • I21
  • J12
  • J13
  • J16
  • Z13