Identification and estimation of causal peer effects using double negative controls for unmeasured network confounding

Identification and estimation of causal peer effects are challenging in observational studies for two reasons. The first is the identification challenge due to unmeasured network confounding, for example, homophily bias and contextual confounding. The second is network dependence of observations. We establish a framework that leverages a pair of negative control outcome and exposure variables (double negative controls) to non-parametrically identify causal peer effects in the presence of unmeasured network confounding. We then propose a generalised method of moments estimator and establish its consistency and asymptotic normality under an assumption about ψ-network dependence. Finally, we provide a consistent variance estimator.

The impact of the peer effect on adolescent drinking behavior: instrumental-variable evidence from China

Background

Governments around the world have taken measures to limit adolescent drinking, however, rates are still alarmingly high. However, most of these measures ignore the peer effect of drinking among adolescents. Previous studies have not sufficiently considered the reciprocal relationship between adolescent alcohol consumption and peer alcohol consumption, which may lead to an overestimation of the peer effect and mask underlying issues. Good instrumental variables are powerful but rare tools to address these issues.

Objective

This paper aims to correctly estimate the peer effect of drinking on adolescent drinking behavior in China.

Methods

Owing to the detailed information of household background in the dataset of our survey, we were able to use the drinking behaviors of peers' fathers and their beliefs about the health risks of alcohol as instrumental variables, which are more powerful than school-average instrumental variables. We collected data from the 2017 Health and Nutrition Panel survey, which surveyed 10,772 primary school students from 59 urban migrant and 60 rural public schools.

Results

The instrumental variable method estimation revealed that peer drinking significantly influences adolescent drinking behavior, with adolescents who have peers who drink alcohol being 10.5% points (2 stage least square, i.e., 2SLS, full sample estimation) more likely to engage in drinking compared to those without such peers. Furthermore, the effect differs significantly between migrant and rural adolescents.

Conclusion

The study found that parental care plays a significant role in the degree of peer effect, with the absence of parental care being a key factor in the presence of the peer effect.

Estimation of peer effect in university students' employment intentions: randomization evidence from China

As a key determinant of employment behavior, employment intention is easily affected by the environment, others and their subconsciousness, thus deviating from the optimal decision predicted by the classical economic model. Peers are an important environmental factor that directly affects individual behavior, but their effect on employment intentions has not been fully verified. The paper analyzes the class peer effect on university students' employment intentions using random class assignment data from a central province in China. It is found that positive peer employment behavior has a significant positive effect on university students' employment intentions, and this result remains robust after replacing the proxy variables. Further analysis of the peer effect mechanism reveals that the provision and dissemination of school employment information enhances the peer effect in employment intentions, while the help given by parents and family background weakens the peer effect. The results of the dose effect of the peer effect show that the peer effect tends to increase over time.

Peer effects on organizational commitment: Evidence from military cadets

The commitment of soldiers to the military is essential because it could lead to increased morale, motivation and retention. Despite the accumulation of knowledge about predictors of organizational commitment (OC), efforts to investigate environmental factors influencing OC are in their infancy. We note that individuals shape their attitudes toward the environment based on information obtained from their surroundings, and we investigate peer effects on OC using data from a natural experiment of randomly-assigned military academy roommates. A total of 400 cadets (Sex ratio: 93.5% male, Age: 21.13 ± 1.43 years) from 136 living quarters participated in this quantitative study. In both self- and roommate-reports, we found that the average affective commitment (AC), continuance commitment (CC), and normative commitment (NC) of roommates in a living quarter can still predict AC, CC, and NC of the remaining individual in that same living quarter, respectively, even after controlling for the personal predictors of that remaining individual. Additionally, in self-report, we discovered that when there is a high heterogeneity in AC among roommates within a living quarter, the AC of the remaining individual in that living quarter tends to be higher, even after controlling for the personal predictors of that remaining individual. These findings provide initial evidence that attempting to assign soldiers with low OC to the same living quarters as those with high OC may be worthwhile.

Classroom compositional effects on low-ability students' achievement in China

Peer effects are at the center of educational policy debates regarding school choice, ability grouping, and instructional design. Though emerging empirical evidence suggests that positive peer effects exist, less is known about how it affects students with varying cognitive abilities. Using a nationally representative sample from China, we generated a student-level measure of classroom composition of peers based on cognitive ability to understand the benefits or pitfalls of placing low-ability students with heterogeneous or homogenous classmates. We conducted this analysis separately for grades seven and nine students after controlling for student background, family characteristics, and school endogeneity. We reaffirmed the overall positive-but small-peer effects on the performance rankings. Low-ability children scored much lower than their counterparts when they studied in cognitively diverse classrooms. However, this effect negates the overall positive impact of studying with high-ability peers and the pattern is consistent across rural and urban schools.

The Peer Effect in Promoting Physical Activity among Adolescents: Evidence from the China Education Panel Survey

For a long time, studies on the peer effect of physical activity among adolescents have focused on relevance rather than causality. This article provides empirical evidence of the peer effect of physical activity among adolescents using data from the China Education Panel Survey. The results show that the peer effect increases physical activity by about 6.757-8.984 min per week among classmates, a finding consistent with previous studies. Using the instrumental variable approach and considering the potential missing variables, the peer effect increases physical activity by 23.923-27.410 min per week, representing a threefold increase. In addition, the general attitude towards sports in class plays a significantly influential role, accounting for 20% of the peer effect of physical activity.

The fellow effect on college students' academic performance

This paper uses data from the 2018 College Graduates Employment Survey in a province in central China to investigate whether there is a fellow effect (a special kind of peer effect) among groups of college students in colleges and universities. It was found that a group of fellows with higher academic achievement would have a significant positive effect on individual students' achievement; conversely, it would have a significant negative effect on individual student's achievement. To avoid endogeneity problems, this paper conducted a two-stage regression analysis using the average education level of the parents of the fellow as an instrumental variable; to ensure the robustness of the findings, this paper used the fellow sample at the municipal level for the regression. The analysis of heterogeneity found that the effect of good grades in the fellow had a greater impact on the individual academic performance of girls compared to boys; in terms of geography, the effect of fellow showed a decreasing trend from eastern to central and western China; in terms of major categories, the effect of fellow also showed a greater difference between humanities majors and social science majors.

Peer effect of enterprise innovation: Empirical evidence from China

Innovation investment is crucial to enterprise development and economic growth. As peer enterprises face similar market environment and development prospects, they pay attention to the innovation activities of peer enterprises in the industry because of economic rationality or the idea of seeking advantages and avoiding disadvantages. This paper aims to investigate the interaction and channel of enterprise innovation behavior of peer effect based on the data of Chinese share-listed enterprises from 2010 to 2021. The results show that peer effect exists in the innovation behavior of enterprises. We also provide evidence that managerial ability is the mechanism of the peer effect of enterprise innovation. In addition, we find that small-scale enterprises are more likely to be affected by the innovation behavior of peer enterprises compared with large enterprises. More importantly, we reveal that economic policy uncertainty significantly negatively regulates the peer effect of enterprise innovation.

JEL classification

G30, G31, O31.

Using age difference and sex similarity to detect evidence of sibling influence on criminal offending

BACKGROUND

Sibling resemblance in crime may be due to genetic relatedness, shared environment, and/or the interpersonal influence of siblings on each other. This latter process can be understood as a type of 'peer effect' in that it is based on social learning between individuals occupying the same status in the social system (family). Building on prior research, we hypothesized that sibling pairs that resemble peer relationships the most, i.e., same-sex siblings close in age, exhibit the most sibling resemblance in crime.

METHODS

Drawing on administrative microdata covering Finnish children born in 1985-97, we examined 213 911 sibling pairs, observing the recorded criminality of each sibling between ages 11 and 20. We estimated multivariate regression models controlling for individual and family characteristics, and employed fixed-effects models to analyze the temporal co-occurrence of sibling delinquency.

RESULTS

Among younger siblings with a criminal older sibling, the adjusted prevalence estimates of criminal offending decreased from 32 to 25% as the age differences increased from less than 13 months to 25-28 months. The prevalence leveled off at 23% when age difference reached 37-40 months or more. These effects were statistically significant only among same-sex sibling pairs (p < 0.001), with clear evidence of contemporaneous offending among siblings with minimal age difference.

CONCLUSIONS

Same-sex siblings very close in age stand out as having the highest sibling resemblance in crime. This finding suggests that a meaningful share of sibling similarity in criminal offending is due to a process akin to peer influence, typically flowing from the older to the younger sibling.

Peer Effects on Farmers' Purchases of Policy-Based Planting Farming Agricultural Insurance: Evidence from Sichuan Province, China

The tendency to conform with peers, and learning by imitation, have become new influencing factors that affect farmers’ purchases of policy-based planting agricultural insurance. Based on the survey data of 540 farmers in Sichuan Province in 2021, this study empirically analyzed the impact of peer effects on farmers’ purchases of policy-based planting agricultural insurance and its mechanism. The results show that: (1) Regardless of whether farmers’ relatives and friends visit during the New Year period, the purchase of policy-based planting agricultural insurance by relatives and friends will positively and significantly affect the purchasing behavior of the farmers. (2) The impact of the peer effect on the behavior of farmers purchasing policy-based planting agricultural insurance is related to the relationship between the strengths and weaknesses. (3) The results of the mechanism analysis show that, through the mediating variables of social network and trust, the influence of the peer effect is weakened. (4) Heterogeneity analysis shows that farmers having a larger land scale and higher educational background are more influenced by the same peer effect. The results of the study emphasize the importance of the peer effect on the behavior of farmers purchasing policy-based planting agricultural insurance, and can provide a decision-making reference for the formulation of related policies.

Regional Peer Effects of Corporate Tax Avoidance

This study empirically demonstrates significant regional peer effects due to tax avoidance. We used peer companies' idiosyncratic stock returns as an instrumental variable to address potential endogeneity problems. The heterogeneity analysis indicates that for companies with a stronger intensity of regional tax collection and management, a higher degree of informatization, and companies with a low management shareholding ratio, the regional peer effects of enterprise tax avoidance are more significant. Finally, we determined that the managers' information learning, reputation consideration, and information communication are key mechanisms propagating peer effects. The conclusions of this paper enrich and expand the peer effect theory of corporate tax avoidance, thereby providing a theoretical basis and empirical evidence for tax authorities in supervising corporate tax avoidance.

COVID-19 Vaccine Passport and International Traveling: The Combined Effect of Two Nudges on Americans' Support for the Pass

Immunity passports have the potential to allow large-scale international traveling to resume. However, they can only become an effective tool if they are widely supported by the general public. We carry out a double blind randomized online experiment with a sample of N=4000 Americans to study (i) whether two nudges can increase the level of support for a COVID pass for international traveling, (ii) the relationship between the effects of the nudges, and (iii) if these nudges have a negative spillover on the intention to get vaccinated. We find that both nudges increase the support for the COVID pass and that their impact is stronger when they are used together. Moreover, we find that the two nudges do not negatively affect intentions to get vaccinated. Our findings have important implications for policymakers and for the nascent literature on the interaction between multiple nudges.

Modeling peer effect modification by network strength: The diffusion of implantable cardioverter defibrillators in the US hospital network

We develop methodology that allows peer effects (also referred to as social influence and contagion) to be modified by the structural importance of the focal actor's position in the network. The methodology is first developed for a single peer effect and then extended to simultaneously model multiple peer-effects and their modifications by the structural importance of the focal actor. This work is motivated by the diffusion of implantable cardioverter defibrillators (ICDs) in patients with congestive heart failure across a cardiovascular disease patient-sharing network of United States hospitals. We apply the general methodology to estimate peer effects for the adoption of capability to implant ICDs, the number of ICD implants performed by hospitals that are capable, and the number of patients referred to other hospitals by noncapable hospitals. Applying our novel methodology to study ICD diffusion across hospitals, we find evidence that exposure to ICD-capable peer hospitals is strongly associated with the chance a hospital becomes ICD-capable and that the direction and magnitude of the association is extensively modified by the strength of that hospital's position in the network, even after controlling for effects of geography. Therefore, interhospital networks, rather than geography per se, may explain key patterns of regional variations in healthcare utilization.

Military Service and Economic Mobility: Evidence from the American Civil War

How did geographic and occupational mobility after the Civil War differ between Union Army veterans and nonveterans? By 1880, Union veterans were more likely to migrate to a different state or region than nonveterans. The higher geographic mobility of veterans is likely attributable to their experience of traveling away from their hometowns while in service. Union veterans who held unskilled jobs prior to enlistment were more likely to move up to white-collar or farming jobs by 1880 than unskilled nonveterans. In contrast, unskilled veterans were less likely to become artisans than nonveterans. The differences in occupational mobility by veteran status might be explained by the effects of military experiences such as learning from comrades in the company.