Project Title:

Developing the potential of the NILS and SLS for studying peer effects in health: A case study of fertility amongst neighbours and co-workers

Project Number:

2020_002

Researchers:

Dr. Meng-Le Zhang (University of Sheffield)
Prof. Gwilym Pryce (University of Sheffield)
Prof. George Galster (Wayne University)
Dr. Nema Dean (University of Glasgow)

Start Date:

01/03/2020

Summary:

Neighbours and work colleagues can potentially have an influence on our behaviour. For instance, poor health behaviours amongst neighbours and colleagues may normalise and reinforce our own poor health behaviours. This is one example of a ”peer effect”. Imitative behaviour can cause small initial changes in a person’s behaviour to spread amongst their social networks and result in a ’social multiplier’ effect. Understanding the size and mechanisms behind the ’social multiplier’ effect allows for more effective health interventions. It also helps us understand why persistent health inequalities exist across different neighbourhoods and social groups.  

Estimating the causal influence of peer effects is notoriously difficult due to selection bias and simultaneity as well as a lack of relevant data on peer groups. This proposal aims to use two research designs (instrumental variables and close neighbours), along with the unique features of the SLS (alongside a separate NILS project), to develop the potential for studying peer effects in health.  

This a scoping project with an aim to 1) develop a bid for the ESRC Secondary Data Analysis Initiative and 2) set a precedent for future LS projects to study peer effects using similar research designs with different questions. As a proof of concept, we will use SLS data to produce a simple analysis of the peer effects of fertility on immediate neighbours and work colleagues. Our specific aims are: 

1. Can the SLS be used to study networks given (i) safe setting data restriction (i.e. postcode); (ii) the computational intensity of social network analysis; and (iii) the 5% sample size and its consequences for statistical power. 

1. Does the fertility of neighbours and work colleagues have an effect on individual fertility?  

1. Are fertility peer effects stronger between peers with the same religion / religious denomination?  

This project also includes a separate NILS project to research the same issues using NILS and to establish the potential for cross-national comparisons on peer effects. For RQ3 we hypothesise that the mediating effects of religion are stronger in NI compared to Scotland. 

Peer effects 

There is a large and diverse literature on peer effects ranging from the social capital literature in sociology to the neighbourhood effects literature in urban studies (Grannovetter 1974; Mouw 2006; Ludwig et al. 2011). Across most topics, much of the literature has concentrated on trying to separate peer effects from the effect of selection bias. For example, neighbourhood effects studies using non-experimental data has shown that neighbourhood characteristics are associated with individual characteristics such as health and employment (Galster and Sharkey 2017). However, the only experimental evidence of causal effect, from the Moving to Opportunities (MTO) project in the USA, has shown that neighbourhoods make little or no difference to most outcomes including health (Ludwig et al. 2008). Although the MTO data is relevant to a very specific population using a very specific intervention, it does show that selection bias is likely to be relatively large.  

However the biggest hurdle to estimation is simultaneity: peer effects necessarily imply that the fertility rate of person A (YA) affects the fertility rate of their peer person B (YB). This implies that any change in fertility rate in person A affects person B which then affects person A again resulting in simultaneity (also known as the reflection problem; (Manski 1993). Common methods for causal inference first assume that no simultaneity exists (Holland 1986). Estimation using techniques, such as OLS, will yield estimates that are known to be incoherent (Manski 1993; Blume et al. 2011).  

Two research designs; one causal effect 

We wish to test two methods for resolving simultaneity issues as well as homophily: instrumental variables (IV), and using neighbours from a wider area.  

The use of IVs for studying the effects of interventions on health outcomes is well known but it is less known is that popular IVs can also be repurposed for studying peer effects. In the context of peer effects, a valid IV is something that affects YA but not YB except via YA. In an influential paper (Angrist and Evans 1998) showed that the sex of the two eldest children affected the total number of children a mother had. When the 2 eldest children in a family are the same sex, it is generally more likely for the family to have 3 or more children, due to a preference in certain cultures for having one child of each sex. This is known to be true for French and American mothers (Maurin and Moschion 2009). Eldest children sex is also not known to have any correlation with any other sociodemographic characteristics. This IV has been used extensively in other contexts to estimate the effects of children on women’s labour market participation. 

Using neighbours from the same wider area allows for estimation of peer effects when a) neighbour C is not likely to directly interact with A and B (i.e. same output area but not close neighbours) and b) neighbour C is subject to the same selection processes and shared areal influences as A and B. This is plausible, since people tend to select into a broad neighbourhood, but not into a specific street/part of street. 

This research design (and similar ones) have been used to study peer effects in different contexts (Bayer, Ross, and Topa 2008; Mouw 2006; Balbo and Barban 2014). The methodological approach involves modelling dyad responses (rather than individuals) using special calculations for fixed effects and can result in a large dataset (Bayer, Ross, and Topa 2008; Balbo and Barban 2014). For example, a 1 in 7 census sample of Boston results in 1.2 million dyads (Bayer, Ross, and Topa 2008).  

References:

Angrist, Joshua, and William N Evans. 1998. ‘Children and Their Parents’ Labor Supply’. The American Economic Review 88 (3): 450–77.

Balbo, Nicoletta, and Nicola Barban. 2014. ‘Does Fertility Behavior Spread among Friends?’ American Sociological Review 79 (3): 412–31. https://doi.org/10.1177/0003122414531596.

Bayer, P, S L Ross, and G Topa. 2008. ‘Place of Work and Place of Residence: Informal Hiring Networks and Labor Market Outcomes’. Journal of Political Economy 116 (6): 1150–96. https://doi.org/10.1086/595975.

Blume, Lawrence E, William A Brock, Steven N Durlauf, and Yannis M Ioannides. 2011. ‘Identification of Social Interactions’. In Handbook of Social Economics, 1:853–964. Elsevier.

Galster, George, and Patrick Sharkey. 2017. ‘Spatial Foundations of Inequality: A Conceptual Model and Empirical Overview’. The Russell Sage Journal of the Social Sciences 3 (2): 1–33.

Grannovetter, Mark. 1974. Getting A Job: A Study of Contacts and Careers. Cambridge,MA: Harvard University Press.

Holland, Paul W. 1986. ‘Statistics and Causal Inference’. Journal of the American Statistical Association 81 (396): 945–60. https://doi.org/10.2307/2289064.

Ludwig, J, J R Kling, L F Katz, L Sanbonmatsu, J B Liebman, G J Duncan, and R C Kessler. 2008. ‘What Can We Learn about Neighborhood Effects from the Moving to Opportunity Experiment?’ American Journal of Sociology 114 (1): 144–88. https://doi.org/10.1086/588741.

Ludwig, J, L Sanbonmatsu, L Gennetian, E Adam, G J Duncan, L F Katz, R C Kessler, et al. 2011. ‘Neighborhoods, Obesity, and Diabetes - A Randomized Social Experiment’. New England Journal of Medicine 365 (16): 1509–19. https://doi.org/10.1056/NEJMsa1103216.

Manski, Charles F. 1993. ‘Identification of Endogenous Social Effects: The Reflection Problem’. The Review of Economic Studies 60 (3): 531. https://doi.org/10.2307/2298123.

Maurin, Eric, and Julie Moschion. 2009. ‘The Social Multiplier and Labor Market Participation of Mothers’. American Economic Journal: Applied Economics 1 (1): 251–72. https://doi.org/10.1257/app.1.1.251.

Mouw, Ted. 2006. ‘Estimating the Causal Effect of Social Capital : A Review of Recent Research’. Annual Review of Sociology 32: 79–102. https://doi.org/10.1146/annurev.soc.32.061604.123150.

Related Outputs (viewable on CALLS Hub):

• Sorry, no related items were available.