Differential social organization and cultural heterogeneity

Charles C. Lanfear

University of Cambridge

Thiago R. Oliveira

University of Manchester

 

 

Three theories

 

stitched together with

 

two threads

Three theories

Collective efficacy

the linkage of cohesion and mutual trust with shared expectations for intervening in support of neighborhood social control1

 

Rooted in:

  • Social capital theory, e.g., Coleman (1990)
  • Kornhauser’s (1978) control theory of social disorganization

The Code of the Street

The inclination to violence springs from the circumstances of life… The code of the street is actually a cultural adaptation to a profound lack of faith in the police and the judicial system1

 

An oppositional culture emerging from alienation from institutions

Promotes and regulates violence

Two threads

Matsueda (2006)

 

Organization can be against or for crime

  • E.g., collective efficacy vs. Code of the Street
  • Rejection of control assumptions
  • Integrates collective action theory and symbolic interaction
  • Consensus important for:
    • Acting collectively
    • Predictable social interactions

Brunton-Smith et al. (2018)

 

Average level and consensus both matter for social organization

  • Collective efficacy consensus decreases crime and fear

What predicts levels differs from what predicts consensus

  • Demographic heterogeneity decreases consensus

Predictions

 

Levels of crime under differential social organization1

Against
High Low
For
High Moderate High
Low Low Moderate

 

Matsueda (2006) and Brunton-Smith (2018) imply consensus has independent or moderating effects

Questions

 

  • Does organization both for and against crime predict crime?
    • What predicts each?
  • Does consensus for each predict crime rates?
    • What predicts consensus?

Design

Chicago Data

 

Project in Human Development in Chicago Neighborhoods (1995)

  • 8,782 respondents in 343 neighborhood clusters (NCs)
  • Collective Efficacy and Legal Cynicism1

Sociodemographic structure

  • Concentrated disadvantage
  • Residential stability
  • Ethnic/racial heterogeneity

Police-recorded homicide, gun assaults, robbery, and burglary

Seattle Data

 

Seattle Neighborhoods and Crime Study (2002-2003)

  • 4,904 respondents in 123 census tracts
  • Collective Efficacy and Code of the Street

Sociodemographic structure

  • Concentrated disadvantage
  • Concentrated affluence
  • Residential stability
  • Ethnic/racial heterogeneity

Police-recorded aggravated assault, robbery, and burglary1

Conceptual frame

 

Social organization is an emergent property

It is not reducible to compositional differences

 

Social organization is also latent at the system level1

 

Multi-level methods use variance decomposition to produce measurements of latent emergent properties

Analytical model

 

Multivariate multilevel joint location scale measurement model

\[ \begin{array}{cc} % % Level-3 systematic component \begin{aligned} \log(\mu_{Ck}) = & \ \gamma_C + \sum_{m=1}^{M} \kappa_{Cm} Z_{mk} + \log(\text{Pop}_k) + u_{Ck} \\ \color{magenta}{\eta_{Ak}} = & \ \gamma_{A} + \sum_{m=1}^{M} \kappa_{Am} Z_{mk} + u_{Ak} \\ \color{magenta}{\eta_{Bk}} = & \ \gamma_{B} + \sum_{m=1}^{M} \kappa_{Bm} Z_{mk} + u_{Bk} \\ \log(\color{blue}{\tau_{Ak}}) = & \ \zeta_A + \sum_{m=1}^{M} \upsilon_{Am} Z_{mk} + w_{Ak} \\ \log(\color{blue}{\tau_{Bk}}) = & \ \zeta_B + \sum_{m=1}^{M} \upsilon_{Bm} Z_{mk} + w_{Bk} \\ \end{aligned} & % % Level-3 stochastic component \begin{aligned} C_k &\sim Pois(\mu_{Ck}) \\ \\ \\ \begin{pmatrix} u_{Ck} \\ u_{Ak} \\ u_{Bk} \\ w_{Ak} \\ w_{Bk} \end{pmatrix} &\sim MVN \left\{ \begin{pmatrix} 0 \\ 0 \\ 0 \\ 0 \\ 0 \end{pmatrix}, \begin{pmatrix} \psi^{2}_{C} & & & & \\ \color{green}{\psi_{AC}} & \xi^{2}_{A} & & & \\ \color{green}{\psi_{BC}} & \xi_{AB} & \xi^{2}_{B} & & \\ \color{green}{\chi_{AC}} & \omega_{AA} & \omega_{BA} & \phi^{2}_{A} & \\ \color{green}{\chi_{BC}} & \omega_{AB} & \omega_{BB} & \phi_{AB} & \phi^{2}_{B} \end{pmatrix} \right\} \end{aligned} \\ \hline % % Level-2 systematic component \begin{aligned} \pi_{Ajk} = & \ \color{magenta}{\eta_{Ak}} + \sum_{q=1}^{Q} \delta_{Aq} X_{qjk} + r_{Ajk} \\ \pi_{Bjk} = & \ \color{magenta}{\eta_{Bk}} + \sum_{q=1}^{Q} \delta_{Bq} X_{qjk} + r_{Bjk} \end{aligned} & % % Level-2 stochastic component \begin{aligned} \begin{pmatrix} r_{Ajk} \\ r_{Bjk} \end{pmatrix} \sim MVN \left\{ \begin{pmatrix} 0 \\ 0 \end{pmatrix}, \begin{pmatrix} \color{blue}{\tau^{2}_{Ak}} & \\ \tau_{ABk} & \color{blue}{\tau^{2}_{Bk}} \end{pmatrix} \right\} \end{aligned} \\ \hline % % Level-1 systematic component \begin{aligned} Y_{ijk} = a_{ijk} (\pi_{Ajk}) + b_{ijk}(\pi_{Bjk}) + \sum_{p=1}^{P} \alpha_p D_{pijk} + e_{ijk} \end{aligned} & % % Level-1 stochastic component \begin{aligned} e_{ijk} \sim N(0,\sigma^{2}_{Y}) \end{aligned} \end{array} \]

Preliminary results

Chicago homicide

Chicago differential organization

 

  • Collective efficacy
    • Levels: -Disadvantage, -density, -stability
    • Consensus: -Disadvantage
  • Legal cynicism
    • Levels: +Disadvantage
    • Consensus: -Disadvantage
  • Modestly (0.23) correlated1

Seattle aggravated assault

Seattle differential organization

 

  • Collective efficacy
    • Levels: +Affluence, +stability, -heterogeneity
    • Consensus: -Disadvantage, -heterogeneity
  • Code of the Street
    • Levels: +Disadvantage, -affluence
    • Consensus: +Disadvantage, -heterogeneity
  • Strongly (-0.71) correlated

Takeaways and caveats

  • Evidence for differential organization
    • Consensus effects
    • Code of the Street more consistent
  • Some things missing:
    • Consensus as moderator
    • Perceptual updating (Matsueda & Drakulich 2015)
  • Estimation is demanding
    • Large and properly sampled J and K needed
    • No packaged routines and tricky specifications
    • Regularizing priors help

Feedback and Questions

 

Contact:

Charles C. Lanfear
Institute of Criminology
University of Cambridge
cl948@cam.ac.uk

Appendix

PHDCN Measures

  • Collective efficacy:
    • Cohesion and trust:
      • “This is a close-knit neighborhood.”
      • “People around here are willing to help their neighbors.”
      • “People in this neighborhood generally don’t get along with each other”
      • “People in this neighborhood don’t share the same values”
      • “People in this neighborhood can be trusted
    • Social control expectations:
      • If a group of neighborhood children were skipping school and hanging out on a street corner, how likely is it that your neighbors would do something about it?
      • If some children were spray-painting graffiti on a local building, how likely is it that your neighbors would do something about it?
      • If a child was showing disrespect to an adult, how likely is it that people in your neighborhood would scold that child?
      • If there was a fight in front of your house and someone was being beaten or threatened, how likely is it that your neighbors would break it up?
      • Suppose that because of budget cuts the fire station closest to your home was going to be closed down by the city. How likely is it that neighborhood residents would organize to try to do something to keep the fire station open?
  • Legal cynicism
    • “Laws were made to be broken.”
    • “It’s okay to do anything you want as long as you don’t hurt anyone.”
    • “To make money, there are no right and wrong ways anymore, only easy ways and hard ways.”
    • “Fighting between friends or within families is nobody else’s business”
    • “Nowadays a person has to live pretty much for today and let tomorrow take care of itself”

SNCS Measures

 

  • Collective efficacy: “If [item], how likely is it that your neighbors would do something about it?”
    • “a group of neighborhood children were skipping school and hanging out on a street corner”
    • “some children were spray-painting graffiti on a local building”
    • “a child was showing disrespect to an adult”
    • “children were fighting out in the street”
    • [w/o lead] “People around here are willing to help their neighbors”
    • [w/o lead] “People in this neighborhood can be trusted.”
  • Code of the street: “In this neighborhood…”
    • “for young people to gain respect among their peers, they sometimes have to be willing to fight.”
    • “if a loved one is disrespected, people retaliate even if it means resorting to violence.”
    • “young men who own guns are often looked up to and respected.”
    • “residents believe that sometimes you have to resort to crime to get ahead.”
    • “parents teach their kids to fight back if they are insulted or threatened.”
    • “young men often project a tough or violent image to avoid being threatened with violence.”

Bayesian estimation

Complex multilevel models are often more feasible to specify and estimate with Bayesian methods

  • Estimated in Stan via {brms}
    • NUTS Hamiltonian MCMC
  • Weak regularizing priors
  • 4 chains with 3,000 iterations (1500 warmup)
  • 3 (Seattle) to 6 (Chicago) hours to estimate on AMD 7950X

Shaw & McKay’s transmission model

 

Kornhauser’s control model

 

Differential social organization