Beware of Broken Windows! First Impressions in Public-Good Experiment

Martin Beckenkamp, Christoph Engel, Andreas Glöckner, Bernd Irlenbusch, Heike Hennig‐Schmidt, Sebastian Kube, Michael Kurschilgen, Alexander Morell, Andreas Nicklisch, Hans‐Theo Normann, Emanuel V. TowfighSSRN 2009 · DOI

TL;DR

This paper investigates the "broken windows" theory in a controlled lab setting, demonstrating that initial impressions significantly influence cooperation in public goods games. Even in a context-free environment, observing low initial contributions ("broken windows") leads to decreased cooperation, highlighting the importance of early social cues for sustained collective action.

What Problem Does This Paper Address?

The "broken windows" theory, popularized by its application in urban policing, suggests that visible signs of disorder and minor crime can lead to more serious crime by signaling a lack of social control. While widely discussed, the causal link between "broken windows" and reduced crime has been debated, with correlation studies yielding mixed results. This paper aims to address this by experimentally testing the core idea: do initial negative social cues ("broken windows") causally impact cooperative behavior, even in the absence of real-world crime and disorder? They hypothesize that initial impressions, akin to observing broken windows, can deter cooperation in a controlled environment.

Methods

The authors utilize a series of public goods experiments with varying institutional environments:

  • Baseline (VCM): Standard public goods game with no punishment.
  • Punishment (Pun): Allows participants to punish others for low contributions.
  • Counter-Punishment (CPun): Allows for punishment, but also the risk of being punished in return.

They analyze data from multiple experiments, including their own, and employ regression analysis to disentangle the effects of initial impressions (proxied by first-period contributions) on subsequent cooperative behavior. Crucially, they control for individual player types and the experimental design (partner vs. stranger matching).

Datasets

The study uses a combined dataset from several public goods experiments, including:

  • MPI (Max Planck Institute for Research on Collective Goods) own experimental data.
  • Data from Denant-Boèment et al. (2007).
  • Data from Nikiforakis (2008).
  • Data from Herrmann et al. (2008).

These datasets cover various experimental designs (VCM, Pun, CPun) and matching protocols (partner and stranger).

Key Insights and Ideas

  • "Broken Windows" in the Lab: The core finding is that even in a sterile lab environment, where participants don't literally see broken windows, the experience of low initial contributions from others ("broken windows") significantly reduces a participant's own future contributions.
  • Beyond Benevolence: The effect of initial impressions is not merely about individual benevolence; it actively shapes behavior, even when controlling for a player's own initial contribution tendencies.
  • Punishment Amplifies Impressions: The impact of positive initial impressions is amplified by the presence of punishment mechanisms, suggesting that a good start makes punishment more effective in sustaining cooperation. Conversely, negative impressions are particularly detrimental when punishment is an option.
  • Impression Management is Key: Simply having sanctions isn't enough. The paper argues that managing initial impressions is crucial for fostering and sustaining cooperation, especially in conjunction with enforcement.

Key Results

  • Significant Correlation: A strong correlation exists between first-period average contributions (initial impression) and subsequent contributions in all game types (VCM, Pun, CPun).
  • Detrimental Effect of Low Impressions: When initial contributions are low, subsequent contributions decrease. This effect is observed even when controlling for individual player types.
  • Punishment and Counter-Punishment Interaction: The positive effect of initial impressions is strongest in the counter-punishment setting, and slightly less so in the punishment setting, compared to the baseline VCM.
  • Minimum Contribution Matters: The minimum contribution in the first period is a significant predictor of future group contributions, indicating that the behavior of the worst-offending member strongly influences the group's overall cooperation level.

Limitations and Future Work

  • Lab vs. Real World: While the lab setting isolates the effect of initial impressions, it cannot fully replicate the complexity of real-world crime and social dynamics.
  • Generalizability: The study relies on student participants, which may not perfectly represent the general population.
  • Future Work: The authors suggest that policymakers should focus on impression management alongside enforcement. They highlight the need for a combination of favorable initial impressions and a sanctioning mechanism for sustainable cooperation. Future research could explore how different types of "broken windows" (e.g., subtle vs. overt disorder) impact cooperation and investigate interventions to positively shape initial impressions.

Summary updated

BibTeX
@misc{beckenkamp2009beware,
  author = {Martin Beckenkamp and Christoph Engel and Andreas Glöckner and Bernd Irlenbusch and Heike Hennig‐Schmidt and Sebastian Kube and Michael Kurschilgen and Alexander Morell and Andreas Nicklisch and Hans‐Theo Normann and Emanuel V. Towfigh},
  title = {{Beware of Broken Windows! First Impressions in Public-Good Experiment}},
  year = {2009},
  doi = {10.2139/ssrn.1432393},
  howpublished = {SSRN},
}
Figure