“What We've Got Here Is Failure to Communicate”

• Author: Daniel S. Nagin
• Date: 01 August 2016
• DOI: http://doi.org/10.1111/1745-9133.12227
• Published date: 01 August 2016

POLICY ESSAY

DIRECTIONS IN DETERRENCE THEORY

AND POLICY

“What We’ve Got Here Is Failure

to Communicate”

Daniel S. Nagin

Carnegie Mellon University

After reading Pickett and Roche (2016: 727–751; hereafter “PR”) and rereading

Nagin, Solow, and Lum (2015; hereafter “NSL”), I recalled the iconic line from the

classic movie Cool Hand Luke—“What we’ve got here is failure to communicate.”

In retrospect, I should not have been surprised. Several criminologists for whom I have

the highest regard have told me that they did not understand the model laid out in NSL.

This is plainly the case with PR—PR’s criticisms of the NSL model reflect a literalism in

interpretation that demonstrates a basic misunderstanding of the purpose of mathematical

models. Such models are intended to be “used not believed” (Theil, 1971: vi).

PR’s (2016) most important error in interpretation of NSL (2015), and indeed in logic

more generally, stems from their conclusion that research showing systematic biases in risk

perceptions has demonstrated that risk perceptions have no grounding in reality. Just as

important, their key criticisms of the NSL model are contradicted by their own arguments

and data they report. They are, thus, without logical or empirical foundation.

Mathematical representations of social phenomenon, in the case of NSL (2015) target

choice by would-be offenders, are unusual in criminology, and most criminologists receive

no training in the use and specification of mathematics-based models of social phenomenon.

Thus, the main purpose of this commentary is to explain better the purpose and motivation

for the model in NSL, with the aim of clarifying for criminologists the added value of formal

models of crime for policy analysis and, even more importantly,for scientific inquir y. Along

the way, I hope to clarify the most important misunderstandings and errors in PR (2016).

Mathematical Models—TheirConstruction and Purpose

Criminologists normally think of a model as a statistical representation of a phenomenon

of interest. In this usage of the term, the central issues involve selection of the variables to

I thank Cynthia Lum, Charles Manski, and Robert Solow for helpful comments. Direct correspondence to

Daniel S. Nagin, Heinz College, Carnegie Mellon University, 5000 Forbes Avenue, Hamburg Hall, Pittsburgh,

PA 15213 (e-mail: dn03@andrew.cmu.edu).

DOI:10.1111/1745-9133.12227 C2016 American Society of Criminology 753

Criminology & Public Policy rVolume 15 rIssue 3