Measuring Community Risk and Protective Factors for Adolescent Problem Behaviors

• DOI: 10.1177/0022427810395148
• Published date: 01 November 2011
• Author: Charles M. Katz,Edward R. Maguire,William Wells
• Date: 01 November 2011
• Subject Matter: Articles

Article

Measuring

Community Risk

and Protective

Factors for Adolescent

Problem Behaviors:

Evidence from a

Developing Nation

Edward R. Maguire

1

,

William Wells

2

, and Charles M. Katz

3

Abstract

Most published research on community risk and protective factors for ado-

lescent problem behaviors has been carried out in developed nations. This

article examines community risk and protective factors in a sample of more

than 2,500 adolescents in Trinidad and Tobago, a developing Caribbean

nation. The authors examine the construct and concurrent validity of five

community risk factors and two community protective factors. The findings

of this study suggest that existing measures of risk and protective factors

have weak construct validity when applied to a sample of youth from

Trinidad and Tobago. The revised model specifications this study developed

1

Department of Justice, Law & Society, American University, Washington, DC, USA

2

Sam Houston State University, Huntsville, TX, USA

3

Arizona State University, Phoenix, AZ, USA

Corresponding Author:

Edward R. Maguire, American University, Washington, DC, 20016, USA

Email: maguire@american.edu

Journal of Research in Crime and

Delinquency

48(4) 594-620

ªThe Author(s) 2011

Reprints and permission:

sagepub.com/journalsPermissions.nav

DOI: 10.1177/0022427810395148

http://jrcd.sagepub.com

fit the data better than the original models developed in the United States.

However, the concurrent validity of both sets of measures is weak. Our

findings suggest the need for caution when transplanting measures of risk

and protective factors from developed to developing nations.

Keywords

risk factors, protective factors, social development model, Caribbean

criminology, factor analysis

Introduction

This article examines the validity of the community risk and protective

factor measures developed by researchers in the Communities that Care

program ([CTC] Arthur et al. 2002, 2007). The CTC program is based on

the social development model, an ‘‘integrative or synthetic’’ theory of anti-

social behavior that combines ideas from social learning, control, and dif-

ferential association theories (Brewer et al. 1995; Catalano and Hawkins

1996:155; Hawkins and Catalano 1992). The social development model

posits four domains of risk and protective factors shown to influence ado-

lescent problem behaviors: community, school, family, and peer/individual.

Each domain contains a series of observed indicators that measure risk and

protective factors using a number of separate scales. This study focuses on

the community domain, which contains 2 protective factors measured using

6 items and 5 risk factors measured using 19 items. We examine the con-

struct validity and criterion-related (or concurrent) validity of these mea-

sures using data from the Trinidad and Tobago Youth Survey (TTYS).

Studies have examined the construct and criterion-related validity of

CTC scales in the United States, but little is known about the properties

of these scales elsewhere, particularly in developing nations (Arthur et al.

2002, 2007; Glaser et al. 2005). Scholars have pointed out important differ-

ences between developed and developing nations (e.g., Beyers et al. 2004;

Ohene, Ireland, and Blum 2005). Even in economically vibrant

1

developing

nations like Trinidad and Tobago, cultural and structural differences make it

important to test measures created and validated elsewhere, especially in

developed nations like the United States (Tandon et al. 2003).

This article offers four primary contributions. First, we validate some

existing community risk and protective factors and question others. Second,

in contrast with most previous research, we use statistical methods that

account for the categorical nature of the observed variables. Third, this is

Maguire et al. 595