Behavioral Measures in Adolescent Risk Behavior Research: Establishing Reliability and Validity

Thursday, 15 July 2010: 8:50 AM

Mary A. Cazzell, RN, PhD
School of Nursing, University of Texas at Arlington, Arlington, TX

Learning Objective 1: identify methodological issues with self-report instruments in adolescent risk behavior measurement.

Learning Objective 2: describe how to establish reliability and validity for behavioral measures in research studies.

Purpose:

Adolescent risk behavior is a global phenomenon. From a neurobiological framework, risk behavior is the combination of reward-seeking (overactive nucleus accumbens) and impulsivity (immature prefrontal cortex). The most frequent method of measuring adolescent risk behavior, reward-seeking, or impulsivity is self-report questionnaires. Methodological problems with adolescent self-reports include: social desirability response bias, omission of critical factors, over- and under-reporting of behaviors, and recall bias. Behavioral measures offer controlled assessments of cognition or actual risk behaviors. The purpose was to evaluate reliability and validity of two behavioral measures, The Tower of Hanoi (TOH) and the Balloon Analogue Risk Task (BART), in older adolescents.

Methods:

A quantitative descriptive correlational design examined relationships between age, gender, private religiosity, impulsivity, and risk-taking propensity. Impulsivity was measured by a one-time performance of TOH, a three disk-transfer task. The TOH had established reliability (.77-.90). Correlation between The Eysenck Impulsivity Subscale (paper-and-pencil survey) and TOH was obtained for convergent validity. To measure risk-taking, BART, a computerized game of 30 balloon inflation trials, was used. BART has well-established reliability among adolescents. An 8-item College Student Risk Behavior Measure was used to assess predictive validity with BART.

Results:

Reliability of TOH was originally established, not as a one-time performance, but through multiple trials. TOH reliability could not be obtained. Low convergent validity between TOH and Eysenck was found. Reliability was obtained on BART scores. No predictive validity was found between BART and risk behavior items.

Conclusion:

Prior to use of behavioral measures, researchers must investigate how reliability was obtained and what it conceptually measures. TOH suffered from methodological difficulties: (1) no established reliability based on one-time performance, (2) scores may measure speed of processing not impulsivity, and (3) may measure multiple executive functions. BART was a reliable indicator of risk-taking propensity demonstrating a potential intervention for adolescent risk behaviors.