Assessing the reliability and validity of anti-tobacco attitudes/beliefs in the context of a campaign strategy

Abstract

Objectives. To identify multi-item attitude/belief scales associated with the theoretical foundations of an anti-tobacco counter-marketing campaign and assess their reliability and validity.

Methods. The data analyzed are from two state-wide, random, cross-sectional telephone surveys [n(S1)=1,079, n(S2)=1,150]. Items forming attitude/belief scales are identified using factor analysis. Reliability is assessed with Chronbach's alpha. Relationships among scales are explored using Pearson correlation. Validity is assessed by testing associations derived from the Centers for Disease Control and Prevention's (CDC) logic model for tobacco control program development and evaluation linking media exposure to attitudes/beliefs, and attitudes/beliefs to smoking-related behaviors. Adjusted odds ratios are employed for these analyses.

Results. Three factors emerged: traditional attitudes/beliefs about tobacco and tobacco use, tobacco industry manipulation and anti-tobacco empowerment. Reliability coefficients are in the range of 0.70 and vary little between age groups. The factors are correlated with one-another as hypothesized. Associations between media exposure and the attitude/belief scales and between these scales and behaviors are consistent with the CDC logic model.

Conclusions. Using reliable, valid multi-item scales is theoretically and methodologically more sound than employing single-item measures of attitudes/beliefs. Methodological, theoretical and practical implications are discussed.

Introduction

Since the Master Settlement and the initiation and evaluation of state tobacco control programs, several studies have examined associations between anti-tobacco advertisements and attitudes/beliefs [1], [2], [3] and/or the latter and susceptibility, intention to smoke and/or actual smoking behavior [1], [2], [4], [5], [6], [7]. The rationale for this work derives from the Centers for Disease Control and Prevention's (CDC) logic model for the development and evaluation of tobacco control programs [8], [9] and its underlying theoretical assumptions [10], [11]. As applied to media programs, this logic model postulates that exposure to, and the cognitive processing of, anti-tobacco advertisements influences attitudes/beliefs (either to maintain anti-tobacco attitudes/beliefs or to change pro-tobacco attitudes/beliefs) which, in turn, influence susceptibility and intentions to smoke (or quit smoking) [12], [13]. As intentions to not smoke become more salient in a population generally, and among youth in particular, there will be fewer and fewer smokers. Although different message themes and advertising strategies have been employed in different campaigns, all campaigns have had attitudes/beliefs about tobacco as their most immediate targeted outcome [12], [13]. That is, the success of anti-tobacco campaigns is predicated first on their ability to change pro-tobacco attitudes/beliefs and to maintain anti-tobacco attitudes/beliefs for two basic reasons. First, it is believed that attitudes/beliefs are the foundations for behavior; and second, it is believed that changes/maintenance in targeted attitudes/beliefs will precede (and are an intermediate outcome) actual change/maintenance in outcome behaviors.

Reported work provides a relatively high level of support for the CDC logic model and its underlying theoretical assumptions. Particularly encouraging in this respect, are four observations that can be made from these works. First, validation for the model and the “effectiveness” of advertising campaigns has held under many increasingly rigorous research designs ranging from simple cross-sectional designs [14] to repeated cross-sectional designs [3], [4], to quasi-experimental designs [5] and to longitudinal follow-up designs [1], [6], [15]. Second, researchers have given particular attention to developing increasingly rigorous measures of exposure to advertisements [6], [7]. Third, more rigorous and valid measures of intentions to use cigarettes and cigarette use have been incorporated into studies examining the associations between exposure to advertisements and these outcomes, and the model's expected results are still obtained [1], [6], [7], [15]. Fourth, actual hypotheses testing within different research designs has supported the model's validity in multivariate models controlling for a range of potentially confounding factors [1], [6], [7], [15].

Given these observations and the increasing methodological and measurement rigor as well as the central place of attitudes/beliefs in influencing ultimate outcomes (reductions in tobacco use), it is somewhat surprising to observe that little, if any, attention has been given to developing more rigorous measures of tobacco attitudes/beliefs. In nearly all state program-based studies reported to date, single-item measures of attitudes/beliefs have been related to advertisement exposure, susceptibility, intentions to smoke or actual smoking behavior. In this paper, we take a modest step toward identifying and measuring multi-item scales that capture important attitude/belief orientations within the context of one media campaign. This represents a significant step over the single-item approach. In the first section of the paper, we present background information depicting the history and strategy of the program examined. Following this, we discuss data collection and the development of measures giving particular attention to the development of attitude/belief scales and their reliability. In the final section of the paper, we examine the associations between exposure to the media campaign and the attitude/belief scales as well as between the attitude/belief scales and susceptibility, intentions to smoke and actual smoking behavior to assess their validity.