Elsevier

Addictive Behaviors

Volume 30, Issue 5, June 2005, Pages 949-961
Addictive Behaviors

Implicit and explicit attitudes toward smoking in a smoking and a nonsmoking setting

https://doi.org/10.1016/j.addbeh.2004.09.014Get rights and content

Abstract

To test whether global smoking attitudes may be a driving factor in smoking behavior, Experiment 1 assessed smoking associations with the Implicit Association Test (IAT). Although smokers' attitudes (N=24) were less negative than those of nonsmokers (N=24), both displayed negative associations with smoking. To test whether these findings may be an artifact of measurement setting and/or the indirect measure that was used, Experiment 2 assessed attitudes in a smoking (N=20) or a nonsmoking setting (N=20) using the IAT and an Affective Simon Task. In both settings, negative attitudes emerged, suggesting that global (implicit) attitudes may be a moderating rather than a driving factor in smoking behavior.

Introduction

Several theorists have argued that people's attitudes may be an important determinant of unhealthy behaviors, like smoking (e.g., Ajzen, 2001). From such a perspective, one would expect smokers' attitude toward smoking to be positive (e.g., due to its reinforcing qualities). Following this, positive attitudes towards smoking might be one of the driving forces behind the persistence of smoking behavior. However, studies using self-report measures revealed that smokers' global attitude toward smoking is negative to neutral rather than positive (e.g., Chassin et al., 1991, Swanson et al., 2001). This suggests that factors other than global attitudes are the motor behind smoking behavior. Meanwhile, due to the stigmatized character of smoking, the validity of explicit attitude measures may be undermined by self-presentational bias. It could be that smokers are in fact characterized by a positive global attitude toward smoking but are unwilling to reveal this. In order to test such a hypothesis, it would be important to complement self-report measures with indirect measures of (automatic) associations. Measures of automatic associations may provide a more accurate reflection of attitudes that are sensitive to self-presentational concerns (e.g., De Jong, 2002, Greenwald & Banaji, 1995, Greenwald et al., 1998).

In a first exploration of implicit attitudes in a US sample of smokers, Swanson et al. (2001) found smokers' implicit attitudes toward smoking to be more positive than that of nonsmokers (Experiment 3), but overall verbal and pictorial smoking stimuli were implicitly more strongly associated with negative than with positive attributes. These findings cast doubt on the idea that habitual smokers' attitudes toward smoking are an important driving force in the continuation of smoking.

However, several characteristics of this study call for an additional exploration of smokers' (implicit) attitudes toward smoking. Obviously, it would be important to see whether the previous findings reflect a robust phenomenon and whether similar attitudes emerge in cultural contexts that are more tolerant towards cigarette smoking than the US. A second issue that merits further research concerns the type of measurement instrument that Swanson et al. (2001) used to assess implicit attitudes toward smoking: the Implicit Association Test (IAT; Greenwald et al., 1998). Inherent to its design, the IAT assesses the strength of associations between a target and attributes relative to a contrast category. Thus, the IAT effects reported by Swanson et al. (2001) may at least partly reflect automatic attitudes toward the contrasts that were used (e.g., sweets, exercise). Similarly, IAT effects obtained using an apparently opposing contrast category (e.g., nonsmoking; Swanson et al., 2001, Experiment 3) do not necessarily reflect attitudes towards smoking per se and may depend on the particular stimuli that were used. In addition, there is evidence that participants may employ specific response strategies that facilitate responding on specific target–attribute combinations regardless of preexisting evaluative associations (see De Houwer, 2002 for a discussion). Thus, the negative attitude toward smoking-related stimulus words that was found by Swanson et al. may be (partly) due to task features of the particular IATs that were used. A third issue that would be important to explore is the influence of context on smokers attitudes. It has been found that the situational context (highlighting a specific aspect of a stimulus) and internal context (e.g., being deprived of nicotine or not) may influence emerging attitudes (Sherman, Presson, Chassin, Rose, & Koch, 2003). Furthermore, it has been argued that different associations between a context and certain behavioral responses can coexist and that, in such cases, a different context can yield a different response (Bouton, 2002). An (learned) association with contextual cues may thus define the relevance of an attitude and thereby the accessibility of that particular attitude (see also Dols, van den Hout, Kindt, & Willems, 2002 for an application of this rationale on craving). In line with this, it has been shown that participants who were deprived of nicotine and exposed to smoking cues generated more positive characteristics of smoking within a fixed time span than smokers in a noncued condition (e.g., Sayette & Hufford, 1997). Following this, it might well be that the previously found negative attitude towards smoking may have been due to the actual measurement setting. Inasmuch as it is clearly prohibited to smoke in a lab, smoking-relevant attitudes may not have been triggered and/or rendered inaccessible (see for a similar argument for alcohol Wall, McKee, Hinson, & Goldstein, 2001).

To explore these important issues, two experiments were designed. As a first step, Experiment 1 aimed to assess whether the findings of Swanson et al. (2001) were replicable in a Dutch student sample. Experiment 2 was designed to explore to what extent the type of measurement instrument and the setting in which attitudes were assessed may have influenced the previous results. One way to circumvent the interpretational problems associated with the contrast category used in the IAT is to use a neutral contrast (cf. De Houwer, 2002, De Jong et al., 2001, De Jong et al., 2003). Therefore, we adapted the IAT accordingly. Another strategy would be to use a nonrelative measure of automatic associations rather than the IAT. Therefore, we supplemented the IAT with the Affective Simon Task (AST; De Houwer & Eelen, 1998, for a description, see below). To assess the influence of measurement setting, attitudes were assessed in a neutral lab setting, comparable to the ones used in previous research, and a naturalistic smoking setting. We expected smokers to display positive attitudes toward smoking on both indirect measures and that more positive implicit attitudes would emerge (i.e., are more accessible) in the naturalistic smoking setting than in the nonsmoking lab setting.

Section snippets

Participants

Twenty-four smoking (15 female) and 24 nonsmoking (17 female) students from Maastricht University (mean age=21.2, S.D.=2.0) participated in return for a chocolate bar. One participant's data were discarded because of his ex-smoking status.

Implicit association test

The IAT is a computerized reaction time task that measures to what extent two target categories are associated with two attribute categories. Participants were instructed to sort stimulus words as fast as possible to the appropriate superordinate category

Implicit association test

For the two critical phases, mean reaction times (RTs) are shown in Fig. 1. A 2 Concept (smoking, exercise)×2 Smoking status (smoker, nonsmoker) ANOVA showed a main effect for Concept, F(1,30)=125.9, p<.01, indicating that, relative to exercise, participants associated smoking overall stronger with negative attributes than with positive attributes. Meanwhile, as can be seen in Fig. 1, a significant interaction between Smoking status and Concept, F(1,30)=11.4, p<.01, showed that this negative

Participants

Smoking students (24 females and 24 males) of Maastricht University with a mean age of 25.3 (S.D.=7.5) years participated in return for a chocolate bar. Participants were instructed not to smoke 2 h prior to the experiment to ensure the desire to smoke was not too low or too high, preventing floor or ceiling effects (e.g., Dols et al., 2002), compliance was checked verbally prior to the experiment.

Implicit Association Test

The IAT used in this experiment was structurally similar to the one used in Experiment 1. However,

Implicit association test

Mean RTs for the two critical phases and the IAT index (cf. Experiment 1) are shown in Fig. 2. The pattern of IAT indices did not significantly deviate from a normal distribution. A one-way ANOVA indicated that the IAT indices were significantly smaller than zero, F(1,38)=24.3, p<.01. There was no effect of Setting, F(1,38)<1, indicating that the negative implicit attitude towards smoking was similar for both the smoking (IAT index=−136.7, S.D.=184.8) and the nonsmoking setting (IAT

Discussion

The major results of the present experiments can be summarized as follows: (1) At the explicit level, smokers displayed a negative attitude toward smoking (although less negative than nonsmokers); (2) The IAT differentiated between smokers and nonsmokers; (3) However, the IAT indicates that smokers' global implicit attitude toward smoking is negative rather than positive; (4) The AST revealed a neutral implicit attitude toward smoking cues; (5) Both implicit and explicit attitudes toward

Acknowledgement

The authors wish to thank Bert Hoekzema for his technical assistance.

References (22)

  • M. Dols et al.

    The urge to smoke depends on the expectation of smoking

    Addiction

    (2002)
  • Cited by (0)

    The experiment reported in this paper was conducted at the Department of Medical, Clinical, and Experimental Psychology, Maastricht University, The Netherlands.

    View full text