The relational responding task and smoking : the role of implicit beliefs in smoking addiction

Graduate School of Psychology

RESEARCH MASTER’S PSYCHOLOGY THESIS

Status (1st _{draft or revision) : Revision Date: 14 / 08 / 2017}

1. WHO AND WHERE Student

Name : Giovanni A. Giaquinto

Student ID number : 10169601

Address : Jan den Haenstraat 22-1

Postal code and residence : 1055 WE Amsterdam Telephone number : 0620286294

Email address : giaquinto1991@gmail.com Supervisor(s)

Within ResMas (obligatory) : Bram van Bockstaele, Reinout Wiers

Specialisation : Clinical Psychology, Developmental Psychology Second assessor : Elske Salemink

Research center / location : UvA Roeterseiland Number of credits Thesis (1 ec = 28 hrs) : 30

At least 25 ec, Excluding 4 ec Thesis Proposal

Ethics committee reference code :

The Relational Responding Task and Smoking;

The Role of Implicit Beliefs in Smoking Addiction

Giovanni A. Giaquinto

Bram Van Bockstaele, & Reinout W. Wiers Contributions:

Helle Larsen, Helen Tibboel, & Bas van den Putte

Abstract

Despite inconsistent results, the implicit association test (IAT) is generally used to assess implicit associations. This study aimed to extend the findings of De Houwer, Heider, Spruyt, Roets, and Hughes (2015) and Tibboel, De Houwer, Dirix, and Spruyt (2016) regarding the relational responding task (RRT), which measures the automatic activation of complex propositional knowledge, by using a self-developed RRT that measured implicit motivation to quit smoking to predict smoking behavior and motivation to quit smoking.

Measures of implicit motivation to quit smoking, implicit smoker identity, explicit motivation to quit smoking, and self-efficacy regarding smoking cessation were used to predict nicotine dependence.

In line with our expectations, the RRT proved to significantly predict nicotine dependence, while the IAT did not. Furthermore, although the IAT proved to be a reliable measure, the IAT also proved that it could lead to confusing results and thus advocated the use of the RRT.

As such, using the RRT to measure implicit beliefs about quitting motivation is likely to further our understanding of smoking behavior. Future research should look into a direct comparison of an IAT and RRT that measure the same construct in order to be able to make a decisive decision about the RRT.

KEY WORDS: IAT, RRT, SMOKING, IMPLICIT COGNITION, MOTIVATION, EXPLICIT, IMPLICIT, SMOKING CESSATION

The Relational Responding Task and Smoking; The Role of Implicit Beliefs in Smoking Addiction.

Tobacco use is related to negative health outcomes and an increased risk for developing other addictive behaviors (Torabi, Bailey, & Majd-Jabbari, 1993; Larsen et al, 2014), and is thus a matter of great concern to modern society. Swayampakala et al. (2014) found that the pictorial health-warning labels on cigarette packs are associated with greater awareness of smoking-related risks. It can thus be assumed that the negative effects of tobacco use are well known. Yet, on a 2014 survey given to the Dutch population, nearly 3.5 million Dutchmen indicated to smoke on a daily basis (Hoeveel mensen in Nederland zijn verslaafd en hoeveel zijn er in behandeling?, March 13th _{2016, retrieved from}

https://www.jellinek.nl/vraag-antwoord/hoeveel-mensen-zijn-verslaafd-en-hoeveel-zijn-er-in-behandeling/). How is it that, despite

knowing about the negative effects of smoking, a huge amount of people keeps smoking? The answer to the question about this seemingly paradoxical behavior may well lie in implicit cognition.

Implicit cognition regarding addictive behavior is often measured using the implicit association test (IAT). The IAT, developed by Greenwald, McGhee, and Schwartz (1998) is a computerized reaction time sorting task that verifies an individuals’ implicit associations by letting them classify stimuli in four contrasting sorting conditions (Larsen et al., 2014). It is believed that individuals will respond faster when the contrasting sorting conditions are in line with the individuals implicit associations (Greenwald et al., 1998). Since its creation, a number of alternative IAT versions aiming to measure several implicit associations have been developed including identity associations (e.g. smoker me/ not me), affect associations (e.g. reward/relief, excitement/depression), and appetitiveness (e.g. approach/avoidance) (Lindgren et al., 2016). For example, Lindgren et al (2016) implemented a version of the IAT where participants had to classify stimuli (i.e. partyer, abstainer, drunk, my, other, etc.) into the contrasting sorting conditions drinker/ not non drinker or me-non drinker/not me-drinker in order to assess an individuals’ implicit drinker identity and use that as a prospective predictor of future alcohol consumption. They indeed

found that an individuals’ implicit drinker identity was predictive of future alcohol consumption

A bulk of studies using the IAT to assess implicit associations to smoking found that smokers have more positive attitudes towards smoking than non-smokers (e.g. McCarthy & Thompson, 2006; Huijding, de Jong, Wiers, & Verkooijen, 2005). Moreover, Rooke, Hine, and Thorsteinsson (2008) performed a meta-analysis to analyze the scope of the relationship between implicit cognition and substance use. In this meta-analysis an average effect size of r = .31 was found, showing that implicit cognition is a reliable predictor of substance use. Nevertheless, Rooke et al. (2008) also found that the average effect size when only looking at studies that used an IAT as their measurement strategy dropped to r = .18, which was the second to lowest effect size that was found (only studies using the Extrinsic Affective Simon Task scored lower

r = .15) showing that implicit attitudes might not be the best predictor of substance

use. Moreover, Larsen et al. (2014) found no differences between smokers and non-smokers, using an IAT that measured associations between smoking and positively/negatively valenced constructs. Also, Spruyt et al. (2015) examined the relationship between implicit attitudes toward smoking related cues in smokers and non-smokers using a valence IAT – participants had to classify pictures related to smoking and non smoking to positive and negative – and long-term relapse in abstaining smokers. None of the behavioral outcome measures were found to correlate with IAT scores. Finally, Robinson, Meier, Zetocha, and McCaul (2005) performed an experiment with two valence IAT’s. In the first IAT the categories smoking and nonsmoking were contrasted with each other, and participants had to classify them in the sorting conditions smoking-bad and nonsmoking-good (or nonsmoking-bad and smoking-good). With regards to this IAT Robinson et al. found that non smokers had negative smoking attitudes and smokers were ambivalent. The second IAT used smoking and stealing as the contrasting categories. When stealing was used as the contrast category, smokers and nonsmokers had identical attitudes towards smoking. Thus, from Robinson et al. (2005) it can be concluded that the contrast category used in an IAT determines the theoretical conclusions of the study. Implicit attitude thus

seems a reliable predictor of substance use, but it also seems that there is room for improvement regarding the way that we measure implicit attitudes.

A novel strategy of measuring implicit cognition is by looking at implicit beliefs. The significance of differentiating between different beliefs at the implicit level is demonstrated by the findings of Remue, De Houwer, Barnes-Holmes, Vanderhasselt, and De Raedt (2013). They found that implicit self-esteem IAT scores were equally positive for depressed patients and non-depressed control participants. Depressed patients interpreted ‘I am good’ as ‘I want to be good’, which led to the equally positive IAT scores, meaning that the association between ‘I’ and ‘Good’ was interpreted differently depending on the context of the participant. The relational responding task (RRT), a new implicit measure developed by De Houwer et al. (2015), is designed to examine automatic beliefs, in contrast to the IAT that is designed to measure automatic associations (Tibboel et al., 2016).

The RRT involves the presentation of full statements (e.g. ‘Smoking makes me unhappy) and participants are required to indicate whether these statements are true or false. In a first block the participants are explicitly instructed to act as if a general rule (e.g. answer as if you like smoking) is true and categorize the full statements they are presented with as either true or false. In a second block this rule is reversed (De Houwer et al., 2015). Thus, similar to the IAT, the RRT assigns four categories of stimuli to two responses in a way that varies across blocks. Unlike in the IAT, where participants can select the correct response by merely identifying the individual category that the presented stimulus is a member of, in the RRT participants can only select the correct response based on the way that different stimuli are related to one another. The RRT therefore requires participants to respond in a complex relational manner (De Houwer et al., 2015). It is assumed that participants will respond faster in blocks in which they are required to respond in line with the rule that matches their own beliefs (Tibboel et al., 2016). In line with this idea, De Houwer et al. created an RRT that measured individuals implicit attitudes towards immigrants by presenting them with statements about the intelligence of immigrants in regards to the intelligence of Flemish people and asking them to act as if they believed that Flemish people are more intelligent then immigrants or as if they believed that immigrants are

more intelligent then Flemish people. De Houwer et al. found that individuals’ RRT scores correlated positively with their explicit racism scores. Tibboel et al. (2016) then extended these findings by depriving smokers of smoking and have them complete a smoking urge RRT, where participants had to act as if they felt the urge to smoke or act as if they did not feel the urge to smoke, and complete a valence IAT where participants had to classify words in the contrasting categories smoking-good/not smoking-bad or smoking-bad/not smoking good. Tibboel et al. found that the IAT scores were not influenced by smoking deprivation, but that the RRT scores increased when participants were deprived of smoking indicating a stronger implicit urge to smoke. Not only do these finding once more show that the IAT might not be an ideal way of measuring implicit attitudes, but they also show that the RRT is a sensitive measure of measuring implicit beliefs.

Intrigued by these findings, we set out to test whether the RRT has an added value in research regarding implicit beliefs and smoking addiction. More specifically, we created a smoker-ideal RRT where participants had to act as if they wanted to quit smoking or act as if they did not want to quit smoking and classify statements regarding smoking cessation as true or false (e.g. I want to quit smoking). We then tested whether our smoker-ideal RRT was more predictive of motivation to quit smoking than a identity IAT and whether our ideal RRT and smoker-identity IAT were predictive of current nicotine dependence. In addition, we investigated whether the smoker-identity IAT successfully discriminated between current smokers, ex-smokers, and non-smokers. Lastly, we tested whether our smoker-ideal RRT was able to predict smokers’ explicit motivation to quit smoking and interest in participating in a smoking cessation program.

We hypothesized that a regression with the RRT-scores would explain additional variance in nicotine dependence compared to a regression with just the IAT-scores. Second, in line with Vangeli and West’s (2012) finding that former smokers did not transition from a smoker to a non-smoker identity but from a smoker to an ex-smoker identity we hypothesized that the IAT-scores would decrease gradually from smokers to ex-smokers to non-smokers. Third, we hypothesized that the RRT-scores, unlike the IAT-scores, would be able to predict explicit motivation to quit smoking.

Fourth, we hypothesized that the RRT-scores would be predictive of an individuals’ interest to participate in a smoking-intervention. Fifth and last, we hypothesized that explicit motivation to quit smoking scores, self-efficacy scores (regarding quitting attempts), and interest in joining a quitting program would correlate positively with the RRT-scores.

Method

Participants

A total of 320 students at the University of Amsterdam participated in our study as part of a large group testing session. Due to time constraints the study was performed in two seperate sessions, one week apart. Participants who missed either of the two sessions (N = 59) and/or had conflicting answers about their smoker status (i.e. they indicated to be smokers during the questionnaire session but indicated to be non-smokers during the reaction time session, or vice versa) (N = 24), were excluded from further analysis. Furthermore, due to a software bug, the data of three participants were scrambled and could not be recovered. This led to a total of 234 unique participants (Mage = 20.31, SDage = 3.07; 67 men). Of these 234 participants 47

participants were smokers (indicated to smoke at least one cigarette a day), 65 participants were ex-smokers, and 122 participants were non-smokers.

Questionnaires

Nicotine dependence was assessed with a self-translated to Dutch version of the Fagerström test for Nicotine Dependence (Fagerstrom & Schneider, 1989). This questionnaire consisted of six questions, scored on two and three point Likert scales. In the present study the Cronbach’s alpha coefficient was .73. Explicit motivation was assessed with a self-constructed scale based on an explicit motivation to quit smoking scale from Hummel et al. (2017). This scale consisted of three theory of planned behavior questions (e.g. I’m planning to quit smoking with the next 4 weeks) scored on a 5-point Likert scale and one stages of change question scored on a 9-point Likert

scale. In the present study the Cronbach’s alpha coefficient was .93. Self-efficacy was assessed with a self-constructed self-efficacy scale based on a self-efficacy scale from Hummel et al. (2017). It consisted of five questions (e.g. How easy would it be for you to quit smoking?) scored on a 5-point Likert scale. In the present study the Cronbach’s alpha coefficient was .83. Interest in joining a quitting program was assessed by a single item scored on a 5-point Likert scale.1

Implicit Smoker Identity: Implicit Association Test

The IAT, measuring smoker identity, was implemented using the INQUISIT Millisecond 4 (2014) software package. The attribute labels used for the smoker identity IAT were the Dutch words for ‘ME’, and ‘NOT ME’. We used four words related to me (Dutch translations of I, me, self, and mine) and to not me (Dutch translations of not me, they, others, and them). The target category labels were the Dutch translation of ‘SMOKING’ and ‘NOT SMOKING. The target stimuli that were used were Dutch translations of four words related to smoking (‘SMOKING’, ‘PUFFING’, ‘CIGARETTE’, and ‘SMOKER’) and Dutch translations of four words related to not smoking (‘NOT SMOKING’, ‘NOT PUFFING’, ‘NO CIGARETTE’, ‘NON SMOKER’). Participants responded by pressing either the left or right button. The IAT was an adapted version of the drinking identity IAT used in Lindgren et al. (2013) applied to smoking. It consisted of seven blocks, and was modeled after the standard 7-block procedure as described in Greenwald, McGhee, and Schwartz (1998) with one exception. Each block started with the presentation of a black screen with the relevant labels of that block for three seconds, that stayed on the screen for the remainder of the block. In the first block participants practiced categorizing the attribute stimuli, each attribute stimulus was presented three times for a total of 24 trials in the first block. The assignment of attribute labels to the left or right key was counterbalanced across participants. Participants practiced categorizing the target stimuli in the second block, each target stimulus was presented three times for a total of 24 trials in the second block. In the third block, participants practiced the compatible categorization (me-smoker versus not me-non smoker). Each stimulus was presented twice for a total

1

of 32 trials. The fourth block was the test block of the compatible categorization; each stimulus was presented four times for a total of 64 trials. The fifth block was similar to the first block, differing only in the reversed response mapping of the attribute labels. Participants practiced the incompatible categorization (non smoker versus not me-smoker) in the sixth block. Each stimulus was presented twice for a total of 32 trials. The seventh and final block was the test block for the incompatible categorization. Each stimulus was presented four times for a total of 64 trials. During the entire task, we used a random order to present the trials and the inter trial interval was 350 ms. Moreover, participants were obligated to give a correct answer in each trial, if an incorrect answer was given a red X appeared on the screen and stayed there until the participants had given the correct response. We did not counterbalance the order of the blocks, because this is known to increase error variance and lower correlations (Perugini & Banse, 2007), and our main goal was to correlate the RRT scores with other measures.

Implicit Motivation to Quit Smoking: Relational Responding Task

The RRT, measuring implicit motivation to quit smoking, was implemented using the INQUISIT Millisecond 4 (2014) software package. The RRT consisted of five blocks and followed a procedure similar to that of De Houwer et al. (2015) and Tibboel et al. (2016). The same ten inducer words that were used in De Houwer et al. (2015) were used in our study (Dutch translation of five words related to true; ‘TRUE’, ‘RIGHT’, ‘CORRECT’, ‘EXACT’, ‘OK’, and five words related to false; ‘MISS’, ‘FAULTY’, ‘INCORRECT’, ‘WRONG’, ‘FALSE’). Twenty statements were used as stimuli. The statements were all related to smoking and smoking cessation. Ten statements implied the beliefs that an individual would want to quit smoking (e.g., a Dutch translation of the statement ‘Quitting smoking would make me happy’), while the other ten statements implied the beliefs that an individual did not want to quit smoking (e.g., a Dutch translation of the statement ‘I like to keep smoking’)2_{. All of the statements} were presented in blue, while all the inducer words were presented in yellow. Each block started with a black screen with the two categorical labels TRUE (top-left corner)

2

and FALSE (top-right corner) on the screen. Similar to the implicit smoker identity IAT, participants were obligated to give the correct response in each trial. If an incorrect response was given, a red X appeared on screen until the correct response was given. In the first block participants practiced the categorization of the inducer words. Each word was presented four times, for a total of 40 trials. The second block, served as a practice block for the presented statements. Participants were instructed to act as if they wanted to quit smoking and categorize the statements according to this rule. Each statement was presented twice for a total of 40 trials. The third block served as the test block for the compatible categorization rule (true-I want to quit smoking versus false-I do not want to quit smoking). Both inducer words and statements were presented and participants had to follow the same rule as instructed in the first and second block Each inducer word was presented four times and each statement twice for a total of 80 trials. The fourth block was similar to the second block, differing only in the rule that participants had to follow. In the fourth block participants had to act

as if they did not want to quit smoking and categorize the presented statements

according to this rule. Each statement was presented twice for a total of 40 trials. The fifth and final block served as the test block for the incompatible categorization rule (true-I dont want to quit smoking versus false-I want to quit smoking). All inducer words were presented four times and all statements twice for a total of 80 trials. During the entire task, we used a random order to present the trials and the inter trial interval was 500 ms. As for the IAT, we did not counterbalance the order of the blocks based on Perugini and Banse’s (2007) findings.

Procedure

In the first session, students completed the questionnaires. The questionnaires were preceded by several questions aimed to distinguish smokers, ex-smokers, and non-smokers and to measure (if applicable) smoking behavior. Ex-smokers were asked about the average amount of cigarettes they used to smoke, when they stopped smoking, and why they stopped smoking. Smokers were asked how many cigarettes they smoked per day. All smokers that indicated to smoke on average at least one cigarette a day, completed the Fagerström test for nicotine dependence and indicated

how many cigarettes per day they smoked for the last week. This was followed by some questions aimed at measuring their explicit motivation to quit smoking, and their assumed self-efficacy regarding smoking cessation. In the second session, all participants completed a smokers’ identity IAT. After completing the smokers’ identity IAT, the participants were prompted with the question: ‘Do you, at this moment of your life, on average smoke 1 or more cigarettes per day?’. For participants who indicated not to do so, this was the end of the study. Participants, who indicated to smoke 1 or more cigarettes per day, were then redirected to the RRT. Lastly, participants were prompted with the question: ‘How interested are you in help with smoking cessation?’

Results

Scoring and Outliers

Following the procedure of both De Houwer et al. (2015) and Tibboel et al. (2016), inducer trials were used to prevent response recoding (De Houwer et al., 2015). Therefore, our data analysis was primarily focused on the target trials of the mixed blocks (block 3 and block 5) regarding RRT scores. RRT and IAT scores were calculated in two ways, first using the D-measure with built-in error penalty (D-biep) algorithm as described in Greenwald, Nosek & Banaji (2003). Trials with latency larger then 10.000 ms were deleted. Participants for whom more than ten percent of trials had latency less 300 ms were excluded from further analysis. Next, we calculated the inclusive standard deviation of all trials in block three and six, and block four and seven for the IAT-scores and for block three, and block five for the RRT-scores. This was followed by the computation of the mean latency of these blocks. For the IAT scores we then computed two mean differences by subtracting the mean latency of the compatible test block (block four) of the mean latency of the incompatible test block (block seven), and subtracting the mean latency of the compatible practice block (block three) of the mean latency of the incompatible practice block (block six). Finally, the mean differences (IAT) and mean latency (RRT) were divided by its associated inclusive standard deviation. Higher scores indicated a stronger implicit motivation to quit smoking (RRT) or a stronger smoker identity (IAT). This method

was chosen because it is the preferred measure when participants are allowed to correct errors and latency is recorded until the correct response is given.

Second we calculated the conventional log measure as described by Greenwald et al (1998). Trials with errors, trials of the practice blocks, and the first two trials of each test block were excluded. All latencies larger than 3000 ms were recoded to 3000; all latencies smaller than 300 ms were recoded to 300. The latencies were then log-transformed and the RRT/IAT-scores were calculated by subtracting the mean reaction on the compatible test block (block 4 for the IAT; block 3 for the RRT) from the mean reaction time on the incompatible test block (block 7 for the IAT; block 5 for the RRT). As for the D-measures, higher scores indicated a stronger implicit motivation to quit smoking (RRT) or a stronger smoker identity (IAT).

For all regression analyses that were performed, cases were marked as outliers and not used for the analysis if the standardized residuals were ≤ -3.00 or ≥ 3.00, the leverage score was ≥ 3 * (k + 1)/n, or the cooks distance was ≥ 4/n (where k is the number of independent variables in the model and n is the number of observations) based on Karadimitriou & Marshall (n.d.). Lastly, if ten percent of a participants’ trial latencies was below 300 ms, they were flagged and excluded from further analyses in analyses including D-scores.

A split-half reliability analysis, based on the log scores, was performed to assess the internal reliability of the IAT and RRT. With this analysis we found a correlation of

rs = .46, p < .001 for the IAT, which shows that our IAT was a reliable measure of implicit smoker identity. For the RRT we found a correlation of rs = .52, p < .001, reflecting a good reliability that was also in line with the reliability indices reported by De Houwer et al. (2015) (r = .64) and Tibboel et al. (2017) (r = .38).

Implicit Motivation, Implicit Smoker Identity, and Nicotine Dependence.

The data of all smokers in our sample was eligible for this part of the data analysis (N = 47). Four participants had a percentage of trials with latencies below 300 ms bigger than ten percent and were excluded from further analysis, which left us with a total of 43 participants for analysis including D-scores. A multiple linear regression was calculated to predict nicotine dependence based on the RRT-scores and the

IAT-scores. Based on the previously mentioned thresholds for being labeled an outlier, five participants were left out of the regression with the IAT D-scores and RRT D-scores including inducer trials scores as independent variables, four participants were left out of the regression with the log IAT scores and log RRT excluding inducer trials as independent variables, and four participants were left out of the regression with the IAT D-scores and RRT D-scores excluding inducer trials as the independent variables.

Using the log IAT scores and log RRT scores including inducer trials in a multiple regression analysis, five participants were left out of the analysis based on the outlier thresholds. Mean nicotine dependence was M = 1.71 (SD = 2.12). Using these measures, a significant regression equation was found (F(2, 39) = 5.02, p = .011) with an

R2 _{of .21. Only the log RRT including inducers score was a significant predictor of}

nicotine dependence. Comparing the model with the log RRT including inducers scores and the log IAT scores as predictors to the model with only the log IAT scores as a predictor of nicotine dependence showed a significant increase in R2 _(R2_{-change = .19,} F(1, 39) = 9.38, p = .004). See Table 1 for an overview of the model parameters. Non

significant results were found when RRT D-scores including inducer trials (F(2, 39) = 2.35, p = .109), RRT D-scores excluding inducer trials (F(2, 38) = 2.66, p = .084), or log RRT scores excluding the inducer trials (F(2, 42) = 1.39, p = .26) were used as the independent variables.

Table 1. Model Parameters of Significant Regression Analysis

Model B Std. Error Beta t Significance

1 Constant 2.32 .88 2.63 .012

log IAT -2.62 3.54 -.12 -.74 .463

2 Constant .806 .81 3.12 .003

log IAT 3.29 3.29 -.02. -.14 .886

Implicit Smoker Identity and Smoker status

The data of all participants was eligible for this part of the data analysis. Before conducting the linear contrast analysis, we checked for the assumption of normality for each level of the independent variable (Smoker, Ex-Smoker, Non-Smoker). For the IAT D-scores, the assumption was met for all levels. For the log IAT scores, the assumption of normality was not met for the Non-Smoker level of smoker status (p = .028) and the Smoker level of smoker status (p = .001) . However, since the number of participants was fairly high (N = 122 and N = 47 respectively) and the Q-Q plots looked fairly normal, this should not influence the conclusion.

The one-way ANOVA with smoker status (Smoker, Ex-Smoker, Non-Smoker) as the independent variable and IAT D-scores as the dependent variable showed a significant effect of smoker status on IAT D-scores FWelch(2, 128.50) = 12.96, p < .001. Levene’s test indicated unequal variances (F = 5.40, p = .005) so degrees of freedom were adjusted from 231 to 128.50. Planned contrasts revealed that smokers had a significant stronger smoker identity than ex-smokers (t(108.48) = 2.89, p = 0.005) and a significant stronger smoker identity than non-smokers (t(132.08) = 5.04, p < .001). There was no significant difference between ex-smokers and non-smokers t(132.62) = 1.37, p = .17. For mean IAT D-scores see Table 2. A linear trend analysis showed however that a significant linear trend was present t(111.51) = 4.95, p <.001. See Figure 1 for a means plot of the IAT D-scores.

The one-way ANOVA with smoker status (Smoker, Ex-Smoker, Non-Smoker) as the independent variable and log IAT scores as the dependent variable showed a significant effect of smoker status on log IAT scores F(2, 231) = 6.50, p = .002. Planned contrasts revealed that smokers had a significant stronger smoker identity than ex-smokers (t(231) = 2.19, p = .029) and a significant stronger smoker identity than non-smokers (t(231) = 3.60, p < .001). There was no significant difference between ex-smokers and non-ex-smokers t(231) = 1.29, p = .198. For mean log IAT scores see Table 2. A linear trend analysis showed however that a significant linear trend was present F(1, 231) = 12.96, p <.001. See Figure 1 for a means plot of the log IAT scores.

Table 2. One-sample T-tests for Log-scores and D-scores

M (SD) T df

Log scores Smoker .23 (.11) 14.18** 46

Ex-smoker .17 (.12) 11.90** 46

Non-smoker .15 (.14) 11.42** 64

D-scores Smoker .64 (.20) 21.58** 64

Ex-smoker .50 (.32) 12.58** 121

Non-smoker .43 (.32) 14.64** 121

^ One-sample t-test tested against a value of zero ** p < .001

Implicit Motivation to Quit Smoking, Explicit Motivation to Quit Smoking, and Implicit Smoker Identity

The data of the smokers in our sample was eligible for this part of the data analysis (N = 46, one participants’ answer did not get recorded). We calculated a regression equation with explicit motivation to quit smoking scores as the dependent variable and RRT D-scores and IAT D-scores as the independent variables. On the aforementioned outlier thresholds, nine participants were excluded from further analysis (percentage of > 10% for latencies below 300 ms, N = 4), which left us with a total of 38 participants for the analysis. Mean explicit motivation was 2.67 (SD = 1.16), mean implicit motivation to quit smoking was -.20 (SD = .29), and mean implicit

smoker identity was .62 (SD = .17). No significant regression equation was found F(2, 35) = .139, p = .870. We also calculated a regression equation with explicit motivation to quit smoking scores as the dependent variable and RRT log scores and IAT log scores as the independent variables. Based on the outlier thresholds four participants were excluded from this analysis. No significant regression equation was found F(2,39) = .092, p = .912.

Correlations

Before the correlational analysis, a test of normality was performed to check for the assumption of normality. This assumption was not met for the nicotine dependence scores (Shapiro-Wilk = .82, p < .001), explicit motivation to quit smoking scores Wilk = .88, P < .001), number of cigarettes smoked per week (Shapiro-Wilk = .89, p < .001), log IAT scores (Shapiro-(Shapiro-Wilk = .94, p = .025), log RRT scores (Shapiro-Wilk = .84, p < .001), RRT D-scores (Shapiro-Wilk = .93, p = .006), and interest in participating in a quitting program scores (Shapiro-Wilk = .76, p < .001). Table 2 shows that the nicotine dependence scores were significantly positively correlated with the log RRT scores (rspearman = .34, p = .021) and the number of

cigarettes smoked per week (rspearman =.73, p < .001); number of cigarettes per week was

significantly negatively correlated with the self-efficacy scores (rspearman = -.29, p =

.049); interest in participating in a quitting program scores were significantly positively correlated with explicit motivation to quit smoking scores (rspearman = .50, p

Table 3. Non-Parametric Correlation Matrix

Self-Efficacy Explicit

Motivation Number of cigarettes Interest Log IAT scores IAT D-scores Log RRT scores RRT D-scores

Nicotine Dependence -.27 .24 .73** .09 -.11 -.25 .34* .25

Self Efficacy - -.10 -.29* -.13 .04 -.04 -.01 .02

Explicit Motivation - .07 .50** .01 .04 -.12 -.06

Number of cigarettes - .11 .12 .00 .12 .10

Interest - -.09 -.02 -.10 .02

Log IAT scores - .62** -.11 .02

IAT D-scores - -.34* -.15

Log RRT scores - .86**

* p < .05 ** p < .001

Supplementary analyses

Supplementary analyses were conducted in order to investigate the diverging results – a significant regression equation with the log RRT scores including inducer trials versus a non significant regression equation with the RRT D-scores including inducer trials – that we obtained with the log and D-scores. The main differences between the log and D-scores are the removal of the first two trials of each test block, the exclusion of participants for whom more than ten percent of trials had latency less 300 ms, and the recoding of latencies larger than 3000 ms and smaller than 300 ms

First, we recalculated the D-scores by first removing the first two trials of each test block and then following the procedure that was described above. This did however not yield similar results. Second, we recalculated the D-scores by recoding the latencies as was done for the IAT, again no similar results were found.

Third, depending on the variables that were put in the regression, different participants were marked as outliers and not used for the analysis. We therefore tried excluding the same participants that were excluded in the multiple regression analysis with log scores as the independent variable from the multiple regression analysis with the D-scores including inducer trials as the independent variable; Using these criteria in a multiple regression analysis, a significant regression equation was found (F(2,39) = 3.72, p = 0.033) with an R2 _{of .12. As was found before, only the RRT D-score was a} significant predictor of nicotine dependence. Therefore, it seems likely that the outlier criteria are the cause of the diverging results between the log measures and the D-measures.

Discussion

In this study, measures of implicit motivation to quit smoking and implicit smoker identity were used to predict individuals’ nicotine dependence, explicit motivation to quit smoking, and interest in help with smoking cessation. Moreover, we tested whether the strength of an implicit smoker identity differed between smokers, ex-smokers, and non-smokers.

Using the conventional log measure we found that implicit motivation to quit smoking significantly predicted nicotine dependence. Higher implicit motivation to

quit smoking was related to higher nicotine dependence. In contrast, implicit smoker identity was not predictive of nicotine dependence. These results support the idea, suggested by de Houwer et al (2015), that the RRT can add to our understanding of implicit measures and that it has an added value compared to the regular IAT

Regarding the differing strength of implicit smoker identity we found that smokers indeed had a stronger implicit smoker identity than ex-smokers and non-smokers. Contrary to our expectations, ex-smokers did not have a stronger implicit smoker identity than non-smokers. These findings oppose previous findings regarding explicit smoker identity (e.g. van den Putte et al., 2009). In contrast to our expectancies, all participants had stronger associations between self and smoking than between self and not smoking. Meaning that even if an individual did not smoke, he or she identified more as a smoker than as a non-smoker. These results support the idea that although the IAT is a good measure of implicit smoker identity, there still is room for improvement. As Blanton, Jaccard, Christie, and Gonzales (2007) criticized, one area of improvement for the IAT is the scoring of the IAT. Blanton et al.

(2007) noted that the scores on the compatible block and the incompatible block should be negatively correlated, based on the remark of Maison, Greenwald, and Bruin (2001) that reaction times in the compatible block should be shortened and reaction times in incompatible blocks should be prolonged (regarding log scores). Blanton et al. (2007) tested this remark and their paper and found no support. As Blanton et al. (2007), we too tested for a negative correlation between the incompatible and compatible block reaction times and found no support for this remark (r = .10, p = .104). Although the D score was developed as an improvement of the log score, these scores too have flaws. As Blanton et al. (2007) note, the D scores were designed to minimize the correlation between the IAT scores and an index of general processing speed in Greenwald’s et al. (2003) research to factor out the effect of general processing speed. Unless a researcher examines the general processing speed in their research, they will not know whether this also works for their research.

Third and last, we found a significant association between explicit motivation to quit smoking and interest in help with smoking cessation. A higher explicit motivation to quit was associated with more interest in help with smoking cessation. In contrast

to our expectations, we found no association between implicit motivation to quit smoking and interest in help with smoking cessation, nor between implicit and explicit motivation to quit smoking. According to Banaji (2001) however, this is not a surprising results. She argues that a low correlation between implicit and explicit measures within ‘the same family’ should be seen as evidence of validity, and that implicit and explicit measures should be independent of each other.

As we pointed out earlier, not only was the IAT unable to predict nicotine dependence, but non-smokers also had stronger associations between the self and smoking than between the self and not-smoking, even though they had the least positive association between the self and smoking. One possible explanation for the unexpected results of the IAT could be the way that a smoker identity arises. A meta-analysis by Tombor et al. (2015) found that many factors influence an individuals’ smoker identity (e.g. social environment, defensive rationalizations, possibility of a shift between identities), for which we did not control. For example, Levinson et al (2007) found that more than half of the students that participated in their study denied being smokers despite their smoking behavior. These ‘deniers’, as they were dubbed, were found to be prone to infrequently smoking and saying they were not addicted to cigarettes. It is therefore likely that, despite our efforts, some participants who were labeled as ex-smoker or non-smoker, should have been labeled smokers. Although we found a significant regression equation when trying to predict nicotine dependence, we also found in our supplementary analysis that the difference in outliers led to different results. It is therefore possible that the outlier thresholds that we used were not adequate. Our motivation for these outliers criteria was to keep the exclusion criteria as unbiased as possible. Several sources (e.g. Stevens, 1984; Cook, 1977; Chatterjee & Hadi, 1986) pointed out that a Cooks distance higher than a certain value might indicate a problematic case and the same goes for Leverage values. However, few sources then elaborate on when a possible problematic case, is actually problematic. Also, a lot of studies don’t even mention any outlier criteria regarding their regression analyses (e.g. Dierker, Hedeker, Rose, Selya, & Mermelstein, 2015; Lindberg et al., 2015; Kurti, Davis, Skelly, Redner, & Higgins, 2016). Because there seems – as far as we know – to be no clear consensus on what makes a possible

problematic case truly problematic, we decided to use the strict unbiased rule that we used to exclude participants from the analysis.

Furthermore, although some promising results were found in the multiple regression analysis regarding the RRT, for the most part no significant results were found. The fact that no relation was found between the implicit and explicit measures of motivation to quit smoking is provocative, but not necessarily problematic. As mentioned earlier, follow Banaji’s (2001) reasoning it is not absolutely necessary for implicit and explicit measures to correlate highly. If implicit and explicit measures would correlate highly, the need for implicit measures would be nullified. After all, why would one use implicit measures if explicit measures measure the exact same thing. Different explanations for the absence of a significant correlation exist however. The most likely explanation seems to be that our measure of explicit motivation did not measure explicit motivation but instead measured explicit intention to quit smoking. Our explicit motivation to quit smoking questions were phrased as ‘I’m

planning to quit smoking’, while the RRT rule was phrased as ‘I want to quit smoking’.

Indeed, several studies make a distinction between intention and motivation (e.g. Athamneh, Essien, Sansgiry, & Abughosh, 2015; Herzog, Pokhrel, & Kawamoto, 2014). This argument is especially compelling since the strongest feature of the RRT is that it uses the complex relationship between two concepts, making it likely that this is the cause of the lack of a relation.

Another problem that we encountered during our data analysis was that, similar to the study of Tibboel et al. (2016), most smokers in our study were considered to be ‘light’ smokers (30 out of 59). Even though no clear consensus exists about what makes a smoker a light smoker (Husten, 2009; Tibboel, 2016) according to our Fägerstrom test for nicotine dependence most smokers had low dependency. It is possible that our results were clouded, due to the high volume of low dependence smokers. We therefore decided to perform one more supplementary analysis with a binary dummy variable for nicotine dependence based on the norm scores of the Fägerstrom test for nicotine dependence (low dependence vs. the rest). We then performed a binary logistic regression to see if we would find any different results. No significant results were found, but the effects did approach significance (p = .063).

A second possible explanation for our results could be that our threshold to code someone as a smoker was to low. As mentioned before, if participants smoked at least one cigarette per day they were coded as smokers. In the similar study by Tibboel et al. (2016), smokers had to smoke at least 10 cigarettes per day. In future studies a stricter threshold, similar to the one just in Tibboel et al (2017), should be used.

Finally, our study had some limitations. First, due to the fact that our study was fitted within the limited time-constraints of large group testings, our study was cut in half and performed in two parts. As such, we lost a considerable amount of data because some participants only participated in one of the two session. Second a measure of explicit smoker identity was missing in our study. We therefore had no possibility to check wheter the IAT actually measured smoking identity.

In sum, our research shows that the relational responding task could be a useful tool when investigating implicit cognitions. Although the smoker-identity IAT could differentiate between groups of smokers, and ex-smokers and non-smokers, the RRT was the only significant predictor of nicotine dependence. As such, using the RRT to measure implicit beliefs about quitting motivation is likely to further our understanding of smoking behavior. Future research should look into a direct comparison of an IAT and RRT that measure the same construct in order to be able to make a decisive decision about the RRT.

References

Athamneh, L., Essien, E. J., Sansgiry, S. S., & Abughosh, S. (2017). Intention to quit water pipe smoking among Arab Americans: Application of the theory of planned behavior. Journal of Ethnicity in Substance Abuse, 16(1), 80-90. doi: 10.1080/15332640.2015.1088423.

Banaji, M. R. Implicit attitudes can be measured in Roediger H.L. ,Nairne, J.S., Neath, I., Surprenant, A. (Eds.), The nature of remembering: Essays in honor of Robert G.

Crowder, American Psychological Association, Washington, DC (2001), 117-150

Blanton, H., Jaccard, J., Christie, C., & Gonzales, P. M. (2007). Plausible assumptions, questionable assumptions and post hoc rationalizations: Will the real IAT, please stand up?. Journal of Experimental Social Psychology, 43(3), 399-409. doi: 10.1016/j.jesp.2006.10.019.

Chatterjee, S., & Hadi, A. S. (1986). Influential observations, high leverage points, and outliers in linear regression. Statistical Science, 379-393.

Cook, R. D. (1977). Detection of influential observation in linear regression.

Technometrics, 19(1), 15-18.

De Houwer, J., Heider, N., Spruyt, A., Roets, A., & Hughes, S. (2015). The relational responding task: toward a new implicit measure of beliefs. Frontiers in

Psychology, 6, 319. doi:10.3389/fpsyg.2015.00319

Dierker, L., Hedeker, D., Rose, J., Selya, A., & Mermelstein, R. (2015). Early emerging nicotine dependence symptoms in adolescence predict daily smoking in young adulthood. Drug and Alcohol Dependence, 151, 267-271. doi: 10.1016/j.drugalcdep.2015.03.009

Fagerstrom, K. O., & Schneider, N. G. (1989). Measuring nicotine dependence: a review of the Fagerstrom Tolerance Questionnaire. Journal of Behavioral Medicine,

12(2), 159-182. doi: 10.1007/BF00846549

Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. (1998). Measuring individual differences in implicit cognition: the implicit association test. Journal of

Personality and Social Psychology, 74(6), 1464.

Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding and using the implicit association test: I. An improved scoring algorithm. Journal of

Personality and Social Psychology, 85(2), 197-216. doi: 10.1037/0022-3514.85.2.197

Herzog, T., Pokhrel, P., & Kawamoto, C. T. (2015). Short-term fluctuations in motivation to quit smoking in a sample of smokers in Hawaii. Substance Use &

Huijding, J., de Jong, P. J., Wiers, R. W., & Verkooijen, K. (2005). Implicit and explicit attitudes toward smoking in a smoking and a nonsmoking setting. Addictive

Behaviors, 30(5), 949-961. doi: 10.1016/j.addbeh.2004.09.014

Hummel, K., Candel, M. J., Nagelhout, G. E., Brown, J., van den Putte, B., Kotz, D., ... & de Vries, H. (2017). Construct and predictive validity of three measures of intention to quit smoking: Findings from the International Tobacco Control (ITC) Netherlands Survey. Nicotine & Tobacco Research, ntx092. doi: 10.1093/ntr/ntx092

Husten, C. G. (2009). How should we define light or intermittent smoking? Does it matter? Nicotine & Tobacco Research, 11(2), 111-121. doi: 10.1093/ntr/ntp010

Inquisit 4.0 [Computer software]. (2014). Retrieved from http://www.millisecond.com

Karadimitriou, S. M., & Marshall, E. (n.d.). Outliers, Durbin-Watson and interactions for regression in SPSS. Retrieved from http://statstutor.ac.uk

Kurti, A. N., Davis, D. R., Skelly, J. M., Redner, R., & Higgins, S. T. (2016). Comparison of nicotine dependence indicators in predicting quitting among pregnant smokers. Experimental and Clinical Psychopharmacology, 24(1), 12-17. doi: 10.1037/pha0000056

Larsen, H., Kong, G., Becker, D., Cousijn, J., Boendermaker, W., Cavallo, D., ... & Wiers, R. (2014). Implicit motivational processes underlying smoking in American and Dutch adolescents. Frontiers in Psychiatry, 5, 1-9. doi: 10.3389/fpsyt.2014.00051

Levinson, A. H., Campo, S., Gascoigne, J., Jolly, O., Zakharyan, A., & Vu Tran, Z. (2007). Smoking, but not smokers: identity among college students who smoke cigarettes. Nicotine & Tobacco Research, 9(8), 845-852. doi: 10.1080/14622200701484987

Lindberg, A., Niska, B., Stridsman, C., Eklund, B. M., Eriksson, B., & Hedman, L. (2015). Low nicotine dependence and high self-efficacy can predict smoking cessation independent of the presence of chronic obstructive pulmonary disease: a three year follow up of a population-based study. Tobacco Induced Diseases, 13(1), 27. doi: 10.1186/s12971-015-0055-6

Lindgren, K. P., Foster, D. W., Westgate, E. C., & Neighbors, C. (2013). Implicit drinking identity: Drinker+me associations predict college student drinking consistently. Addictive Behaviors, 38(5), 2163–2166. doi: 10.1016/j.addbeh.2013.01.026

Lindgren, K. P., Neighbors, C., Teachman, B. A., Baldwin, S. A., Norris, J., Kaysen, D., ... & Wiers, R. W. (2016). Implicit alcohol associations, especially drinking identity, predict drinking over time. Health Psychology, 35(8), 908-918. doi: 10.1037/hea0000396

Maison, D., Greenwald, A. G., & Bruin, R. (2001). The Implicit Association Test as a measure of implicit consumer attitudes. Polish Psychological Bulletin, 2, 61-79. doi: 10.1066/S10012010002

McCarthy, D. M., & Thompsen, D. M. (2006). Implicit and explicit measures of alcohol and smoking cognitions. Psychology of Addictive Behaviors, 20(4), 436-444. doi: 10.1037/0893-164X.20.4.436

Perugini, M., & Banse, R. (2007). Personality, implicit self-concept and automaticity.

European Journal of Personality, 21(3), 257-261. doi: 10.1002/per.637

Remue, J., De Houwer, J., Barnes-Holmes, D., Vanderhasselt, M. A., & De Raedt, R. (2013). Self-esteem revisited: Performance on the implicit relational =assessment procedure as a measure of self-versus ideal self-related cognitions in dysphoria.

Cognition & Emotion, 27(8), 1441-1449. doi: 10.1080/02699931.2013.786681

Robinson, M. D., Meier, B. P., Zetocha, K. J., & McCaul, K. D. (2005). Smoking and the Implicit Association Test: When the contrast category determines the theoretical conclusions. Basic and Applied Social Psychology, 27(3), 201-212. doi: 10.1207/s15324834basp2703_2

Rooke, S. E., Hine, D. W., & Thorsteinsson, E. B. (2008). Implicit cognition and substance use: a meta-analysis. Addictive Behaviors, 33(10), 1314-1328. doi: 10.1016/j.addbeh.2008.06.009

Spruyt, A., Lemaigre, V., Salhi, B., Van Gucht, D., Tibboel, H., Van Bockstaele, B., ... & Nackaerts, K. (2015). Implicit attitudes towards smoking predict long-term relapse in abstinent smokers. Psychopharmacology, 232(14), 2551-2561. doi: 10.1007/s00213-015-3893-2

Stacy, A. W., & Wiers, R. W. (2010). Implicit cognition and addiction: a tool for explaining paradoxical behavior. Annual Review of Clinical Psychology, 6, 551-575. doi: 10.1146/annurev.clinpsy.121208.131444

Stevens, J. P. (1984). Outliers and influential data points in regression analysis.

Psychological Bulletin, 95(2), 334-344.

Swayampakala, K., Thrasher, J. F., Hammond, D., Yong, H. H., Bansal-Travers, M., Krugman, D., ... & Hardin, J. (2014). Pictorial health warning label content and smokers’ understanding of smoking-related risks—a cross-country comparison.

Health Education Research, 30 (1), 35-45. doi: doi.org/10.1093/her/cyu022

Tombor, I., Shahab, L., Herbec, A., Neale, J., Michie, S., & West, R. (2015). Smoker identity and its potential role in young adults’ smoking behavior: A meta-ethnography. Health Psychology, 34(10), 992-1003. doi: 10.1037/hea0000191

Torabi, M. R., Bailey, W. J., & Majd-Jabbari, M. (1993). Cigarette smoking as a predictor of alcohol and other drug use by children and adolescents: evidence of the “gateway drug effect”. Journal of School Health, 63(7), 302-306.

Tibboel, H., De Houwer, J., Dirix, N., & Spruyt, A. (2016). Beyond associations: Do implicit beliefs play a role in smoking addiction? Journal of

Psychopharmacology, 31(1), 43-53. doi: 10.1177/0269881116665327

Vangeli, E., & West, R. (2012). Transition towards a ‘non-smoker’ identity following smoking cessation: An interpretative phenomenological analysis. British Journal

of Health Psychology, 17(1), 171-184. doi: 10.1111/j.2044-8287.2011.02031.x

van den Putte, B., Yzer, M., Willemsen, M. C., & de Bruijn, G. J. (2009). The effects of smoking self-identity and quitting self-identity on attempts to quit smoking.

Health Psychology, 28(5), 535-544. doi: 10.1037/a0015199

Appendix A

Synonyms of not smoking: • Niet roken • Niet paffen • Geen sigaret • Niet roker Appendix B. Appendix B – IAT Start instructions:

In deze taak zullen stimuli één voor één op het scherm verschijnen.

Jouw taak bestaat erin om op de linker (A) of rechter (L) toets te duwen afhankelijk van de categorie waartoe de stimulus behoort. Er zijn vier mogelijke categorieën: (1) IK: Woorden die verwijzen naar jezelf

(2) NIET IK: Woorden die verwijzen naar anderen (3) ROKEN: Woorden die gerelateerd zijn aan roken

(4) NIET ROKEN: Woorden die gerelateerd zijn aan niet roken

Welke categorieën aan welke knop zijn toegewezen, verschilt van fase tot fase. De categorieën die aan de linkerknop (A-toets) zijn toegewezen, zullen in de linker-bovenhoek van het scherm staan.

De categorieën die aan de rechterknop (L-toets) zijn toegewezen, zullen in de rechter-bovenhoek van het scherm staan.

Probeer steeds zo snel mogelijk te antwoorden zonder te veel fouten te maken. Block 1 instructions:

Dit is een fase met enkel IK en NIET IK. IK: Druk op de linker toets (A)

NIET IK: Druk op de rechter toets (L) Block 2 instructions:

Dit is een fase met enkel ROKEN en NIET ROKEN ROKEN: Druk op de linker toets (A)

NIET ROKEN: Druk op de rechter toets (L) Block 3 & 4 instructions:

Dit is een fase met alle stimuli. IK: Druk op de linker toets (A)

NIET IK: Druk op de rechter toets (L) ROKEN: Druk op de linker toets (A)

Synonyms of other: • Niet ik • Zij • Anderen • Hun Synonyms of smoking: • Roken • Paffen • Sigaret • Roker Synonyms of me: • Ik • Mij • Zelf • Mijn

NIET ROKEN: Druk op de rechter toets (L) Block 5 instructions:

Dit is een fase met enkel ROKEN en NIET ROKEN ROKEN: Druk op de rechter toets (L)

NIET ROKEN: Druk op de linker toets (A) Block 6 & 7 instructions:

Dit is een fase met alle stimuli. IK: Druk op de linker toets (A)

NIET IK: Druk op de rechter toets (L) NIET ROKEN: Druk op de linker toets (A) ROKEN: Druk op de rechter toets (L) Appendix B – RRT Synonyms of True: • Goed • Juist • Correct • Exact • In orde Want to quit-statements:

• "IK WIL STOPPEN MET ROKEN"

• "IK WIL VAN MIJN BEHOEFTE OM TE ROKEN AF" • "HET ENIGE WAT IK NU WIL IS STOPPEN MET ROKEN" • "VAN BLIJVEN ROKEN WORD IK NIET BLIJ"

• "STOPPEN MET ROKEN ZOU ME GELUKKIG MAKEN" • "MIJN BEHOEFTE OM TE ROKEN STOORT MIJ"

• "IK WORD ONGELUKKIG VAN ROKEN"

• "IK STEL MEZELF TELEUR ALS IK BLIJF ROKEN" • "IK WIL NIET MEER ROKEN"

• "BLIJVEN ROKEN PAST NIET BIJ WIE IK BEN"

Don’t want to quit-statements:

• "IK WIL NIET STOPPEN MET ROKEN"

• "HET ENIGE WAT IK NU NIET WIL, IS STOPPEN MET ROKEN" • "STOPPEN MET ROKEN STAAT NIET IN MIJN AGENDA" • "STOPPEN MET ROKEN ZOU ME ONGELUKKIG MAKEN" • "IK GENIET VAN MIJN BEHOEFTE OM TE ROKEN"

• "VAN BLIJVEN ROKEN WORD IK BLIJ"

• "ROKEN IS IETS WAT IK GRAAG BLIJF DOEN" • "IK VIND HET LEUK OM TE BLIJVEN ROKEN"

• "IK VOEL ME GOED BIJ HET IDEE OM TE BLIJVEN ROKEN" • "BLIJVEN ROKEN PAST BIJ MIJ"

Synonyms of False: • Mis • Onjuist • Incorrect • Verkeerd • Fout

Block 1 Instructions:

Welkom en bedankt voor je deelname aan dit experiment.

In dit reactietijdexperiment zal je moeten reageren op woorden en stellingen die aangeboden worden op het scherm.

In de eerstvolgende fase zullen woorden aangeboden worden in het GEEL.

Het is jouw taak om zo snel mogelijk te bepalen of het woord verwijst naar WAAR of naar NIET WAAR.

Druk op de A-toets als het woord een synoniem van WAAR is. Druk op de L-toets als het woord een synoniem van NIET WAAR is.

Mocht je een fout maken, druk dan alsnog zo snel mogelijk de juiste toets in!

Leg je wijsvingers op de A- en L-toetsen en druk op Enter om te starten.

Block 2 Instructions: Heel goed!^^

In de volgende fase zullen alleen stellingen worden aangeboden. Deze stellingen worden in het BLAUW aangeboden en gaan altijd over stoppen met roken.

Als je een stelling in het BLAUW ziet, moet je reageren ALSOF je GRAAG wilt stoppen met roken.

Druk op de linker toets (A-toets) als de stelling volgens de bovengenoemde regel WAAR is. Druk op de rechter toets (L-toets) als de stelling volgens de bovengenoemde regel NIET WAAR is.

Block 3 Instructions: Goed gedaan!

Dit is een fase met zowel woorden als stellingen.

Losse woorden zullen in het GEEL worden aangeboden terwijl stellingen in het BLAUW zullen worden aangeboden.

Als je een stelling in het BLAUW ziet, moet je reageren ALSOF je GRAAG wilt stoppen met roken.

Druk op de linker toets (A-toets) als de stelling volgens de bovengenoemde regel WAAR is. Druk op de rechter toets (L-toets) als de stelling volgens de bovengenoemde regel NIET WAAR is.

De regel voor de GELE woorden blijft ook hetzelfde.

Druk op de A-toets als het woord een synoniem van WAAR is. Druk op de L-toets als het woord een synoniem van NIET WAAR is. Block 4 Instructions:

Heel goed!

In de volgende fase zullen opnieuw alleen stellingen worden aangeboden.

Als je een stelling in het BLAUW ziet, moet je reageren ALSOF je NIET wilt stoppen met roken.

Druk op de linker toets (A-toets) als de stelling volgens de bovengenoemde regel WAAR is. Druk op de rechter toets (L-toets) als de stelling volgens de bovengenoemde regel NIET WAAR is.

Als je bijvoorbeeld de stelling 'Ik wil niet stoppen met roken' te zien krijgt, dien je op de A-toets te drukken.

Als je bijvoorbeeld de stelling 'Ik wil stoppen met roken' te zien krijgt, dien je op de L-toets te drukken.

Block 5 instructions: Goed gedaan!

Dit is opnieuw een fase met zowel woorden als stellingen.

Losse woorden zullen in het GEEL worden aangeboden terwijl stellingen in het BLAUW zullen worden aangeboden.

Als je een stelling in het BLAUW ziet, moet je reageren ALSOF je NIET wilt stoppen met roken.

Druk op de linker toets (A-toets) als de stelling volgens de bovengenoemde regel WAAR is. Druk op de rechter toets (L-toets) als de stelling volgens de bovengenoemde regel NIET WAAR is.

De regel voor de GELE woorden verandert NIET.

Druk op de A-toets als het woord een synoniem van WAAR is. Druk op de L-toets als het woord een synoniem van NIET WAAR is.