Smokers’ Thoughts and Worries: A Study Using Ecological Momentary Assessment

Abstract

Objective:

The purpose of this study was to investigate smokers’ thoughts and worries about their smoking behavior. Researchers have sometimes asked smokers to make such self-assessments but typically using retrospective summary judgments.

Design:

Using ecological momentary assessment, community and student smokers reported five times daily during two separate 1-week intervals.

Main Outcome Measures:

Smokers reported their thoughts about smoking, worries about smoking, and level of contemplation to quit smoking.

Results:

Smokers reported thinking negatively about their smoking 26.8% of the time they had a cigarette. The most frequent thoughts reported by smokers related to immediate reinforcement of smoking (e.g., “How I smell like cigarettes”). However, smokers reported more intense worry about thoughts related to health concerns (e.g., “Symptoms I’m having because of smoking”). The occurrence of negative thoughts was significantly and positively related to contemplation about quitting, worry about smoking, and risk perceptions. Finally, self-reported worry intensity was more strongly related to contemplation of quitting than negative thought occurrence.

Conclusion:

Our results show that thoughts about smoking (i.e., cognitions) and feelings about smoking (i.e., worry) are loosely connected and it is feelings rather than cognitions that are most related to contemplation to quit.

Persons who smoke cigarettes in the United States likely confront negative reactions to their behavior many times each day. People may nag them to quit, they could see ads about the dangers of second-hand smoke, they could be forced to leave a smoke-free building to have a cigarette, and they might experience symptoms possibly caused by their smoking. Given the almost constant state of dissonance that most smokers must experience, one wonders what negative thoughts occupy them and whether they worry about those thoughts. These questions are important because having certain negative thoughts about smoking is related to motivation to quit (Curry, Grothaus, & McBride, 1997). Similarly, some data suggest that worrying about smoking is related to motivation to quit (Dijkstra & Brosschot, 2003; Manfredi, Pratt Lacey, Warnecke, & Petraitis, 1998). In turn, motivation to quit predicts intentions and plans to quit (e.g., Kozlowski et al., 1999) which, of course, are crucial for smoking cessation. Therefore, if we can learn more about what smokers think and worry about, we might be able to use this information to encourage smokers’ movements toward the quitting stage. Increasing smokers’ motivation to quit is an important goal and has been called a “tremendous and virtually untapped opportunity” (Furmanski, 2003).

The purpose of this study was to provide evidence about smokers’ negative daily thoughts and worries and to connect such data to reports of motivation to quit. We collected these data from college students and community smokers who were asked to report their thoughts and feelings five times daily during two different weeks. To explain the importance of this work, we next consider what is presently known about smokers’ thoughts and worries, and we explain our rationale for relying on ecological momentary assessment (EMA) methodology.

Smokers’ Thoughts

Most research connecting thoughts to quitting motivation has examined smokers’ beliefs rather than their naturally occurring thoughts. For example, Clark, Rakowski, Kviz, and Hogan (1997) correlated participant beliefs with motivation to quit. They found the only belief that consistently predicted desire to quit was the conviction that smoking harms health. Similarly, Hymowitz et al. (1997) interviewed ex-smokers who had quit in the previous five years, asking them about the reasons why they quit. Nearly all (90%) agreed that health concern was paramount, with the second most frequently acknowledged reason being cost (53%). Concern for others, including the effects of passive smoking (52%), and setting a good example (52%) were the next most important reasons.

Curry and colleagues (1997) categorized 20 reasons for quitting into two classes: intrinsic reasons (health concerns and self-control issues) and extrinsic reasons (immediate reinforcement and social pressure). They asked smokers how often they had these different quitting thoughts, proposing that smokers motivated to quit for intrinsic reasons would be more likely to stay quit and they observed this effect one year following cessation (see Rose, Chassin, Presson, & Sherman, 1996). In the present study, we relied primarily on the reasons for quitting generated by Curry and her colleagues to construct a “thought list” for participants to report about daily.

Smokers’ Worries

Borkovec, Robinson, Pruzinsky, and Dupree (1983) conceived of worry as a chain of thoughts and images, which are relatively uncontrollable and accompanied by negative affect. Reports of the time spent worrying each day are significantly associated with trait and state anxiety (Borkovec et al., 1983; Metzger, Miller, Cohen, Sofka, & Borkovec, 1990), and trait worry correlates with anxiety and depression (Boehnke, Schwartz, Stromberg, & Sagiv, 1998). However, nonpathological worry is associated with constructive problem solving (Davey, 1994). Although people sometimes say that worrying can “make things worse,” they also report important benefits of worrying—motivating action and stimulating preparatory and analytic thinking (Tallis, Davey, & Capuzzo, 1994). These latter data suggest that smokers who worry about their smoking might also be more likely to quit as a problem-solving strategy. In line with that speculation, McCaul et al. (2006) observed modest correlations between worry and motivation to quit.

But do smokers worry about their smoking behavior? Despite the many studies of smokers over the last 50 years, a good answer to this question is unavailable, because investigators have rarely asked direct questions about worry. We have no idea when, how often, or how intensely smokers think and worry about the consequences of their smoking. Similarly, we have no knowledge of what things smokers worry about most. This study will begin to redress the knowledge gaps concerning thoughts about smoking and worry.

Ecological Self-Monitoring

Most researchers who have investigated smokers’ thoughts relied on the typical methodology of having participants judge, retrospectively, whether and how often they had certain thoughts. Research conducted over the last decade has shown that such recall measures are often inaccurate (Stone & Shiffman, 1994) but that measurement validity can be improved using EMA, a technique in which people report their thoughts, feelings, and behavior as they are experienced. Although researchers have not asked the types of questions we are interested in here, they have used EMA successfully to address other thoughts that smokers have. Salkovskis and Reynolds (1994), for example, recruited smokers in the midst of trying to quit or cut down and asked them about the intrusions they were experiencing. Smoking-related images and thoughts (e.g., “I would really enjoy a cigarette”) occurred frequently and intensely in this group. Shiffman and colleagues (2000) employed palm-top computers to ask smokers to report self-efficacy after a quit attempt, demonstrating that self-efficacy predicted risk of relapse. Shiffman’s work is especially important for the present research because it demonstrates the value of self-monitoring. Still, we wish to ask different questions: “What thoughts do smokers have and which of those thoughts worry them?”

In summary, the literature suggests that the thoughts and worries that smokers have about their behavior are probably related to their motivation to quit. The present study begins to fill an important gap in our knowledge by assessing smokers’ thoughts and worries in their natural environment and relating these variables to motivation to quit. This study also emphasizes the worry that accompanies thoughts about negative consequences of smoking, and we contrast thoughts and worry as they relate to smoking. Specifically, we asked, (1) how often and in what circumstances do cigarette smokers think about negative aspects of their smoking behavior? (2) which negative aspects of smoking are most prevalent in smokers’ thoughts? (3) how do negative thoughts relate to how smokers feel about smoking? In particular, do cigarette smokers worry about their smoking behavior? (4) how do negative thoughts and worry relate to motivation to quit? and (5) are there subsets of smokers who worry more or less than others?

Method

Community-based and college-student smokers attended three laboratory assessments and recorded their thoughts and worries about smoking using a personal digital assistant (PDA) for two 1-week intervals. The recording weeks were separated by one week of no recording. During the two assessment weeks, participants completed PDA questionnaires five times daily. Three of these five questionnaires were completed in conjunction with specific smoking instances.

Participants

We recruited approximately equal numbers of college student smokers and adult smokers from the surrounding metropolitan community. The resulting sample is one of convenience but the inclusion of community smokers improves generalizability as opposed to an exclusively college-student sample. Ninety-four smokers began the study, but six were excluded before data analysis because their data were either lost due to technological malfunction (n = 3) or they dropped from the study before completion (n = 3). Thus, participants included 88 (52 male and 36 female) daily smokers. Student smokers (n = 46) were recruited from a screening questionnaire distributed to undergraduates in psychology classes at the beginning of each semester. Community smokers (n = 42) were obtained in five ways: from posters placed throughout the metropolitan community (n = 14), through a 14-word personal advertisement placed in local newspapers (n = 14), through a 14-word advertisement placed in the alumni newsletter (n = 5), from the faculty/staff email listserv (n = 5), and by word of mouth (n = 4). None of the smokers in the community sample were full-time students at the same university from which our student sample was obtained.

Procedure

Participants came into the lab individually or in groups up to eight. They were told that the purpose of the study was to learn about smokers’ daily thoughts. After providing informed consent, smokers completed a pretest questionnaire and were then asked to carry a PDA to record their thoughts about smoking five times a day for two 1-week intervals. After the first week of PDA recording, all smokers turned in their PDA and had a 1-week break. The participants returned at the end of the off-week for a second week of PDA questionnaires identical to the first. Following the last week of recording, participants returned the PDA and completed a posttest questionnaire.

Measures

Pretest

The pretest included demographic items, an addiction scale, the Penn State Worry Questionnaire (PSWQ), 10 worry and risk items, an intrusive thoughts scale, motivation to quit, quitting self-efficacy, and open-ended assessments about quit attempts.

Demographics.

Background data collected at pretest included age, gender, education, marital status, and ethnicity. We assessed smoking behavior by asking about the number of cigarettes smoked each day, the age of smoking initiation, the smoking status of spouse/partner, and the number of smokers in the participant’s household.

Addiction.

We administered the Fagerstrom Test of Nicotine Dependence-Revised (FTND-R; Heatherton, Kozlowski, Frecker, & Fagerstrom, 1991). The FTND-R has six items yielding a summary score ranging from 1 to 10 (higher scores indicate higher dependence; α = .64).

PSWQ.

The PSWQ is a 16-item questionnaire assessing trait worry (Meyer, Miller, Metzger, & Borkovec, 1990). Examples of items are “My worries overwhelm me” and “Once I start to worry, I cannot stop.” Responses ranged from 1 (not at all typical of me) to 5 (very typical of me). Items were summed, with higher numbers indicating higher levels of trait worry (α = .94).

Worry.

Four items assessed worry about developing a medical problem and about harming others. For instance, smokers were asked “How worried are you about developing a smoking-related medical condition?” The two items assessing worry about developing a medical problem were averaged (α = .85) as were the two items assessing worry about harming others (α = .83).

Risk.

Six items assessed absolute and comparative risk from smoking. For instance, smokers were asked “What are the chances that you might develop some type of cancer within your lifetime if you continue to smoke?” These items were combined to create three risk measures: absolute risk of illness (α = .87), comparative risk of illness (α = .58), and risk to others (α = .59).

Intrusive thoughts.

The intrusive thoughts subscale of the impact of event scale (Horowitz, Wilner, & Alverez, 1979) includes seven items. These items were changed to pertain directly to smoking. For instance, one item originally stated, “I thought about it when I didn’t mean to” (with “it” referring to a serious life event) and was changed to, “I thought about smoking when I didn’t mean to.” Smokers rated how often each item had been experienced in the last seven days (0 = not at all, 1 = rarely, 3 = sometimes, and 5 = often). The summed items produce a total score ranging from 0 to 35 (α = .84).

Contemplation ladder.

Readiness to quit smoking was assessed with the contemplation ladder (Biener & Abrams, 1991), an 11-point scale, with 0 = No thought of quitting and 10 = I’m taking action to quit (e.g., cutting down, enrolling in a program). This scale has good discriminant validity (Abrams & Biener, 1992) and has been used extensively in the smoking literature (e.g., Amodei, & Lamb, 2004; Martin, Rohsenow, MacKinnon, Abrams, & Monti, 2006). The contemplation ladder predicts smoking cessation at a similar level as intentions to quit smoking (Abrams, Herzog, Emmons, & Linnan, 2000).

Quit attempts.

Participants indicated the number of times they had tried to quit smoking for at least 24 hours in their lifetime, including the date and length of their most recent quit attempt.

Posttest

The posttest reassessed intrusive thoughts, worry and risk, motivation to quit, plans to quit, and self-efficacy. This questionnaire also assessed smoking behavior changes.

Thoughts.

Participants read 15 smoking-related thoughts on the PDA questionnaires. Fourteen thoughts were drawn from the 20-item Reasons for Quitting scale (RFQ; Curry et al., 1997) with the fifteenth thought coming from pilot testing (”How I appear to others when I’m smoking”). The thoughts were reworded from reasons to quit smoking to negative thoughts about smoking (e.g., “people will stop nagging me” became “people nag me about my smoking”).

EMA Procedure

Smokers completed five questionnaires a day to gather a cross-section of smokers’ daily thoughts and worries while also minimizing respondent burden. To capture the cross-section of smokers’ daily thoughts and worries, we used event-, signal-, and interval-contingent recordings (Wheeler & Reis, 1991). Event-contingent recordings were completed after smoking specific cigarettes, while signal-contingent recordings were completed at a random time everyday and interval-contingent recordings were completed before going to sleep each night.

PDAs.

Smokers carried one of two models of PDA manufactured by Palm (Sunnydale, CA): The M105 and the Zire 31. The software program Satellite Forms 5.2 was used to create the PDA application used for data collection.

PDA questionnaire structure.

Participants were asked “Where did you smoke?” (home, work, other’s home, bar/restaurant, vehicle, or other), “Did you have to move to where smoking was permitted?” (yes, no), and “Were you with others?” (yes, no). If participants indicated they were with others, they were automatically directed to a screen asking them to check all the people who they were with at the time of smoking (family, friends, coworkers, strangers, other smokers, and/or nonsmokers).

Next, participants were asked “Did you have any negative thoughts before, during, or after smoking this cigarette?” If participants answered “No” to having negative thoughts, they were finished with the questionnaire. If participants answered “Yes”, they were directed to a negative thought checklist consisting of the 15 RFQ thoughts. In addition, they were given the option to check “None of these thoughts” to indicate they had experienced negative thoughts about smoking but none of those listed.

Upon checking a RFQ thought, smokers were automatically directed to a new screen where they were asked “How intense was the worry about this thought?” Responses were made on a 3-point scale labeled “no worry”, “some worry”, or “a great deal of worry.” Then participants were taken back to the thought checklist and given the chance to check other negative thoughts. Once each day, in the afternoon, smokers were required to read through each negative thought and answer “yes” or “no” to having experienced them. The remaining four questionnaires allowed smokers to bypass the thoughts if they indicated they did not have any negative thoughts.

Event-contingent recording.

Event-contingent recording, the practice of linking questionnaires with specific events, was used three times a day. Participants completed a questionnaire after the first cigarette smoked in the morning (Morning questionnaire), the first cigarette smoked after 12:00 p.m. (Afternoon questionnaire), and the first cigarette smoked after 5:00 p.m. (Evening questionnaire). Smokers were asked to complete these recordings immediately after smoking. Reminder messages, accompanied by alarms, were issued at 10:30 a.m., 1:00 p.m., and 7:00 p.m. on the PDA to ensure participants did not forget the event-contingent questionnaires. Morning questionnaires had to be completed anytime between when the smoker woke up and 11:30am, afternoon questionnaires had to be completed between 12 p.m. and 4 p.m., and evening questionnaires had to be completed between 5 p.m. and 10 p.m. Questionnaires completed outside these time limits were considered noncompliant. In addition, if a smoker did not smoke during one of these time periods, they were instructed to skip the recording and inform the experimenter to avoid being penalized for noncompliance. We did not wish to change their smoking behavior.

Signal-contingent recording.

Signal-contingent recording, the practice of signaling a questionnaire with an alarm, was used once daily (Random Beep questionnaire). Smokers were randomly signaled between 2:00 p.m. and 5:00 p.m. to complete a questionnaire. This questionnaire was shorter than the others, consisting of two questions: “Have you smoked a cigarette in the last hour?” (yes, no) and “Have you had any negative thoughts about smoking in the last hour?” (yes, no). If they indicated they had experienced negative thoughts about smoking in the last hour, they were directed to the thought checklist. Smokers had to complete the questionnaire within one hour of the random beep to be considered compliant.

Interval-contingent recording.

The last questionnaire of the day (end-of-day questionnaire) was an interval-contingent recording in which the questionnaire was tied to a predetermined period: immediately before going to bed. This questionnaire had to be completed between 8 p.m. and 4am for the smoker to be compliant. First, smokers indicated whether they had smoked in the last hour and whether they had experienced any negative thoughts about smoking in the last hour. If they had experienced negative thoughts in the last hour, they completed the thought checklist and intensity questions about that time period.

Compliance

An overall study compliance percentage was computed by dividing the total number of questionnaires smokers completed over the course of the two weeks of EMA recording by the number of questionnaires that could have been completed. This compliance percentage was then categorized by week, day, and individual questionnaire percentages. Overall, smokers completed 82.4% of the EMA questionnaires within time guidelines. They were slightly more compliant in week 1 (83.6%) versus week 2 (81.3%) and slightly less likely to complete the end of day questionnaire (76.8%) versus all others (all percentages >79.8%).

Results

Questionnaire Data

We used 2 (student or community) × 2 (male or female) analyses of variance to assess volunteer group and pretest gender differences. We obtained only one pretest gender difference, and no reliable volunteer group × gender interactions. Compared to men, women believed that the risk to others from their own smoking was greater, F(1, 84) = 7.06, p = .009. Because of identifying only a single gender difference, we collapsed across gender for all subsequent analyses.

We also compared students and community volunteers. Most students were in their first or second year of college, whereas 13 of the community volunteers (31.0%) reported that they were taking college courses. Students were more likely to be single (90.0% vs. 54.8%), and all of the students were white, whereas five of the community volunteers (10.9%) reported minority status. The students (M = 22.0) were also much younger than the community volunteers (M = 36.8), F(1, 84) = 43.33, p < .001. Aside from these demographic differences, the volunteer groups differed significantly on only three out of 19 pretest variables (see Table 1). Specifically, the students a) were less addicted, F(1, 84) = 6.65, p = .01, b) reported greater self-efficacy about quitting, F(1, 84) = 6.32, p = .01, and c) reported fewer intrusive thoughts about smoking, F(1, 83) = 3.90, p = .05. Because our volunteer groups strongly differed in average age, we conducted an analysis of covariance controlling for age, and only one pretest difference remained. Specifically, the college students (M = 8.29, SD = 4.94) reported fewer intrusive thoughts than the community volunteers (M = 10.81, SD = 8.69), F(1, 84) = 5.89, p = .02. Overall, the sampling procedure achieved its intended aim: producing community smokers who differed in some ways from the student smokers. On the other hand, one could account for almost all the differences by referring to the smokers’ ages. Thus, rather than separating the two groups in the subsequent analyses, we combined them into a single sample of smokers who differed on several dimensions.

Table 1.

Means and SDs for Smokers at Pretest and Posttest

	Students (n = 46)	Community volunteers (n = 42)	Total (n = 88)
Pretest smoking behavior
Cigarettes smoked/day	15.18 (7.54)	17.83 (8.99)	16.45 (8.32)
Addiction (FTND)	3.22 (1.82)	4.45 (2.61)	3.81 (2.30)
Age smoking began	15.17 (2.20)	15.29 (4.40)	15.23 (3.41)
Number of times quit in past	5.62 (7.88)	5.17 (6.93)	5.40 (7.39)
Time since last quit attempt (months)	18.27 (29.97)	40.54 (65.22)	29.23 (51.33)
Length of last quit attempt (months)	1.20 (2.18)	4.90 (14.53)	2.99 (10.32)
No. of people in household who smoke	.59 (.93)	.58 (1.39)	.58 (1.16)
Pretest attitudes
Self-efficacy/quitting (1–10)	7.04 (2.40)	5.64 (2.86)	6.38 (2.71)
Worry about medical condition (1–7)a	3.85 (1.34)	4.51 (1.72)	4.16 (1.56)
Worry about harming others (1–7)a	3.64 (1.41)	4.43 (1.60)	4.02 (1.54)
Percent risk of medical conditionb	71.08 (20.81)	71.36 (22.79)	71.21 (21.65)
Comparative risk of medical condition (1–7)b	3.15 (.65)	3.33 (.63)	3.24 (.64)
Risk of harming others (1–7)a	4.63 (1.23)	4.96 (1.39)	4.79 (1.31)
Contemplation of quitting (1–10)	5.91 (2.18)	5.48 (3.13)	5.70 (2.67)
Penn State Worry Questionnaire (PSW)	43.83 (7.51)	43.52 (11.35)	43.68 (9.48)
Intrusive thoughts	8.29 (4.94)	10.81 (8.69)	9.51 (7.08)
Posttest attitudes
Self-efficacy/quitting	7.33 (2.08)	6.17 (3.12)	6.77 (2.67)
Worry about medical conditiona	3.73 (1.49)	4.18 (1.81)	3.94 (1.65)
Worry about harming othersa	3.59 (1.54)	3.75 (1.79)	3.66 (1.66)
Percent risk of medical conditionb	69.20 (22.15)	67.88 (25.74)	68.57 (23.80)
Comparative risk of medical conditionb	3.09 (.66)	3.20 (.89)	3.14 (.78)
Risk of harming othersb	4.15 (1.44)	4.27 (1.69)	4.21 (1.55)
Contemplation of quitting	6.48 (2.44)	5.90 (3.46)	6.20 (2.97)
Intrusive thoughts	6.07 (5.18)	6.29 (6.31)	6.17 (5.71)

Note. FTND = Fagerstrom Test of Nicotine Dependence.

^aFor descriptive purposes, worry means are given for each type of worry assessed. An averaged score for these two forms of worry are used for all statistical analyses.

^bRisk means are given for each type of risk assessed. z scores are used for all statistical analyses.

How Often Do Cigarette Smokers Think About Negative Aspects of Their Smoking?

Across two weeks of self-monitoring, each participant had approximately 60 total opportunities to report a negative thought about smoking, resulting in a total of 5280 possible recording instances across all 88 participants. As a whole, participants responded to the queries to complete the PDA questionnaires 82.4% of the time, resulting in a total of 4351 completed questionnaires. Of the times smokers responded to their PDAs, they reported having had at least one negative thought 26.8% of the time. Over the two weeks of recording, smokers reported a total of 4297 negative thoughts. Individual smokers reported anywhere from 2 to 350 thoughts, and five of the 88 smokers failed to report a single negative thought over the 2-week period. The distribution of thoughts reported was positively skewed, with 80% (n = 70) of participants reporting a thought 0% to 10% of the time they completed a questionnaire.

Table 2 shows the total and individual proportion of thoughts reported for the two weeks. Smokers reported thinking most often about how much cigarettes cost and least about whether they were addicted. The proportion of total thoughts that a smoker reported was positively correlated with the intensity of total worry they reported across the two weeks (r = .40, p < .01). Thus, thoughts and worry are related but not identical constructs.

Table 2.

Total Number of Thoughts Reported and Worry of Online Individual Thoughts

Thought	Total of thoughts	Individual proportion of thoughts (%)	Worry intensity
How much money cigarettes cost.	471	11.2	1.07 (.44)
Symptoms I’m having because of smoking.	434	10.1	1.24 (.47)
People I know want me to quit.	408	9.6	1.03 (.52)
How I smell like cigarettes.	405	9.3	1.08 (.51)
How my clothes smell bad from smoking.	375	8.7	1.10 (.45)
Bothering other people with my smoke.	315	7.3	1.14 (.41)
Not sure that I can quit smoking.	302	6.9	1.12 (.52)
Getting sick because of my smoking.	282	6.4	1.29 (.50)
I don’t like myself as a smoker.	271	6.3	1.25 (.49)
Not sure I can control my smoking life.	244	5.5	1.18 (.51)
How I appear to others when I’m smoking.	219	5.1	1.19 (.52)
Quitting would stop people from nagging me.	202	4.7	1.11 (.54)
Dying because of my smoking.	187	4.3	1.12 (.61)
People I know who are sick because of smoking.	94	2.2	1.18 (.53)
Wonder if I’m addicted.	88	2.1	1.31 (.61)

Note. Worry Intensity: 0 = never, 1 = sometimes, 2 = often. Frequency of negative thoughts was computed by summing the number of times each smoker reported a particular thought and then summing across all participants. Proportions of individual thoughts reported were computed by dividing the number of times an individual reported having a specific thought by the number of times he or she completed the online questionnaire. These were then averaged across all participants.

Are There Subsets of Smokers Who Have More or Fewer Negative Thoughts?

Table 3 presents correlations between pretest smoker characteristics and the proportion of negative thoughts reported. None of the background variables was a significant predictor of the proportion of negative thoughts reported by smokers, although a weak relationship (p = .09) was obtained between smoking more each day and fewer thoughts. Smokers who worried more about smoking (r = .26, p = .015) and who had higher risk perceptions (r = .26, p = .017) reported more negative thoughts. Also, smokers who were thinking more about quitting (r = .22, p = .04) reported more negative thoughts about quitting during the 2-week period.

Table 3.

Correlations Between Pretest Items With Proportion of Thoughts and Worry Intensity

	Proportion of thoughts reported	Average worry intensity
Age	.14	−.01
Cigarettes/day	−.18	−.15
Addiction (FTND)	.04	.04
Age began smoking	−.15	−.09
Worry	.26*	.54**
Risk (z score)	.26*	.34**
Contemplation	.22*	.46**
PSW	.19a	.22*
Intrusive thoughts	.20b	.44**

Note. FTND = Fagerstrom Test of Nicotine Dependence.

^ap = .07.

^bp = .06.

^*p < .05.

^**p < .01.

When and in What Circumstances Are Negative Thoughts Most Likely?

A mixed-effects random regression model, or random effects modeling, (REM; Crosby, Pearson, Eller, Winegarden, & Graves, 1996) was used to analyze differences between community and student participants. This model takes into account all missing data and creates an estimate of the predicted mean for the dependent variable controlling for covariates in the model. Results of this initial REM analysis established that student and community participants did not differ in either the proportion of reported thoughts (p = .30) or the intensity of worry (p = .59).

REM was then used to examine the frequency of negative thoughts reported across week and day of the week (i.e., Monday, Tuesday, etc.). Because only two questionnaires were completed on Day 1, Day 1 was removed from the analysis for both weeks (Shiffman et al., 2002). Thus, the analysis only included Days 2 to 7. In addition, to control for Type II error, Tukey’s HSD tests and all main effects for REM were set at p = .01. The following analyses used the Proc Mixed procedure (SAS 9.1). The analyses of total reported thoughts revealed no significant effects for week (p = .25), day of the week (p = .49), or the week X day interaction, (p = .98). We also tested whether the thoughts reported during the week varied by whether it was the weekday or weekend. Thoughts reported Monday through Friday were averaged to create a “weekday” score, and thoughts reported Saturday through Sunday were averaged to create a “weekend” score. Smokers reported a comparable number of thoughts regardless of weekend or weekday, p = .12.

Situational Circumstances

REM was used to assess the impact of situational variables on the number of negative thoughts reported. Four different models were tested: whether smokers reported different numbers of negative thoughts (a) depending on the assessment situation (i.e., home, work, someone else’s home, bar/restaurant, vehicle, or another location), (b) depending on who they were with (i.e., family, friends, coworkers, strangers, another smoker, a nonsmoker, or a combination of these), (c) when they had to move to another location to smoke, and (d) whether they reported smoking at the time of questionnaire completion.

Smokers’ situational contexts played a role in how many overall thoughts they reported, F(5, 2595) = 9.12, p < .001. Half (49.9%) of the PDA assessments were completed at home, followed by 20.5% while in a vehicle, 13.0% at an unidentified location, 9.9% at work, 5.3% at someone else’s home, and 1.4% in a bar or restaurant. Post hoc tests (Tukey’s HSD) revealed that when smokers were at work they reported more negative thoughts about smoking (est = .126, SE = .012) than if they were at home (Mean est = .067, SE = .010), at someone else’s home (est = .063, SE = .015), or in a vehicle (est = .077, SE = .010; all ps < .001). Additionally, more thoughts were reported in an “other” (unidentified) location (est = .100, SE = .012) than at home or at another’s home (all ps < .01); unfortunately, we cannot identify these “other” situations.

Smokers reported more negative thoughts about smoking when they were with others than when alone (est = .093, SE = .010 vs. est = .073, SE = .009), F(1, 2596) = 10.79, p = .001. The number of thoughts also varied by whom the smokers were with, F(18, 731) = 2.05, p = .006. Post hoc tests (Tukey’s HSD) revealed that when smokers were with “friends and family” they reported more negative thoughts than when they were with any other combination of individuals.

Smokers also reported more thoughts when they had to move to smoke (est = .097, SE = .010) than with they did not have to move (est = .068, SE = .009), F(1, 2597) = 18.81, p < .001.

In general, when smokers reported moving, they more often reported thinking about how “people I know want me to quit.” In contrast, smokers reported thinking more about the cost of cigarettes when they did not move to smoke.

Finally, recall that two of the five questionnaires (random beep and end of day) were not event-contingent and therefore could be completed even when participants had not smoked recently. Comparisons of the number of thoughts revealed that smokers had more thoughts after smoking (est = .054, SE = .006) than when they did not smoke (est = .081, SE = .008), F(1, 1549) = 35.50, p < .001.

How Often Do Smokers Worry About Their Smoking?

When smokers reported having one of the 15 thoughts, they were prompted to report the intensity of their worry about that thought. Overall, smokers reported having had at least some worry 94% of the time they reported a negative thought. Smokers reported more intense worry for “wonder if I’m addicted” (the least frequent thought) and the least intense worry for “people I know want me to quit”. As the data in Table 2 suggest, the thoughts that smokers reported most frequently were not those that they worried most about. In fact, correlating the last two columns in Table 2 produced the opposite relationship: thoughts reported most frequently were less worrisome, r = −.49, p = .06.

When and in What Circumstances Do Smokers Worry?

REM was used to assess differences in worry across week and day of the week. When worry about all thoughts was combined, a main effect for week occurred, F(1, 1076) = 7.55, p = .006. Smokers reported more worry about their thoughts during week 1 (est = 1.03, SE = .037) than during week 2 (est = .98, SE = .038). Smokers’ worry about smoking did not differ across the days of the week (p = .79) nor was there a significant week × day interaction (p = .23). In addition, smokers’ worry about their thoughts did not differ by weekend or weekday, p = .44.

Situational Circumstances

Similar to analyses on the number of negative thoughts, the amount of worry was evaluated in terms of (1) where smokers were, (2) whom smokers were with, (3) if smokers had to move to a new location to smoke, and (4) whether smokers had smoked recently. No significant effects were obtained for these variables.

Are There Subsets of Smokers Who Worry More or Less Than Others?

Table 3 presents correlations between mean worry intensity ratings and pretest smoker characteristics. Smokers were more likely to report greater worry intensity about smoking thoughts the more they worried about smoking in general (r = .54, p < .001) and the higher their risk perceptions associated with smoking (r = .34, p = .001). Also, smokers who were thinking more about quitting (r = .46, p < .001), and who had higher scores on the PSW were more likely to report greater worry intensity about smoking thoughts (r = .22, p = .048). Finally, smokers who experienced more intrusive thoughts reported greater worry about their smoking thoughts (r = .44, p < .001).

How Do Negative Thoughts and Worry Relate to Contemplation to Quit?

An important question is whether the proportion of negative thoughts smokers reported and the worry associated with these thoughts during the EMA recording predicted posttest contemplation about quitting smoking. The proportion of all thoughts was correlated with contemplation to quit at posttest, r = .28, p = .009. Worry intensity, averaged across the 2-week assessment, was also correlated with contemplation about quitting, r = .46, p < .001.

Cognitive Versus Affective Predictors of Contemplation to Quit

We asked whether contemplation to quit smoking was best predicted by worry intensity and the proportion of thoughts reported. A simultaneous regression revealed that worry intensity, t(82) = 4.05, p < .001, β = .43, but not the proportion of thoughts, t(82) = .76, p = .45, β = .08, predicted posttest contemplation to quit. The greater a smoker’s worry about negative smoking thoughts, the more he or she was contemplating quitting. This effect, which was based on EMA worry reports, held even when controlling for trait worry (PSWQ scores worry intensity β = .46). A second regression in which pretest contemplation was controlled for in step 1 before entering worry intensity and proportion of thoughts at step 2 negated the previous thought/worry/contemplation relationships. Neither proportion of thoughts, t(81) = .87, p = .39, β = .07, nor worry intensity, t(81) = 1.65, p = .10, β = .16, predicted posttest contemplation after controlling for pretest contemplation levels, t(81) = 6.72, p < .001, β = .60.

Discussion

The purpose of this study was to add to our understanding of the thoughts and worries that smokers have about their smoking behavior. The study is especially novel because of its reliance on “online” measurement as opposed to the more typical methodology of asking smokers to recall their thoughts and feelings. Other positive features of the study include a varied sample of both lighter and heavier smokers and the inclusion of multiple recording weeks.

Smokers reported having an average of just over four negative thoughts each day, although this mean value could be high given that smokers typically reported their thoughts after smoking a cigarette, and other data showed that thoughts were more frequent following smoking episodes. Given that possible constraint, the most frequent negative thoughts were those related to “immediate reinforcement,” such as how one appears to others and how much the cigarette costs. This pattern of thoughts in which immediate consequences were considered more frequently than distant consequences was also evident for health concerns. Smokers rarely reported thinking about dying from smoking (4.3% of their thoughts) compared to thinking about symptoms they were presently experiencing (10.1%).

We also learned something about when negative thoughts are more likely for smokers. In particular, they were more likely to report a negative thought when they (a) were smoking at work as opposed to at home, someone else’s home or in a vehicle, (b) were with other people, especially friends and family, and (c) had to move somewhere to find an appropriate place to smoke. These data suggest that “public smoking” (at work; with others) is more likely to produce negative thoughts, in line with the notion that smoking behavior is becoming increasingly stigmatized (e.g., Farrimond & Joffe, 2006). The data also suggest that laws mandating no smoking in the workplace could cause smokers to have more negative thoughts about their own behavior (although not necessarily to worry more).

Other implications of the “thoughts” data for motivating smoking cessation are not so obvious. One might suggest, for example, an advertising strategy emphasizing the immediate consequences of smoking because that is what smokers think about most often. But, because our sample consists of continuing smokers, the thoughts they reported experiencing often are not necessarily the thoughts that will lead to successful cessation; otherwise, they might have already quit. Instead, one might propose increasing efforts to emphasize consequences that smokers do not report often. In the latter case, for example, advertisements might target reminders of the addictive nature of smoking or the fact that smokers do die earlier (and often with a poorer quality of life). Such an approach might be more attention-grabbing and motivation-inducing than one based on reminding smokers of the more prevalent negative thoughts they are already having.

Yet another way of looking at the thoughts data is to consider the connection of the different thoughts to worry. Of note, frequent thoughts were not the ones that produced the most intense worrying. Smokers reported the most worry about self-control (wonder if I’m addicted; don’t like myself as a smoker) and health (symptoms I’m having; getting sick because of my smoking). A focus on worry would suggest that to motivate smoking cessation, we should emphasize the lack of self-control that smokers have and the many symptoms and illnesses attributable to smoking. This suggestion is bolstered by the clear finding that average worry about one’s thoughts is highly related to contemplation of quitting; indeed, once we accounted for the worry-contemplation connection, the number of thoughts smokers reported no longer accounted for meaningful variation in contemplation. It is also worth noting that, intuition to the contrary, smokers are somewhat naïve about both the addictive nature of their behavior and the illness consequences accompanying smoking. Although smokers recognize that quitting is a difficult task (Molyneux et al., 2006), many persist in thinking that they can quit relatively easily (Romer, Jamieson, & Ahern, 2001) and they are unaware of many smoking-disease associations (Weinstein, Slovic, Waters, & Gibson, 2004).

One limitation to our data is that it is impossible to disentangle the causal relationships between smokers’ thoughts, worries, and levels of contemplation to quit. That is, we cannot determine whether the thoughts and worries reported during the two weeks of EMA are a product of, or are the precursor to, contemplation to quit smoking. An explication of these relationships would be an important direction for future research. Do changes in the frequency of thinking or worrying about smoking move people further along the contemplation ladder or rather does more contemplation of quitting lead to thinking and worrying about smoking more often? Also, when queried about their thoughts, the smokers in our sample most frequently reported not having negative thoughts about smoking—this response occurred 76% of the time. Moreover, even this relatively low level of thinking about smoking could be artificially high, given that smokers were typically questioned shortly after a smoking episode. On the other hand, with only five measurements daily, when smokers were consuming three times that many cigarettes on average, we may have missed important instances when negative thoughts were more likely. Moreover, when smokers reported having had a thought, they also knew that they would have to complete the subsequent questionnaire; thus, underreporting might have been reinforcing. Finally, we need to emphasize that even EMA methodology cannot provide us with the “true” content of an individual’s inner state: The methodology is better than others but still imperfect.

The EMA methodology relies on frequent assessments and thus raises the possibility of reactivity. Frequent exposure to the various thoughts one could have about smoking might cause the smokers to have those thoughts more frequently or to increase their worry about those thoughts. Using a strategy similar to that used by Stone et al. (2003), we implemented the EMA protocol for two weeks, reasoning that any reactivity that is present should be more apparent earlier in the recording process when the procedures are novel. No differences emerged for number of thoughts across week or day of the week, and day of week did not affect worry. However, our smokers did report greater worry during week 1 versus week 2, raising the possibility that the reported worry levels early in the study were artificially high.

One final limitation of the study is that the implications of this work for smoking cessation depend on the contemplation of quitting self-report outcome measure. The reliance on self-reports is a weakness of the present study, but it is important to point out that the contemplation measure is strongly related to intentions to quit smoking (Biener & Abrams, 1991). In turn, intentions to quit smoking are strongly related to quit attempts (Norman, Conner, & Bell, 1999).

In summary, this study presents the first data regarding smokers’ day-to-day negative thoughts and worries about their smoking behavior. Our results suggest that a minority of smokers can go without thinking at all about the negative consequences of their smoking and that most smokers ignore thinking about the negative features of the behavior most of the time they smoke. Our results also show that thoughts about smoking (i.e., cognitions) and feelings about smoking (i.e., worry) are loosely connected and it is feelings rather than cognitions that are most related to contemplation to quit. For investigators who are interested in increasing motivation to quit and the quit attempts that follow, we suggest increased attention to methods of increasing worry among smokers about the consequences of their actions.