OUP user menu

Time preferences, socioeconomic status and smokers’ behaviour, attitudes and risk awareness

Patrick Peretti-Watel, Olivier L’Haridon, Valerie Seror
DOI: http://dx.doi.org/10.1093/eurpub/cks189 783-788 First published online: 23 January 2013

Abstract

Background: Tobacco control policies have succeeded in reducing tobacco use, but the negative correlation between smoking prevalence and socioeconomic status (SES) has increased. This study focused on the relationships between time preferences, SES, and smoking behaviour, attitudes and risk awareness. Methods: A cross-sectional telephone survey was conducted in France in 2008 on a representative national sample of people aged 18–75 (N = 2000, including 621 smokers) years. Two scales measuring planning horizon and impulsivity and various indicators of SES were introduced into the logistic regressions performed on smoking status and smokers’ attitudes to anti-smoking campaigns, quitting attempts, attempts to quit or smoke less, fear of smoking-related cancer and risk perception. Results: Indicators of lower SES and smoking status were correlated with present time orientation and impulsivity. On modelling smoking status, time orientation and lower SES were found to be significant predictors. Among smokers, lower SES and present time orientation were predictive of smoking-related outcomes: little personal concern with anti-tobacco campaigns, not reporting recent behavioural changes, not expressing personal fear of smoking-related cancer and low risk awareness. When time-related preferences were introduced into the analysis, the effects of several lower SES indicators (especially a low educational level) became non-significant. Conclusion: The relationship between SES and smoking behaviour, attitudes and beliefs may be partly mediated by time preferences. Time preference is strongly correlated with smoking status, risk perceptions and attitudes towards anti-smoking campaigns. Tobacco control policies should include messages targeting present time-oriented smokers and/or interventions designed to enhance more future-oriented attitudes among smokers.

Introduction

Tobacco control policies have successfully decreased the prevalence of smoking in many industrialized countries during the past decade, but this trend has been accompanied by increasing inequalities as far as smoking is concerned, to the detriment of people with a lower socioeconomic status (SES). This tendency already occurred in the UK between 1970 and 1990,1 and the negative correlation between smoking prevalence and educational level increased in several European countries between 1985 and 2000.2,3 Between 2000 and 2007, the smoking rates decreased in France by 22% among executive managers and professionals and by 11% among manual workers, but remained stable among the unemployed.4

Why are people with a lower SES more prone to smoking? Firstly, less educated people may experience greater difficulty in understanding the information conveyed by preventive campaigns. This is consistent with the positive correlation found to exist between a low educational level and the tendency to underestimate smoking-related health hazards.5,6 Secondly, underprivileged smokers may be less likely to quit because they regard tobacco as a necessity good that helps them cope with the hardships in their lives.7

In addition, the most serious effects of smoking on health occur many years ahead, whereas the perceived benefits are immediate. People with a lower SES may therefore be less inclined to worry about the adverse future effects: classical sociological studies have shown that poverty shrinks people’s temporal outlook.8–10 People with a low SES may therefore be more prone to indulge in persistent smoking because they are more present-oriented.

In economic studies, time discount rates are often used to reflect willingness to exchange present for future utility. Time preferences are a key component of the theory of rational addiction. According to this theory, time preferences are stable and specific to each individual, and present-oriented people are more prone to addiction.11 Time preferences are shaped by learning and culture,12,13 and schooling also increases people’s patience and their ability to look ahead.14,15

The importance of time preferences as a factor contributing to people’s decisions about smoking is self-evident: the decision to smoke involves an intertemporal trade-off between the immediate benefits and the delayed costs, and conversely, the decision to quit involves an intertemporal trade-off between the immediate costs and the subsequent rewards. However, it is not easy to assess the impact of time preferences on smoking empirically because this factor is difficult to measure. The most widely used method consists in presenting subjects with hypothetical scenarios about willingness to exchange money today for (more) money in the future. However, the results obtained on these lines differ considerably from one study to another,16 and the associated estimates are rather noisy. In a recent study based on this approach, no significant relationship was found to exist between time discounting and smoking status, whereas a strong correlation was observed between smoking and the respondents’ impulsivity and planning horizon scores, based on their level of agreement with various statements.17 Impulsivity can be said to reflect a short-term outlook, whereas planning horizon reflects the ability to envisage medium-/long-term consquences.18,19

In this study, we firstly investigated the relationships between SES and time preferences, and then analysed simultaneously the impact of a low SES and time preferences on smoking status, as well as various smokers’ behaviour, attitudes and risk awareness. Rather than using a single time discounting score, we used two separate scores to measure planning horizon and impulsivity, assuming that these scores would capture at least some of the explanatory power of indicators of lower SES.

Methods

Sample

The study was based on data collected in a national telephone survey conducted by the French Institute for Health Promotion and Health Education (INPES) in June to July 2008. A random digit dialling process was used to obtain listed and unlisted telephone numbers. The households contacted were forewarned by a letter describing the survey. In each household, one person aged 18–75 years was randomly selected to undergo the interview. Overall, 2000 people were interviewed (response rate: 71%).

The resulting sample was weighted by the inverse of the household size. We also used data from the National Institute of Statistics and Economic Studies to calculate weights so that our sample would be representative of the general French population aged 18–75 (in terms of age, gender, education, size of town and geographical area) years. Among the participants, 621 reported that they smoked cigarettes at least occasionally at the time of the survey. These smokers were asked specific questions, and the non-smokers were questioned retrospectively about their past smoking habits.

Questionnaire

The questionnaire included the following indicators to the respondents’ socio-demographic characteristics and SES: gender, age, educational level, job status, occupation (manual worker or other occupation), household income and how these resources had changed during the past 5 years (using a dummy to indicate those who reported that their income levels had greatly decreased). These indicators were used to identify several aspects of SES, which are known to be predictive of smoking status.20 Concerning the income level, a dummy was used instead of a continuous variable because the aim was to identify low-SES people (<€1500 vs. ≥€1500 per month, a threshold commonly used in French surveys).21–23 Concerning educational level, after completing high school, most young people in France go to university, as there is no real selection process at French Universities and the tuition fees are usually low. The selection process occurs later on, and many students therefore leave the University once they have obtained their first degree, or with no degree at all. High school graduates and University undergraduates therefore have fairly similar profiles. We therefore defined the following three levels: <high-school, high-school or first university degree, >first university degree.

Respondents’ time preferences were determined by asking whether they agreed with eight statements borrowed from previous studies17–19: ‘I often think about what my life will be like ten years from now’ (Plan 1), ‘I prefer to enjoy the present instead of worrying about the future’ (Plan 2), ‘I am ready to go without some pleasures in order to live a few years longer’ (Plan 3), ‘I prefer to enjoy spending the money I earn instead of putting it aside for the future’ (Plan 4), ‘sometimes I behave impulsively and I regret it later’ (Imp 1), ‘I know how to keep disruptive emotions and impulses under control’ (Imp 2), ‘I base my decisions on my instinct instead of balancing the pros and cons’ (Imp 3) and ‘Once I have made a decision, I stick to it’ (Imp 4). The response items were encoded from 0 to 3 (‘strongly disagree’ = 0, ‘disagree’ = 1, ‘agree’ = 2 and ‘strongly agree’ = 3) for statements expressing impulsivity (Imp 1, Imp 3) and future orientation (Plan 1, Plan 3), and conversely, from 3 to 0 for statements expressing self-control (Imp 2, Imp 4) and present time orientation (Plan 2, Plan 4). The resulting outcomes were summed together to obtain two scales ranging from 0 to 12 (planning horizon: Plan 1 + Plan 2 + Plan 3 + Plan 4, impulsivity: Imp 1 + Imp 2 + Imp 3 + Imp 4). A high (low) planning horizon score was taken to reflect future (present) time orientation. Likewise, people with a high impulsivity score were assumed to be more impulsive.

Regarding smoking status, former smokers were defined as those who reported that they were not currently smoking, but acknowledged that they had smoked every day for at least 6 months in the past. Among smokers, the following variables were used:

  • cigarette consumption (>10 cigarettes per day, a threshold commonly used to define heavy smokers24–26);

  • time to first cigarette after waking up (an indicator of nicotine dependence27);

  • changes in smoking behaviour during the past 12 months (smoking less, attempting to quit for at least 1 week);

  • attitudes towards anti-tobacco campaigns (feeling personally very concerned, a little, versus not really, not at all);

  • personal fear of smoking-related cancer (yes a lot, yes a little versus not really, not at all);

  • two questions dealing with risk perception: respondents were asked how many out of 100 smokers would die of a tobacco-related disease [responses were collapsed into a binary outcome: (0–49) vs. (50–100)] and how many years of life a smoker who never quit would lose on average [responses were collapsed into a binary outcome: (0–9 years) vs. ≥10 years]. These questions, which were previously used by economists,28 correspond to information conveyed by recent French anti-smoking mass media campaigns.

Analysis

The planning horizon and impulsivity scores were analysed depending on respondents’ socio-demographic background and smoking status, using analysis of variance tests to compare mean scores across the whole sample. The determinants of smoking status were investigated by performing logistic regressions (current smoker versus never/former smokers), using a two-step approach: firstly, the socio-demographic and SES indicators were fed into the model, and secondly, time preferences were added.

Subsequent analyses were conducted on the smokers only. The same two-step approach was used to analyse their attitudes towards anti-tobacco campaigns, changes in smoking behaviour during the past 12 months, personal fear of smoking-related cancer and tobacco risk awareness. As these outcomes may depend on respondents’ consumption level and nicotine dependence, corresponding indicators were introduced into the analysis (except for the consumption levels applied when modelling reduced consumption, as the number of cigarettes smoked per day is impacted by a reduced consumption rate). As the smokers’ subsample was rather small, only the most significant covariates (selected with the stepwise method at a P < 0.1 threshold) were kept in the models. For selected covariates, odds ratios and 90% confidence intervals corresponding to the selection threshold were used.

Results

Time preferences, SES and smoking status

The correlation coefficient between the planning horizon and impulsivity scores was negative and statistically significant (−0.18, P < 0.001). In other words, the most future-oriented people are likely to be the least impulsive.

Table 1 shows the variations in the planning horizon and impulsivity scores, depending on respondents’ socio-demographic profile and smoking status. Less educated people, manual workers, the unemployed and those with low financial resources were found to be more present-oriented and more impulsive than their counterparts. Current smokers were also more present-oriented and more impulsive than ‘never-smokers’ and former smokers.

View this table:
Table 1

Time preferences depending on socio-demographic background, SES and smoking status (N = 2000)

Sociodemographic background, SES and smoking statusMean score on planning horizonMean score on impulsivity
Whole sample5.545.06
Gender
    Male (N = 978)5.444.78
    Female (N = 1022)5.64 (P = 0.069)5.33 (P < 0.001)
Age
    18–24 (N = 612)5.635.03
    25–34 (N = 608)5.565.19
    35–49 (N = 531)5.424.94
    50–75 (N = 248)5.59 (P = 0.360)5.10 (P = 0.255)
Educational level
    <High-school (N = 1167)5.425.37
    High-school, first university degree (N = 581)5.674.81
    >First university degree (N = 252)5.78 (P = 0.031)4.21 (P < 0.001)
Occupation
    Manual worker (N = 462)5.105.36
    Other (N = 1538)5.56 (P < 0.001)4.97 (P < 0.001)
Job status
    Unemployed (N = 98)5.165.45
    Employed, other (N = 1902)6.10 (P = 0.002)5.04 (P = 0.012)
Financial resources of the household
    <€1500/month (N = 404)5.105.42
    ≥€1500/month (N = 1596)5.76 (P = 0.011)4.97 (P < 0.001)
Changes in financial resources (past 5 years)
    Stability/increase/slight decrease (N = 1728)5.245.02
    Strong decrease (N = 272)5.18 (P = 0.712)5.33 (P = 0.033)
Smoking status
    Never smoked (N = 940)5.744.97
    Former smoker (N = 440)5.554.96
    Current smoker (N = 621)5.23 (P < 0.001)5.26 (P = 0.043)
  • P values in brackets have been computed with multiple t-tests for comparisons between means.

  • Reading examples: across the whole sample, the mean planning horizon and impulsivity scores were 5.54 and 5.06, respectively. The mean planning horizon score was 5.44 among men and 5.64 among women, and these means were not significantly different (P = 0.069). The mean impulsivity score was significantly higher among women (5.33 vs. 4.78, P < 0.001). Other example: the mean planning horizon score was significantly higher among the unemployed, which means that they were less future-oriented (or more present-oriented) than those with a different job status.

The second column in table 2 indicates which socio-demographic and SES indicators were found to be associated with current smoking when time preferences were not included in the analysis. Manual workers, the unemployed and those whose household’s financial resources had greatly decreased were found to be more likely to report current smoking.

View this table:
Table 2

Factors associated with current smoking (logistic regressions, N = 2000, INPES 2008)

Sociodemographic background, SES and time preferencesStep 1: Model without time preferencesStep 2: Model including time preferences
Odds ratios (90% CI)
Gender
    Male1.001.00
    Female0.80 (0.68, 0.96)0.79 (0.67, 0.95)
Age
    18–241.001.00
    25–341.10 (0.80, −1.41)1.08 (0.85, 1.45)
    35–490.60 (0.45, 0.77)0.59 (0.45, 0.77)
    50–750.22 (0.16, 0.28)0.22 (0.16, 0.28)
Educational level
    <High-school1.001.00
    High-school, first university degree0.92 (0.74, 1.14)0.98 (0.80, 1.23)
    >First university degree0.77 (0.63, 0.92)0.71 (0.48, 1.06)
Occupation
    Manual worker1.23 (1.02, 1.52)1.22 (1.03, 1.54)
    Other1.001.00
Job status
    Unemployed1.58 (1.09, 2.29)1.48 (1.08, 2.20)
    Employed, other1.001.00
Financial resources of the household
    <€1500/month0.89 (0.72, 1.11)0.99 (0.79, 1.26)
    ≥€1500/month1.001.00
Changes in financial resources (past 5 years)
    Stability/increase/slight decrease1.001.00
    Strong decrease1.45 (1.13, 1.87)1.43 (1.04, 2.13)
Planning horizon score0.93 (0.90, 0.96)
Impulsivity score1.05 (1.01, 1.10)
  • Reading examples: the odds ratio computed on the planning horizon score was significantly <1, which means that a higher score (corresponding to a more future-oriented attitude) is associated with a lower risk of reporting current smoking. The odds ratio computed on the impulsivity score was significantly >1, which means that a higher score (corresponding to more impulsive behaviour) is associated with a higher risk of reporting current smoking.

In the second-step model, planning horizon and impulsivity were found to be significant predictors of current smoking: after controlling for the effects of other indicators, it emerged that the more people were present-oriented or impulsive, the more they were likely to report current smoking. The results obtained here on socio-demographic and SES factors were similar to those obtained in the first step, except for education: once the effects of planning horizon and impulsivity had been controlled, educational level was no longer found to be a significant predictor of current smoking.

Time preferences, SES and smokers’ behaviour, attitudes and perceptions

Table 3 showed the results obtained on smokers’ personal concern with anti-tobacco campaigns and recent behavioural changes. Manual workers and smokers reporting that their income had greatly decreased during the past 5 years were less likely to feel personally concerned by anti-tobacco campaigns. In Step 2, future-oriented smokers were found to be more prone to feeling personally concerned by preventive campaigns, whereas the effect of impulsivity was not significant. The effects of the other factors tested were still similar to those observed in Step 1, except for age, which was no longer a significant factor.

View this table:
Table 3

Factors associated with smokers’ attitudes and behavioural changes (logistic regressions, N = 621, INPES 2008)

Sociodemographic background, SES, smoking behaviours and time preferencesFeeling personally concerned by anti-tobacco campaignsBehavioural changes during the past 12 months
Decreasing consumptionQuitting attempts
Step 1Step 2Step 1Step 2Step 1Step 2
Odds ratios (90% CI)
Age
    18–241.00NS1.00NS1.001.00
    25–341.66 (1.02, 2.67)0.89 (0.58, 1.39)0.29 (0.17, 0.48)0.31 (0.20, 0.51)
    35–491.02 (0.64, 1.63)0.96 (0.62, 1.61)0.33 (0.24, 0.50)0.34 (0.24, 0.50)
    50–750.67 (0.39, 1.10)0.68 (0.58, 0.94)0.26 (0.14, 0.45)0.27 (0.16, 0.46)
Educational level
    <High-schoolNSNSNSNS1.00NS
    High-school, first university degree0.86 (0.56, 1.27)
    >First university degree1.38 (1.15, 1.56)
Occupation
    Manual worker0.65 (0.45, 0.93)0.68 (0.48, 0.95)1.49 (1.10, 2.07)1.47 (1.10, 2.15)NSNS
    Other1.001.001.001.00
Job status
    UnemployedNSNS0.55 (0.35, 0.80)0.50 (0.33, 0.78)NSNS
    Employed, other1.001.00
Financial resources of the household
    <€1500/monthNSNSNSNS1.27 (1.07, 1.67)1.26 (1.05, 1.66)
    ≥€1500/month1.001.00
Changes in resources (past 5 years)
    Stability/increase/slight decrease1.001.00NSNSNSNS
    Strong decrease0.47 (0.30, 0.76)0.40 (0.23, 0.68)
Smoking the first cigarette
    Within 30 min of waking up1.59 (1.09, 2.32)1.53 (1.08, 2.19)0.61 (0.43, 0.85)0.61 (0.41, 0.86)0.54 (0.32, 0.91)0.55 (0.32, 0.92)
    Later1.001.001.001.001.001.00
Cigarette consumption rate
    >10 cigarettes per dayNSNS0.26 (0.15, 0.41)0.26 (0.14, 0.47)
    ≤10 cigarettes per day1.001.00
Planning horizon score1.10 (1.06, 1.20)1.10 (1.01, 1.12)1.06 (1.02, 1.13)
Impulsivity scoreNSNS0.94 (0.87, 0.98)
  • NS: variable not selected by the stepwise procedure. Participants’ gender does not feature in this table because it was never selected.

  • Reading examples: for personal concern with anti-tobacco campaigns, the odds ratio computed on the planning horizon score was significantly >1, which means that a higher score (corresponding to a more future-oriented attitude) is associated with a higher risk of feeling personally concerned by anti-tobacco campaigns. The odds ratio computed on the impulsivity score did not differ significantly from 1, which means that impulsivity neither increases nor decreases personal concern.

Regarding recent behavioural changes, manual workers were more likely to have reduced their cigarette consumption during the previous 12 months, whereas smokers who were unemployed were less likely to do so. Once again, the effects observed in Step 2 were similar to those observed in Step 1 (except for age), and future orientation was found to be a significant predictor of reduced consumption.

Concerning quitting attempts, several significant predictors were detected in Step 1, including high education and low financial resources. In Step 2, the effect of education was no longer significant [when this effect was forced into the model, taking ‘>first university degree’, we obtained odds ratio = 1.19 (0.79, 1.78)]. Future-oriented smokers were more prone to try to quit, and the more impulsive ones were less likely to do so.

Regarding personal fear of smoking-related cancer (see table 4), the more educated smokers and those whose income had decreased considerably during the past 5 years were less likely to express this fear. In Step 2, the effect of a large income decrease disappeared, and people with a higher planning horizon score, in other words people who were more future-oriented, were found to be more prone to report personal fear of tobacco-related cancer.

View this table:
Table 4

Factors associated with smokers’ fear and risk awareness (logistic regressions, N = 621, INPES 2008)

Personal fear of smoking-related cancerRisk awareness
≥50/100 smokers will die of a tobacco-related diseaseLifetime smokers lose at least 10 years of life expectancy
Step 1Step 2Step 2Step 1Step 2
Odds ratios (90% CI)
Age
    18–241.001.001.001.001
    25–342.31 (1.61, 3.28)2.27 (1.48, 3.37)1.58 (1.06, 2.43)1.00 (ns)0.99 (0.48, 1.25)
    35–491.07 (0.73, 1.64)1.02 (0.65, 1.57)1.30 (0.80, 2.15)0.77 (0.48, 1.25)0.78 (0.48, 1.25)
    50–750.75 (0.45, 1.21)0.68 (0.41, 1.12)0.66 (0.46, 0.98)0.41 (0.24, 0.70)0.41 (0.25, 0.69)
Educational level
    <High-school1.001.001.00NSNS
    High-school, first university degree1.31 (0.92, 1.86)1.29 (0.51, 2.87)0.83 (0.71, 1.44)
    >First university degree0.61 (0.38, 0.96)0.64 (0.41, 1.00)0.49 (0.30, 0.82)
Occupation
    Manual workerNSNS0.63 (0.43, 0.95)NSNS
    Other1.00
Financial resources of the household
    <€1500/monthNSNSNS0.67 (0.41, 1.00)NS
    ≥€1500/month1.00
Changes in resources (past 5 years)
    Stability/increase/slight decrease1.00NSNS1.00NS
    Strong decrease0.69 (0.42, 1.00)0.63 (0.38, 0.96)
Smoking the first cigarette
    Within 30 min of waking upNSNSNSNSNS
    Later
Cigarette consumption rate
    >10 cigarettes per day2.19 (1.50, 3.18)2.37 (1.61, 3.47)NSNSNS
    ≤10 cigarettes per day1.001.00
Planning horizon score1.23 (1.15, 1.30)NS1.07 (1.01, 1.13)
  • NS: variable not selected by the stepwise procedure. Participants’ gender, job status and impulsivity scores do not feature in this table because they were never selected as significant predictors of the modelled outcomes.

  • Reading examples: when modelling personal fear of smoking-related cancer, the odds ratio computed on the planning horizon score was significantly superior to 1, which means that a higher score (corresponding to a more future-oriented attitude) is associated with a higher risk of reporting this fear.

Concerning the first indicator of risk perception (believing that at least 50 smokers of 100 will die of a smoking-related disease), only the results of the Step 2 model are presented, as neither of the time preference indicators was a significant predictor. Manual workers and the more highly educated were less likely to endorse this opinion. Concerning the second indicator, low-income smokers and those whose income had decreased considerably were less prone to believe that lifetime smokers lose at least 10 years of life expectancy. In Step 2, these effects became non-significant, whereas the planning horizon score was found to be a significant predictor. Overall, some of the indicators of a low SES and present orientation were predictive of a lower propensity to fear/perceive the dangers of cigarette smoking.

Discussion

The potential endogeneity of time preferences

The present study provides strong evidence that time preferences and SES are correlated: respondents with a low SES were found to be more present-oriented and impulsive. As discussed previously, socioeconomic hardships may shorten people’s time horizons. In addition, schooling focuses students’ attention on the future and may help them learn patience and planning skills.15 But, conversely, a low SES might also result from present time orientation and impulsivity: for example, self-control and the propensity to plan ahead may contribute to academic achievement and the acquisition of wealth.29,30

This study also shows that present-oriented and impulsive people are more prone to cigarette smoking, as reported by Khwaja et al.17 In addition, impulsive smokers and the less present-oriented were found to be less likely to report having made recent quitting attempts, whereas previous studies showed the existence of a positive correlation between impulsivity, present orientation and unsuccessful attempts to quit.5,31 Time preferences may impact smoking behaviourm,5,17,31 but conversely, addictive behaviour may alter people’s time preferences and induce them to discount the future more strongly.15 Some empirical evidence has been published on these lines.32 Likewise, it was established in the present study that former smokers were less present-oriented and less impulsive than current smokers. The potential endogeneity of time preferences suggests that the impact of time preferences on smokers’ behaviour, attitudes and beliefs may have been overestimated here, as this impact may in fact partly reflect the effects of smoking on time preferences.

SES, time preferences and smoking behaviours

Previous studies have shown the existence of relationships between a low SES and cigarette smoking on the one hand, and between present orientation, impulsivity and cigarette smoking on the other hand. But to our knowledge, the relationships between SES, time preferences and smoking have rarely been investigated simultaneously.33 The results obtained here suggest that time preferences partly mediate the effects of SES on smoking behaviour: when smoking status and recent quitting attempts were modelled, the effects of educational level (a key component of SES) became non-significant when the time preferences were introduced into our analyses. In a previous modelling study on smoking behaviour, although the educational coefficient decreased when the time preference was included in the model, it was still statistically significant, which indicated that education still had substantial exogenous effects on smoking.33

In other respects, the results obtained here suggest that it would be worth extending these investigations to aspects of cigarette smoking other than the smoking status alone. SES and time preferences were found here to have a significant impact on smokers’ attitudes, fears and risk perception, and the effects of some low SES indicators disappeared after horizon planning was introduced into the analysis. For example, present-oriented smokers were less likely to feel personally concerned by anti-tobacco campaigns and to fear smoking-related cancer, probably because feeling personally concerned by the risk of smoking-related cancer proclaimed in preventive messages requires the ability to make future adverse effects on health seem less remote.

Limitations of the study

The present study has several shortcomings that have to be acknowledged. First of all, as our smokers’ subsample was rather small, the statistical power of the study was fairly low. Secondly, the planning horizon and impulsivity scales used here need to be studied in greater depth to test their accuracy and their reliability. Although similar statements to those presented to the respondents in this study have been previously used to determine general attitudes,17–19 other studies have suggested that time preferences may be domain-specific (an individual may be present-oriented when dealing with professional issues, but future-oriented when faced with health issues).34 Finally, this study was not designed to take the endogeneity issues discussed previously into account.

Conclusion

Time preferences are a key determinant of smokers’ behaviour, but they also strongly influence smokers’ attitudes and beliefs about prevention and smoking risks. Tobacco control policies might be made more effective by including messages specifically targeting present-oriented smokers and/or interventions designed to enhance awareness of the future in these populations. Interventions of this kind could potentially improve multiple health outcomes, as time perspective is also predictive of health-related behaviour other than smoking, such as eating habits and physical activity.

Funding

The French Institute for Health Promotion and Health Education (INPES) and the French Institute for Public Health Research (IRESP) supported this study financially.

Conflicts of interest: None declared.

Key points

  • In developed countries, the prevalence of smoking is highest in populations with a low SES.

  • Time preferences may partly mediate the effects of low SES on smoking behaviour, as present time orientation (giving little thought to the future/living for the moment and not considering the future consequences of present behaviour) is correlated with both low SES and cigarette smoking.

  • Our results support the hypothesis that time preferences partly mediate the effects of SES on smoking behaviour.

  • Our results also suggest that this hypothesis could be extended to include other aspects of smoking-related outcomes, including risk perceptions and attitudes towards anti-smoking media campaigns.

  • In studies on time preferences, planning horizons and impulsivity should be both taken into account.

  • Tobacco control policies should include messages specifically targeting present-oriented smokers and/or interventions designed to enhance future orientation among the populations targeted.

Acknowledgements

The authors thank all the people who agreed to participate in this survey.

References

View Abstract