Tobacco Documents Online

I I I I I I I I I I I I I I I I Addiction (1994) 89, 321-330 Pharmacological and non-pharmacological smoking motives: a replication and extension JAMES C. TATE, CYNTHIA S. POMERLEAU & OVIDE F. POMERLEAU University of Michigan Behavioral Medicine Program, Ann Arbor, Michigan, USA Abstract Cigarette smokers (n = 387) completed a questionnaire measure of smoking motives, and subgroups of this sample provided external validation information. Seven factors emerged from a primuipal components' analysis: automatic, sedative, addictive, stimulation, psychosocial, indulgent and sensorimotor manipulation. A higher-order principal components analysis revealed the presence of two second-order factors. Inspection of the pattern of correlations between factor scores and criterion variables clearly indicated that the first four factors above and their underlying second-order factor are more closely related to nicotine pharmacology and mood-altering effects of nicotine than the latter three motives and their underlying second-order factor. Moreover, the positive correlations between these pharmacological motives and age, coupled with a negative relationship between age and the non-pharmacological motives, support the description of the smoking career as a progressive transfer of reward from non-pharmacological to pharmacological factors. These findings suggest that self-reported reasons for smoking represent more than bias in verbal report. Introduction The question, "why do people smoke?" concisely summarizes the thrust of a large portion of cigarette smoking research conducted over the last three decades. A direct approach to answer- ing this question is to ask smokers their reasons for smoking. Beginning in the mid-1960s, re- searchers began constructing questionnaires designed for this purpose and analyzing re- sponses to these questionnaires to detect evidence of smoking motives. The development of reliable, valid self-report measures of smok- ing motives is important for both basic and ap- plied reasons. Smoking motive scales would allow measurement of private events mediated by the proposed neuro-regulatory effects of nicotine Correspondence to: Chris Tate, Ph.D., PO Box 87, Depart- ment of PsycholoQy, Middle Tennessee State University, Murfreesboro, TN 37132, USA. 1 (Pomerleau & Pomerleau, 1984). Relating smok- ing motive scores to external criterion variables would help to separate pharmacological and non-pharmacological factors in cigarette smok- ing and further our understanding of smoking. Moreover, the identification of specific smoking motives could guide the tailoring of smoking cessation treatment to the individual needs of patients (Kreitler, Shahar & Kreitler, 1976). Smoking motive questionnaires have been based on affect management models (Ikard, Green & Horn, 1969), situations associated with smoking (McKennell, 1970; Best & Hakstian, 1978) and arousal modulation models (Frith, 1971). Because the questionnaires used in these studies were not identical, different sets of smok- ing motives were found. Nevertheless, the large degree of content overlap resulted in the finding of similar sets of smoking motives, and the re- sults of factor analytic studies demonstrate excellent consistency (Tate, Schmitz & Stanton, 321

I I I I I I I I I I I I I I I I I I I .!' 322 James C. Tate, Cynthia S. Pomerleau & Ovide F. Pomerleau 1991). Different researchers labeled motives dif- ferently, but the following are representative names of the most commonly found motives: automatic (ATM), sedative (SED), addictive (ADD), stimulation (STM), psychosocial (SOC), indulgent (IND) and sensorimotor ma- nipulation (SMM). Russell, Peto & Patel (1974) approached the question of smoking motives from the perspective of pharmacological versus non-pharmacological rewards of smoking. Using a 34-item questionnaire derived from both the Ikard et al. (1969) and McKennell & Thomas (1970) measures, Russell et al. (1974) recovered six of the seven smoking motives (ATM, ADD, STM, SOC, IND & SMM) and found evidence, via a higher-order factor analysis, of two more basic dimensions underlying these smoking mo- tives. The first dimension, described as representing the pharmacological rewards of nic- otine, was composed of the ATM, ADD and STM smoking motives. Although the expected SED factor did not emerge, these investigators suggest that it should lbad on the pharmaco- logical dimension. The second, non- pharmacological dimension was composed of the SOC, IND and SMM smoking motives. Russell et al. (1974) speculated that social and other non-pharmacological rewards motivate smoking initially and account for the stronger role of the SOC, IND and SMM motives early in the smoker's career. Eventually, the positive re- wards of nicotine, due to its direct and indirect actions on the brain, exert more control as the smoker increasingly uses nicotine to modulate arousal and affective tone, thus accounting for the stronger role of the SED and STM motives as the smoker's career progresses. As nicotine intake increases and the pattern of intake be- comes more regular, avoidance and relief of withdrawal become paramount, and the ATM and ADD motives becomes stronger. Thus, smokers progress along two orthogonal dimen- sions as they continue to smoke. Initially, non-pharmacological rewards exert greater con- trol over smoking; however, pharmacological rewards develop greater control. The validity of the pharmacological dimension was supported by moderate positive correlations (i.e., r= 0.50-0.63) between STM, ADD, ATM and SED motive scores and smoking rate. Also, this dimension discriminated between a sample of normal smokers not attempting to quit and a sample of addicted heavy smokers attending smoking cessation clinics. Specifically, clinic smokers scored higher on the pharmacological smoking motives. The non-pharmacological mo- tives did not correlate as highly with smoking rate, and the SOC motive correlated negatively with age as predicted (Russell et al., 1974). Addi- tional validation comes from three subsequent studies (Niaura et al., 1989; West, Hajek & Belcher, 1986; West & Russell, 1985) that demonstrated predicted relationships between pharmacological smoking motive scores and vari- ous criterion variables (i.e., withdrawal symptomatology, smoking rate, expired carbon monoxide, cotinine levels, etc.). These studies lend support to the validity of Russell et al.'s (1974) pharmacological and non- pharmacological dimensions, but there are limitations. Limited sample sizes (n = 29-77) limit the reliability of the findings. Indeed, the correlation between ATM scores and smoking rate is the only consistent finding across the studies. Also, some results contrary to Russell et al.'s model emerged. Russell et al. (1974), West & Russell (1985), and West et al. (1986) found that IND motive scores behaved more like those belonging to the pharmacological group (i.e., correlated positively with criterion variables). Similarly, Niaura et al. (1989) found significant positive correlations between IND and SMM scores and withdrawal and urge to smoke scores. Finally, in Russell et al. (1974), the ratio of subjects to questionnaire items used in making factor estimations (i.e., 5 to 1) is close to the minimum rule of thumb guide of 4 to 1(Kim & Mueller, 1978). It is possible that this fact ex- plains the failure to obtain an SED factor. Given that the stability of a factor solution increases as this ratio increases, replication in a large sample would increase confidence in this model of smoking behavior. The general goal of this study was to examine the relationships between self-reports of smoking motives and external validation criteria in an attempt to establish the pharmacological under- ~ pinnings of specific reasons for smoking. As a~ first step towards demonstrating this, the follow- ~ ing sub-goals were set and specific predictions ~ made. First, we attempted to recover the first-or- ~ der and second-order factor structures described ~ by Russell et al. (1974). Secondly, to establish ~ the pharmacological and non-pharmacological t\L natures of these factors, we investigated the rela- C_~

I I I I I I Pharmacological and non-pharmacological smoking motives 323 tionships between smoking motive scores and an array of pharmacological marker variables. We predicted that the ATM, SED, ADD, STM and the underlying second-order factor would corre- late positively with plasma cotinine, smoking rate, length of time smoking, age and a question- naire measure of nicotine dependence and negatively with latency to the first cigarette of the day. Also, we predicted that the SOC, IND, SMM and their underlying second-order factor would correlate negatively with age and mini- mally, mally, or not at all, with the criterion measures. s Thirdly, we predicted that a sub-sample of sub- jects recruited to participate in smoking cessation I trials would have higher scores on the pharmaco- logical motives and lower scores on the non-pharmacological motives as compared to a r sub-sample of regular smokers not attempting ~ smoking cessation. Finally, we predicted positive relationships between pharmacological motive scores and measures of trait anxiety and de- pression. It was reasoned that individuals ~ reporting higher levels of subjective distress due to personality make up, negative environmental events or nicotine withdrawal would be more likely to report stronger pharmacological reasons ~ for smoking. I Method Sub~'ects Subjects were 387 cigarette smokers (179 fe- males, 208 males) who participated in research ~ projects carried out at the University of Michi- gan's Behavioral Medicine Program between 1986-1993. Two hundred and fifty-three of these subjects (106 females, 147 males) partici- pated pated in laboratory projects, and 134 (73 ~+ females, 61 males) were participants in clinical trials involving smoking cessation. Laboratory ~ subjects were chosen to be healthy, moderately dependent smokers and may not be representa- tive of all smokers. Inclusion criteria for clinic subjects were somewhat less restrictive and ~ specifically included a current desire to quit smoking. I Measures Measures completed by most subjects included a smoking history form, the Fagerstrom Tolerance Lushene, 1970) and depression (CES-D; Weiss- man et al., 1977) measures, and a modified version of the Russell et al. (1974) smoking motives questionnaire (SMQ; Pomerleau et al., 1992). The modified SMQ was constructed by examining the factor structure reported in Rus- sell et al. (1974) and, for each factor, selecting the three items with the highest significant (i.e., _- 0.40) loadings in conjunction with non- significant loadings on the other factors. Only two such items could be found for the ATM and SED motives; consequently, a third item was rationally constructed for each of these scales. Appendix A contains the resulting scales. Re- sponses to items are made on a 0 (not at all) to 3 (very much so) scale. Additionally, 297 sub- jects supplied plasma samples which were assayed for cotinine concentration via either high-pressure liquid chromatography (Hariha- ran, Van Noord & Greden, 1988) or gas chromatography (Jacob, Wilson & Benowitz, 1981). Due to changes in study demands and missing data, criterion variables data were in some instances available for only subsets of the database. Procedure Questionnaire materials were provided at a screening session and subjects completed these prior to actual involvement in a study. Data were inspected for completeness and entered into the database. Plasma samples were obtained prior to any experimental manipulations or provision of treatment. In some cases, these samples were collected after overnight smoking abstinence. Results Subjects Table 1 shows descriptive data on the total sam- ple, by sex, and laboratory/clinic subgroups. Generally, subjects were young, moderate smok- ers who reported smoking for an average of 16 years and being moderately dependent on nicotine. Univariate analyses of variance (ANOVA) testing for sex differences on these variables were non-significant with the exception of plasma cotinine concentration [F(l, 295) = 5.20, p = 0.023] with males having the higher concentration. Clinic and laboratory sam- ples differed significantly on age [F(l, 385) = 236.87], years smoked [F (1, ~ Questionnaire (FTQ; Fagerstrom, 1978), trait anxiety (STAI-Trait; Spielberger, Gorsuch & I ~w

I I I I I i I I I I I I I I I I 1 I I -0 324 ,3`ames C. Tate, Cynthia S. Pomerleau & Ovide F. Pomerleau Table 1. Subject characurran'ss Variable Males Females Lab. Clinic Total Age (yrs) Mean 32.84 32.80 28.86 40.31 32.82 SD (9.10) (8.88) (6.38) (7.95) (8.84) n 208 179 253 134 387 Plasma cotinine (ng/ml) Mean 281.67 250.12 240.95 306.92 268.71 SD (124.22) (106.56) (120.25) (104.05) (118.13) n 175 122 172 125 297 Smoking rate (cigs/day) Mean 27.48 25.83 23.55 32.54 26.71 SD (11.99) (8.38) (7.57) (12.44) (10.47) n 203 179 248 134 382 Years smoked Mean 16.65 17.16 13.00 22.99 16.88 SD (9.06) (8.32) (6.71) (8.02) (8.72) n 189 154 210 133 343 FTQ Mean 7.23 7.32 6.88 7.76 7.27 SD (1.69) (1.74) (1.80) (1.44). (1.71) n 165 158 244 99 323 Latency* (min) Mean 24.04 28.48 25.80 SD 31.85 27.35 30.84 n 73 48 121 * Data available for laboratory subjects only. 341) = 151.67], smoking rate 267) = 76.92], plasma cotinine 295) = 24.36], and FTQ [F(l, 321) = [F(l, [F(1, 18.31] (all p= 0.0001). Clinic subjects had higher val- ues on all of these variables. Factor analysds First-order analysis. Responses to the 21 SMQ items were entered into a principal components analysis. Using Kaiser's (1974) criterion (eigen- values ;~, 1) to determine how many factors to retain, a clear seven-factor solution, accounting for 65% of the total variance, emerged. The seven factors each accounted for 23, 10, 9, 7, 6, 5 and 5% of the total variance, respectively. Following factor extraction, initial rotation was achieved via the varimax method, and rotation to a terminal solution was achieved via the oblique promaz method. Table 2 shows the factor loadings greater than 0.40 after rotation, communalities (It2), and Kaiser's Measure of Sampling Adequacy (MSA; Kaiser, 1970, 1974). Comparison of Table 2 and Appendix A indicates that items loaded mainly on only one factor and the predicted pattern of factor loadings was obtained. Only six of 147 factor loadings were contrary to predictions to a significant degree. Two factors (ATM, IND) emerged exactly as predicted, and the SED and SMM factors emerged with an additional item loading significantly on each. For each of the SOC, ADD and STM scales, two of the three items behaved as predicted. Kaiser's (1970, 1974) MSAs were calculated for each item and overall. This statistic is a measure of the degree to which the variance in each variable is accounted for by common fac- tors, that is, the level of factorial determination (Kim & Mueller, 1978). The MSA can vary between 0 and 1, with higher values indicating greater factorial determination and empirical confirmation of a given factor solution. The last column of Table 2 contains MSAs for individual items. The overall MSA is 0.77. Using Kaiser's guidelines for interpretation, these values are quite acceptable and support the appropriateness of the seven-factor solution. Second-order anaFysis. Inspection of the corre- lation matrix in Table 3 reveals moderate inter-correlations among the seven factors with

I i I Pharmacological and non-pharmacological smoking motives 325 I Table 2. Factor loadings* from first-order factor analysir Factor I Item ATM SED SOC IND ADD SMM STM h2 MSA 3 0.91 0.81 0.67 1 0.84 0.71 0.75 2 0.78 0.64 0.81 4 0.76 0.61 0.84 I 6 0.69 5 0.69 0.64 0.83 0.64 0.84 9 0.67t - 0.46t 0.64 0.81 8 0.81 0.70 0.75 ~ 7 0.81 0.76 0.77 ,~ 20 0.54j- 0.44t 0.55 0.87 11 0.84 0.70 0.61 10 0.84 0.70 0.60 12 0.53 0.46 0.73 I 15 0.77 0.63 0.77 13 0.74 0.65 0.84 16 0.84 0.71 0.66 14 0.53 0.59 0.82 ~ 17 0.52t 0.57 0.63 18 0.40t 0.44 0.58 0.67 19 0.87 0.74 0.71 21 0.68 0.66 0.80 I * Standardized regression coefficients. Loadings < 0.40 not shown. t Loadings contrary to pre- dictions. 1coefficients ranging from - 0.05 to 0.37. Consequently, the seven factors were arranged uch that more highly correlated factors are ad- acent, forming two groups of intercorrelated iii~~~ctors. This prompted a second-order principal components analysis. Two second-order factors, I espectively accounting for 28% and 18% of the otal variance, emerged using Kaiser's criterion to terminate factor extraction. Rotation to simple structure was achieved initially via the varimax tethod and terminally via the promax method. able 4 contains the seven first-order factors and their loadings on the two second-order factors. The two second-order factors demonstrated a ~ow inter-correlation (r = 0.15). '.. alidiry I'able 5 contains correlations between factor scores and criterion variables. Five of these vari- •ables relate directly to cigarette consumption, cotine intake and nicotine dependence (plasma otinine, smoking rate, years sinoked, FTQ, la- tency to the fust cigarette of the day). Age was because Russell et al. (1974) postulated Icluded positive relationship between age and strength of the pharmacological rewards of smoking and a negative relationship between age and the strength of the non-pharmacological rewards of smoldng. Due to the large number of correla- tions, minimum significance was set at pt!G 0.01. Factor I, ATM and ADD scores correlated as predicted with all criterion variables. The STM scores correlated as predicted with four of six criterion variables, and SED scores correlated with number of years smoking. With the excep- tion of a significant positive correlation between SOC scores and smoking rate, none of the corre- lations involving the SOC, IND, SMM and Factor II scores were significant. There was, however, a trend (p < 0.05) for IlVD and Factor II scores to correlate negatively with age. Focus- ing only on the correlations between second-order factor scores and criterion vari- ables, Student's t-statistic was computed to test whether differences between correlation coefficients are significant. In every case, the difference was significant. Correlations between Factor I and pharmacological marker variables are significantly larger than comparable correla- tions involving Factor II. Next, motive factor scores for the clinic and

I I I i I I I 1 I I I I I I I I I I 326 ,fames C. Tate, Cynthia S. Pomerleau f7 Ovide F. Pomerleau Table 3. Inter-facror correlations ATM SED ADD STM SOC IND SMM ATM 1.00 SED 0.25 1.00 ADD 0.25 0.37 1.00 STM 0.23 0.14 0.23 1.00 SOC 0.19 0.24 0.07 0.20 1.00 IIZD 0.00 - 0.05 0.00 - 0.02 0.22 1.00 SMM 0.25 0.07 0.05 0.08 0.20 0.07 1.00 laboratory sub-samples were compared using univariate ANOVAs. Significant results were ob- tained for the STM [F(1, 385) = 27.81, p= 0.0001), ADD [F(i, 385) = 88.62, p= 0.0001], SED [F(1, 385) = 25.09, p= 0.0001], IND [F(1, 385) = 7.08, p= 0.008) and SMM [F(1, 385) = 4.04, p= 0.045] mo- tives. Clinic subjects had higher STM, ADD, SED and IlVD scores and lower SMM scores than the laboratory sub-sample. This analysis was repeated with the second-order factor scores as dependent variables, and similar results were obtained. Clinic subjects had higher Factor I scores• [F(1, 385) = 73.49, p= 0.0001]. The ANOVA involving Factor II scores was non- significant [F(1, 385) = 0.06, p= 0.81]. To assess further the validity of the smoking motive factors, factor scores were correlated with measures of trait anxiety and depression (see Table 6). The SED, ADD, Factor I scores corre- lated significantly with STAI scores. The SED scores correlated with CES-D scores also. Con- trary to predictions, SOC scores correlated significantly with STAI scores. Table 4. Fauor loadi'ngs* frons tlu second- order factor analysu Second-order factors First-order factors Factor I Factor II ADD 0.73 - 0.15 SED 0.70 - 0.02 ATM 0.61 0.13 STM 0.53 0.10 ' SOC 0.23 0.71 IND - 0.25 0.69 SMM 0.09 0.58 * Standardized regression coefficients. Discussion The goal of this study, to exam±ne smoking motives systematically in a larger population, were largely met. The findings replicate and extend the results of previous investigations of self-reported smoking motives and their relation- ships to nicotine pharmacology. Seven smoking motives were identified by factor analysis, and a second-order factor analysis revealed the pres- ence of two more basic dimensions. These first- and second-order factors are strikingly similar to those obtained by Russell et al. (1974) in terms of item composition, inter-factor correlations and variance accounted for. However, we found a much clearer separation between the second- order factors as evidenced by the unequivocal pattern of second-order factor loadings and the low inter-factor correlation. In addition, pre- dicted relationships between smoking motive factor scores and external criterion variables were observed. Correlations between most of these dimen- sions and pharmacological markers demonstrated predicted patterns, thus lending credence to the pharmacologicai and non-phar- macological labels applied by Russell et al. (1974). Factor I clearly seems more pharmaco- logical in nature than Factor II. Furthermore, STM, SMM, ADD, SED and Factor I scores discriminated between a group of heavy, ad- dicted smokers seeking smoking cessation treatment and a group of less'dependent smok- ers. Lastly, subjects with higher SED scores reported being generally more anxious and de- pressed, subjects with higher ADD scores reported being more anxious, and subjects with higher Factor I scores reported being more anx- ious. These results imply that smokers who experience more subjective distress may derive greater reinforcement and/or relief of dysphoric states from smoking. • I

I I I Pharmacological and non-pharmacological smoking motives 327 I Table 5. Correlation.s between factor scores and criterion variables Factor Plasma cotinine (n = 297) Smoking Years rate smoked Age FTQ Latency (n = 382) (n = 343) (n = 387) (n = 323) (n = 121) I ATM 0.24*** 0.48*** 0.33*** 0.29*** 0.27*** - 0.28* SED - 0.03 0.09 0.14* 0.07 0.12 - 0.10 ADD 0.21** 0.39*** 0.34*** 0.32** 0.33*** - 0.28* STM 0.11 0.28*** 0.25*** 0.24*** 0.16* - 0.02 ~ ~ Factorl 0.20** 0.47*** 0.40*** 0.35*** 0.33*** - 0.24* SOC - 0.04 0.15* 0.01 - 0.01 0.14 - 0.10 IND -0.01 -0.08 -0.07 -0.11 0.10 -0.04 ,~ SMM -0.10 0.12 -0.10 -0.09 -0.07 0.10 Factor II - 0,08 0.09 - 0.08 - 0.10 0.09 - 0.02 t 3.96** 6.43** 7.60** 7.33** 3.57** 3.99** I *p<0.01,**p<0.001,***p<0.0001. Not all results were as predicted. Clinic sub- 6ects had higher IND scores than laboratory ~~~ubjects. Otherwise, IND scores behaved as non-pharmacologically based. This motive, originally postulated as straddling the pharma- ological/non-pharmacological dimension, was und by Russell et al. (1974) to load on a non-pharmacological second-order factor and o correlate significantly with daily smoking rate, seemingly paradoxical finding. West et al. 1986) and West & Russell (1985) reported significant positive correlations between II,1D ores and pharmacological marker variables. ~ us, there is precedent for ambiguity concern- g g the IND smoking motive, and further research will be needed to clarify the nature of Ws motive. Also, the non-pharmacological SOC motive correlated with smoking rate and trait anxiety t ores. It is possible that some anuous individu- I I I I Table 6. Relationships between smoking motive factor scores, trait anxaery and depression Trait anxiety Depression Factor (n = 372) (n = 200) ATM 0.05 0.07 SED 0.16* 0.25** ADD 0.14* 0.06 STM 0.03 - 0.07 Factor I 0.14* 0.13 SOC 0.16* 0.06 IND -0.09 -0.05 SM.M 0.09 - 0.03 Factor 11 0.07 - 0.005 *p<0.01,**p<0.001. als score highly on the SOC motive because they experience heightened anxiery when in social situations, and the act of smoking may represent a pharmacological coping device (Pomerleau & Pomerleau, 1984). Two pieces of evidence sup- port this interpretation. First, the two items loading highest on the SOC factor (items 10, 14) contain content suggesting increased ease and confidence via smoking in social situations. This is conceptually similar to the avoidance or re- duction of negative affect central to the concept of SED smoking. Secondly, the SOC and SED motives are correlated significantly in this sample (see Table 3). Although plausible, the present correlational design cannot settle the issue. Finally, the SED factor's showing was not as consistent or as strong as expected. The SED scores correlated as predicted with number of years smoked, trait anxiety and depression; demonstrated a marginal trend with respect to FTQ scores and latency to the first cigarette of the day; but were unrelated to plasma cotinine, smokting rate and age. Russell et al. (1974) failed to find evidence of an SED factor. As with the IND factor, there is precedent for inconsistency, and further research may shed light on the mat- ter. These findings have several implications. First, individuals who report smoking to increase/de- crease arousal or because of habit or addiction are likely to be using nicotine for its pharmaco- logical effects. Thus, the stimulatory and sedative functions of smoking are probably medi- ated by mcotine. Moreover, the automaticity of smoking may serve to maintain nicotine levels above a certain threshold, thus preventing with- I

I I I I I I 1 I I I I I I I I 328 ,3`ames C. Tate, Cynthia S. Pomerleau & Ovide F. Pomerleau drawal. When abstinence is unavoidable, dis- comfort and craving motivate drug seeking. Although the present study does not represent a direct test of these assertions, the general finding that the ATM, SED, ADD and STM smoking motives are more strongly related to external validation criteria than the SOC, IND and SMM motives is consistent with the view of smoking being motivated by both the mood-altering ef- fects of nicotine and avoidance/relief of nicotine withdrawal (Pomerleau & Pomerleau, 1984). The positive correlations between SED and Fac- tor I scores and mood and anxiety further support this interpretation. Secondly, positive correlations between phar- macological factors and length of time smoking and age support the description of the smoking career as a progressive transfer of reward and control from predominantly non-pharmacologi- cal to predominantly pharmacological factors. A longitudinal design would be needed to support this interpretation unequivocally, but the current data show small to moderate significant relation- ships in predicted directions and are consistent with the development of increasing tolerance to nicotine as a function of years of smoking described elsewhere (Henningfield & Nemeth-Coslett, 1988). Stronger negative rela- tionships between age and the non- pharmacological factors might have been obtained if younger smokers (age < 18 yrs) were included in the sample. Russell et al. (1974) recruited subjects as young as 16 years of age and obtained a correlation between age and SOC scores of - 0.23. More recent support for this interpretation comes from a longitudinal study by Stanton et al. (1993) in which reasons for smoking were assessed in a cohort of children at ages 11 and 13 years. They found that the im- portance of "image" smoking ("I look better with a cigarette in my hand") declined significantly during the intervening two years, and smoking because of "friends" ("I smoke because I don't want to be the odd one out in a group") demonstrated a statistically significant, but small, degree of consistency. Possibly, the rapidity with which people become nicotine de- pendent renders psychosocial motives less important in a similarly rapid fashion. A review of the RFS (Tate et al., 1991) docu- mented that groups of smokers report remarkably similar motives with excellent con- sistency; however, analogue, self-monitoring and treatment studies support neither the validity nor clinical utility of the individual motive scales. Although the literature is fraught with methodo- logical problems, this finding raises the issue of the relationship between self-reported smoking motives and actual smoking behavior. Schachter (1978) argued that the "psychological and prob- ably the sensory and manipulative gratifications of smoking are illusory" (p. 112). Because the RFS and SMQ share item content, this issue is relevant to the SMQ. The results presented here demonstrate clearly that some smoking motives have stronger pharmacological ties than other motives. Consequently, motive scores represent more than self-report bias. Nevertheless, the specific nature of these ties remains speculative. Consequently, the answer to the question of why people smoke continues to be elusive. A partial reason for the difficulty may be that nic- otine is a drug with multiple pharmacological properties and that the tenacity of cigarette smoking is based upon the diversity of nicotine's actions (Pomerleau & Pomerleau, 1984). In view of the apparent ability of smokers to secure a wide range of effects, with reinforcement value modulated by ongoing activities and environ- mental context, a productive use of the findings of motives-for-smoking research might be to pro- vide guidance for systematic investigations of the circumstances that trigger smoking and the subjective, behavioral, physiological and neuroendocrine consequences of nicotine self- administration. Such research may resolve some of the contradictions and ambiguities of the ini- tial attempts to classify smokers and, ultimately, should lead to a better understanding of nicotine dependence and cigarette smoking. Acknowledgements Preparation of this manuscript was supported by National Institute on Drug Abuse Grant DA 06529 and by National Cancer Institute Grant CA 42730 to the third author. The authors are grateful to Samuel B. Green and Mary Lutzke for their valuable assistance. Correlation ma- trices are available from the first author. Z11:) References ~ Bnsr, A. J. & HAxsnt+N, A. R. (1978) A situationC' ~ specific model for smoking behavior, Addica%.-, Behaviors, 3, pp. 79-92. © I

i I I Pharmacological and non-pha»nacological smoking motives 329 _ FAGERSTROM, K. O. (1978) Measuring degree of physi- cal cal dependence to tobacco smoking with reference to individualization of treatrnent, Addicrive Behaviours, 3, pp. 235-241. ~ FtuTti, C. D. (1971) Smoking behaviour and its rela- tion to the smoker's immediate experience, British Journal of Social and Clinical Psychology, 10, pp. 73-78. i HARtHAN, M., VANNooRi), T. & GxEDEN, J. F. (1988) ~ A high-performance liquid-chromatographic method for routine simultaneous determination of nicotine and cotinine in plasma, Clinical Chemistry, 34, pp. 724-729. ~ HENNirrGFtEin, J. E. & NEMETH-Costfrr, R(1988) Nicotine dependence: interface between tocacco and tobacco-related disease, Chesr (Supplement), 93, pp. 37S-55S. ~ Ixnttn, F. F., GxEEN, D. E. & HoxN, D. (1969) A ~~ scale to differentiate between types of smoking as related to the management of affect, Inurnational Journal of the Addictions, 4, pp. 629-639. ~ JACOB, P., WILSON, M. & BENOwnz, N. L. (1981) Improved gas chromatographic method for the de- termination of nicotine and cotinine in biologic fluids, Journal of Chromatography, 222, pp. 61-70. KA1SEFt, H. F. (1970) A second-generation little jiffy, Psychometrika, 35, pp. 401-415,, ~ KAtsEtt, H. F. (1974) An index of factorial simplicity, ~ Psychometrika, 39, pp. 31-36. Ki.`s, J. & MvEU.Ett, C. W. (1978) Factor Analysis: Statistical Met3wds and Practical Issues (Beverly Hills, Sage). KxErrt.ER, S., SHAxAR, A. & KRErrt.ER, H. (1976) Cognitive orientation, type of smoker and behaviour therapy of smoking, British Journal of Medical Pry- chology, 49, pp. 167-175. ~ McKENNELt,, A. C. (1970) Smoking motivation fac- tors, British Journal of Social and Clinical Psychology, 9, pp. 8-22. ~ N1AUR+, R, Go1.DsTm,r, M. G., WARD, K. D. & , ABRAMS, D. B. (1989) Reasons for smoking and severity of residual nicotine withdrawal symptoms when using nicotine chewing gum, British Journal of Addiction, 84, pp. 681-687. ~ PomFRLEAu, O. F. & PoMExr.Fr,u, C. S. (1984) Neu- roregulators and the reinforcement of smoking: towards a biobehavioral explanation, Neuroscience & Biobehavioral Reviews, 8, pp. 503-513. ~ PoMERLEau, C. S., PoMERL.EAu, O. F., FtEsstarm, K A. & BASSON, S. M. (1992) Relationship of tridimensional personality questionnaire scores and smoking variables in female and male smokers, Journal of Substancu Abuse; 4, pp. 143-154. ~ RussEU., M. A. H., Psro, J. & PATEL, U. A. (1974) i The classification of smoking by factorial structure of motives, Journal of the Royal Statistical Sociay, A, 137, pp. 313-342. ~ ScHAcxTHx, S. (1978) Pharmacological and psycho- logical determinants of smoking, Annals of Internal Mtdiccine, 88, pp. 104-114. SprEr.BERCEti, C. D., Gotzsucx, R. L & LustrnarE, - R E. (1970) Manual for rlu State Trait Anxiay _'. Invenrory (Palo Alto, CA, Counseling Psychologists Press). STANTON, W. R., MAxntAstct, P. A., McGES, R. & Su.vA, P. A. (1993) Reasons for smoking or not smoking in early adolescence, Addictive Behaviors, 18, pp. 321-329. TATE, J. C., SCHMITZ, J. M. & STANTON, A. L. (1991) A critical review of the reasons for smoking scale, Journal of Substance Abuse, 3, pp: 441-455. WEISSMAN, M. M., SHOt,oMAras, D., PorrENGER, M., PRUSxoFF, B. A. & LocKE, B. Z. (1977) Assessing depressive symptoms in five psychiatric populations: A validation study, American Journal of Epiderniology, 106, pp. 203-214. WES•r, R. J., HATEy,, P. & BEicHER, M. (1986) Which smokers report most relief from craving when using nicotine chewing gum, Psychopharmacology, 89, pp. 189-191. WEST, R. J. & RussEit, M. A. H. (1985) Pre-absti- nence smoke intake and smoking motivation as predictors of severity of cigarette withdrawal symp- toms, Psychopharmacology, 87, pp. 334-336. APPENDIX I Modified smoking motives questionnaire ATM 1. I've found a cigarette in my mouth without re- calling putting it there. 2. I light up a cigarette without realizing I still have one burning in the ashtray. 3. I find myself smoking without remembering lighting up. SED 4. I smoke more when I am worried about some- thing. 5. Smoking calms me down when I fell tense. 6. I light up a cigarette when I feel angry about something. SOC 7. It is easier to talk and get on with other people when smoking. 8. Whi1e smoking I feel more confident with other people. 9. I smoke much more when I am with other people. IND 10. I want to smoke most when I am comfortable and relaxed. 11. I like a cigarette best when I am having a quiet rest. 12. I usually only smoke when I can really sit back and enjoy it. ADD 13. When I have run out of cigarettes I find it almost unbearable until I can get them. 14. Without a cigarette I don't know what to do with my hands. 15. I get a real gnawing hunger to smoke when I haven't smoked for a while. I

I I 330 ,7ames C. Tate, Cynthia S. Pomerleau & Ovide F. Pomerleau SMM STM I 16. I smoke for the pleasure of having something to put in my mouth. 17. Part of the enjoyment of smoking is watching the 19. I like smoking while .I am busy and working hard. 20. I get a definite lift and feel more alert when I smoke as I blow it out. 18. Part of the enjoyment of smoking comes from the 21. smoking. I smoke more when I am rushed and have lots to I I I I I I I I I I I I I I steps I take to light up. do.