1000 Words Rogerian Essay

The Journal of Social Psychology, 1961, 54, 235-248.

CIGARETTE SMOKING AND ATTITUDE TOWARD THE ETIOLOGY OF LUNG CANCER*^

Norristoiun State Hospital, Norristoivn, Pennsylvania

M . POWELL LAWTON AND ALFRED E . GOLDMAN

A. CIGARETTE SMOKING

Cigarette smoking is a widespread practice with compelling psychological and physiological motives, and is markedly resistant to alteration. Recently serious suspicion has been cast upon cigarette smoking as a causative factor in producing lung cancer, with wide variation in the interpretation given to the laboratory and statistical data pertaining to it. This controversy and the attitudes generated by it offer a situation in which what is often a strong intellectual conviction conflicts directly with an overdetermined, highly stable habit pattern. Specifically, tbis study sought to investigate the degree to which involvement in lung cancer research and the opinions about lung cancer causation associated with such research are related to smoking practice. Access to first-hand information about the controversy differs widely; those who are closest to the data are those scientists who are involved in lung cancer research. It is these cancer scientists, many of whom have devoted their entire professional careers to this research area, who have the greatest opportunity to base their opinions regarding lung cancer causation on the evidence itself rather than upon secondhand reports. Also, involvement in lung cancer-smoking research makes it more likely that an opinion to which the researcher is already committed is less subject to modification. On the other hand, those who are not so intimately acquainted with these data may be more likely to base their opinions regarding lung cancer etiology on other than direct empirical evidence.

These considerations generated the following hypotheses: (1) Involve- ment in lung cancer research determines the distribution of opinions regarding cigarette smoking as a cause of lung cancer. It was expected that cancer scien- tists would be more likely to feel that smoking is a causative factor than a group which is not involved in lung cancer research. (2) Opinion regarding the etiology of lung cancer is related to smoking practice. It was expected

* Received in the Editorial Office on July 20, 1959. 1 This paper was presented on August 29, 1958, at the meetings of the American

Psychological Association in Washington, D. C.

235

2 3 6 JOURNAL OF SOCIAL PSYCHOLOGY

that those who feel that cigarette smoking is a cause of lung cancer would smoke less than those who do not feel that smoking is a cause of lung cancer. (3) Opinion about the lung cancer-smoking controversy is consonant with modification of smoking practice. It was expected that opinion about causa- tion would be related to satisfaction with smoking pattern and to change in smoking habit.

B. SUBJECTS

Degree of involvement in lung cancer-smoking research was represented ordinally by the choice of two groups:

Lung cancer scientists. On November 7-8, 1957, a group of 90 interna- tionally renowned experts in lung cancer research were invited to a confer- ence on lung cancer at Virginia Beach, Va., by the American Cancer Society. A mail questionnaire regarding smoking habits and opinion regarding lung cancer causation was sent to 85 participants in this conference excluding only representatives of the American Cancer Society and of the Tobacco Industries Research Committee. This group of scientists was chosen by the American Cancer Society without regard to the individual's opinion about lung cancer causation. It included approximately equal numbers of individuals with the M.D. and with other doctoral degrees. While many of the scientists had done research specifically on the role of tobacco in lung cancer causation, others were concerned with different etiological propositions while a few were mainly concerned with the chemistry of tobacco. Pathologists, chemists, epi- demiologists, physiologists, radiologists, pharmacologists and psychologists were among tbose included, while clinical practitioners were, in the main, absent.

Experimental psychologists. A control sample was sought which would equate some of the factors which are known to be associated with differential smoking practice, such as age, sex, and occupational level (Haenzel, Shimkin, and Miller, 1956). Experimental psychologists were chosen as a readily available group of comparable education and scientific eminence. The psy- chologists were individually matched for sex and age with the lung cancer specialists. A pool of experimental psychologists' names was randomly drawn by selecting every fifth name from among the fellows of the Division of Experimental Psychology of the American Psychological Association. In order to obtain sufficient female subjects who also could be matched in age it was necessary to use some associates of the Division.

A questionnaire was sent to members of the two groups, with a second mailing sent three weeks later in order to increase number of subjects.

M. POWELL LAWTON AND ALFRED E. GOLDMAN 2 3 7

Of the 85 lung cancer scientists polled, 90.6 per cent responded, all but four of them to the first mailing. Of the 72 psychologists originally polled, 84.7 per cent responded. In order to increase the number of matched pairs, a second sample of 20 psychologists was polled, of which 17 responded. Al- though possible nonrespondent bias may have been introduced, the fact that only 11 psychologists did not respond to the first mailing makes this unlikely. Finally, then, 72 pairs were obtained where a match on both age and sex was possible. Seven of the pairs were female. The mean age of the cancer scientists was 45.4, the mean age of the psychologists 43.8, the difference being not significant. None of the psychologists indicated any professional contact with lung cancer beyond acquaintance with mass communication ma- terial on the lung cancer-smoking controversy.

C. PROCEDURE

The questionnaire requested the following: 1. Judgment on a five point scale ranging from extreme agreement to

extreme disagreement of the degree to which evidence suggests that smoking causes lung cancer.

2. Extent of agreement, based on unsupported belief or "hunch," with the statement "Smoking is a cause of lung cancer," rated on a six point Likert scale.

3. An estimate of daily tobacco consumption. 4. An estimate of daily tobacco consumption prior to any change in amount

which may have occurred during the past five years. 5. Indication of any unsuccessful attempt to stop smoking. 6. A checklist of statements which best express the feeling of current cig-

arette smokers about their own smoking; for example, "I would like to give up smoking but doubt that I can," or "The pleasure of smoking is worth the risk of disease." Although information on pipe and cigar smoking was obtained, these data were not treated inasmuch as cigarette smoking was the major focus of interest.

D. RESULTS

In accordance with Hypothesis I, it was expected that more cancer scientists than psychologists would hold opinions which causally link cigarette smoking and lung cancer. Table 1 presents the distribution of 66 matched pairs of subjects who expressed opinions regarding causation.^ While 83.3 per cent

2 In the other six pairs, one of each pair would not express an opinion and was thus eliminated from the analysis.

238 JOURNAL OF SOCIAL PSYCHOLOGY

TABLE 1 DISTRIBUTION OF OPINIONS REGARDING LUNG CANCER CAUSATION EVIDENCE

Frequency Psychologists Cancer scientists

Opinion (n = 66) (n = 66)

Is a cause 4 13 Probably a cause 38 42 Evidence equivocal 22 8 Probably not a cause 1 3 Not a cause 1 0

P<.01

of the cancer scientists thought that cigarette smoking is one of the causes of lung cancer, only 63.6 per cent of the psychologists expressed this opinion. The difiference between groups is most striking in the number of subjects expressing this relationship without qualification: 13 cancer scientists as op- posed to 4 psychologists. These two distributions of opinions differ from each other significantly (/> < .01 — McNemar two-sample test). When the necessity of basing opinion on evidence is removed by phrasing the question in terms of "hunch" or "feeling" (Table 2 ) , the two samples still differ

TABLE 2 Distribution of unsubstantiated beliefs about lung cancer causation

("smoking is a cause of lung cancer")

Opinion

Agree strongly Agree moderately Agree slightly Disagree slightly Disagree moderately Disagree strongly

P<.OS

significantly (/> •< .05 — Sign T e s t ) . However, with greater freedom to express their unsupported beliefs psychologists more closely approximate the distribution of beliefs expressed by the cancer scientists. In further analyses relating unsupported belief in causation to other experimental variables, re- sults essentially similar to those involving evidence were found. Therefore, this variable is excluded from further discussion.

Table 1 presents all of the opinions expressed by the total group of subjects, smokers as well as nonsmokers. In order to ascertain whether this difference in opinion between cancer specialists and psychologists might be due to dif- ferences in individual cigarette smoking between the two groups, two sub-

Psychologists (n = 64)

11 23 20

7 1 2

Frequency Cancer scientists

(n = 64) 23 24

7 3 3 4

M. POWELL LAWTON AND ALFRED E. GOLDMAN 2 3 9

samples of subjects were drawn, one of psychologists and one of cancer re- search scientists, all of whom had been habitual smokers within the past five years. These samples, like the total sample, were matched individually for age and sex and consisted of 33 pairs. These matched samples of recent smok- ers were found to differ in opinion: cancer scientists felt more strongly than psychologists that smoking was a cause of lung cancer (/> < .05 — Sign T e s t ) .

Hypothesis II stated that opinion and smoking habit are related. Accord- ingly, it was expected that the cancer scientists would smoke less than psy- chologists.

Table 3 indicates that the amount of cigarette smoking among cancer scientists is significantly less than that for psychologists, with both lower

TABLE 3 DISTRIBUTION OF AMOUNT OF CURRENT CIOARETTE SMOKING AND PAST INCIDENCE OF

CIGARETTE SMOKING

Frequency Psychologists Cancer scientists

(n = 72) (n = 72) p

Current amount smoked None 34 50 Yi pack or less 10 11 ^ QJ 1 pack or less 8 6 More than 1 pack 20 5

Past incidence of smoking Never cigarette smoker 26 39 ^ QJ Cigarette smoker within past 5 years 46 33

incidence (/> < .01 — McNemar's two-sample test) of smokers in the entire sample (N = 72), and lower amount of cigarette smoking (/> < .05 — Sign Test) among the 33 pairs of recent smokers.

To throw light on the question of whether this difference in smoking habit is of recent origin, past incidence of cigarette smoking was compared for the two groups (Table 3 ) . This analysis revealed that the cancer scientists sample had a significantly greater number who have never smoked than the psychologist sample {p < .05 — McNemar's two-sample test).

Further exploration of the relationship between opinion and smoking be- havior was made by analysis of this relationship within each of the groups taken separately. The first entry of Table 4 indicates that incidence of cur- rent cigarette smoking is significantly related to opinion (/> < .02 — chi square) among psychologists. For the cancer scientists Fisher's exact method indicated that the relationship between opinion and incidence of cigarette smoking did not reach statistical significance. However, the interpretative

68 46 46 38 36

4.75 4.83 9.32 2.92 4.69

<.O2 <.O2 <.O1 <.O5 <.O2

70 33 33 21 22

>.O5 >.O5 >.O5 >.O5 >.O5

240 JOURNAL OF SOCIAL PSYCHOLOGY

TABLE 4 RELATIONSHIP BETWEEN CAUSATION OPINION AND CIGARETTB SMOKING BEHAVIOR

VARIABLES

Psychologists Cancer scientists Smoking variable compared with opinion n chi square p n p* Incidence of current cigarette smoking Cessation of smoking Change in amount of cigarette smoking Attempts to quit cigarette smoking Satisfaction with current smoking

* Fisher's exact method.

significance of this finding is vitiated by the extremely low frequency of can- cer scientists who did not express the opinion that smoking is a cause of lung cancer. Only three of the 70 cancer scientists felt that cigarette smoking is not a cause. The meaningfulness of the failure to establish statistically these relationships for the cancer scientists is further limited by the relatively low incidence of smoking in this group. The severely restricted distribution of both variables made it very unlikely that even the most extreme relationships between opinion and habit could prove statistically significant.

Subsequent consideration of these findings led to the analysis of the con- cordance between opinion and behavior in a way which was less affected by the grossly skewed distributions of smoking habit and of attitude, and which made use of more data. In the previous analyses the measure of the relation- ship between opinion and behavior was obtained by dichotomizing habit and opinion. That is, subjects were classified as either smokers or nonsmokers, without consideration of magnitude of smoking habit. Similarly, subjects were classified in terms of whether or not they believed that smoking was a cause of lung cancer. In this further analysis, use was made of .degree of both opinion and behavior, and thus provided a more sensitive measure of the relative concordance of opinion and behavior between the two groups.

Weights -were established for the degree to which a particular combination of attitude and amount of smoking would be expected in terms of the hypoth- esis. These weights were established by the method of paired comparisons, wherein the authors rated which of two relationships between opinion and smoking amount corresponded more with the hypothesis. Since the correla- tion between raters' judgments was high ( r h o = : . 9 3 ) , the mean of the judges' rankings served as weights. Thus, a case in which a subject strongly links cigarette smoking and lung cancer and does not smoke at all would receive a high weight relative to a case in which a subject strongly links cig- arette smoking and lung cancer but smokes heavily.

M. POWELL LAWTON AND ALFRED E. GOLDMAN 2 4 1

Treated in this manner it was found that the concordance between opinion and current smoking behavior for the cancer scientists was significantly higher than for the psychologists (^ =: .02 — Wilcoxon matched pairs signed ranks test).

It will be recalled that a significant relationship was established between opinion and smoking habit for the psychologists and yet failed to be demon- strated for cancer scientists. In this case, with a test which utilized matched pairs, and also magnitude, rather than incidence alone, a significant difference favoring the cancer scientists was found. This test of concordance between opinion and amount of smoking for matched pairs circumvents the problem of grossly skewed distributions obtained in the within groups analysis.

Hypothesis III stated the expectation that opinion regarding the contro- versy would be reflected in some rnodification of smoking habit, or at least in some desire to modify it.

Table 5 indicates that while a higher percentage of cancer scientists have decreased their daily smoking consumption than psychologists, this difference does not meet the 5 per cent criterion for significance.

The difference between the cancer scientists and psychologists in number who have been successful in complete termination of cigarette smoking (Ta- ble 5) again favors the cancer scientists though not at a significant level (/) > .05 — chi square). Similarly, significant differences were not found

TABLE 5 DIFFERENCES BETWEEN GROUPS IN CHANGE IN CIGARETTE SMOKING HABIT, INCIDENCE

OF QUITTING, AND DEGREE OF DISSATISFACTION WITH CURRENT HABIT

Smoking behavior Change in amount smoked/day

Increase < J4 pack No change Decrease < ^ pack Decrease J/2 to 1 pack Decrease > 1 pack

Number quit smoking Attempts to quit cigarette smoking Number expressing dissatisfaction

Psychologists n

46

46 38 38

Frequency

1 34

3 4 4 8 (17%)

13 (34%) 10 (25%)

Cancer scientists n

33

33 20 22

Frequency

1 19 2 6 5

II (33%) 4 (20%) 5 (27%)

P

>.O5

>.O5 >.O5 >.O5

between the groups in number of unsuccessful attempts to quit. The statements which current smokers checked expressing the attitude they

held toward their own habit were classified in terms of satisfaction or dis- satisfaction. Slightly over 25 per cent of each group were dissatisfied, in various degrees, with their own smoking habit, with no significant difference between groups (/> > .05 — McNemar T e s t ) .

2 4 2 JOURNAL OF SOCIAL PSYCHOLOGY

Those who did quit were asked why they did. Classification of their re- sponses into health-related and non-health-related categories did not serve to reveal any reliable differences (/> > .05 — McNemar Test) between the groups with respect to their reason for quitting.

While Table 5 presents differences between the two groups in change in smoking habit, the last four rows of Table 4 consider the concordance he- tween a subject's opinion toward the controversy and modification in his smoking habit.

For psychologists, opinion is significantly related to change in daily cigarette consumption {p < .01 —chi square), and to cessation of smoking (p < .02 — chi square). Opinion is also related significantly to incidence of unsuc- cessful attempts to stop smoking (/> < .05 — chi square), and to dissatisfac- tion with current smoking hahit {p < .02 — chi square).

While the psychologists' data was evaluated hy means of chi square, the cell entries in the distribution of the cancer scientists required use of Fisher's exact method. For the cancer scientists none of the relationships tested reached the .05 level of significance (Table 4 ) .

E. DISCUSSION

The focus of the present study has been the consequence of two interacting forces: a strong intellectual conviction, on the one hand, and a. powerful psychophysiological habit on the other. The choice of lung cancer specialists and psychologists as subjects did provide us with groups which differed in degree of involvement in lung caticer research and which did, as the results indicate, differ in intellectual conviction regarding the lung cancer-smoking controversy.

1. Attitude Toward Controversy

While both of these groups believe that smoking is a cause of lung cancer, the cancer scientists hold this opinion significantly more strongly than the psychologists. Previous studies also indicate preponderant agreement with statements causally relating smoking and lung cancer. A probability sample of Ann Arbor, Michigan, residents (Cannell and MacDonald, 1956) indi- cated that 42 per cent of this group (highly weighted toward the upper end of the educational scale) were willing to accept as "proven" the proposition that smoking may cause lung cancer. Samples of professionals express agree- ment with the causal statement of the lung cancer-smoking relationship to a greater degree. Snegirefl and Lombard (1954, 1955) indicated that 59 per cent of Massachusetts physicians agreed with the statement, "Heavy smoking

M. POWELL LAWTON AND ALFRED E. GOLDMAN 2 4 3

of cigarettes may lead to lung cancer," while 55 per cent of a group of thoracic surgeons, pathologists and cancer research scientists polled by Horn, Cameron, and Kipnis (1955) agreed with the same statement. A poll con- ducted under much less rigorous conditions than the above two studies by the journal Resident Physician (1957) indicated that 71 per cent of their group of 675 residents in various medical specialties "felt that there was a relationship between smoking and cancer" (p. 54).

The two groups which are the most directly concerned professionally with lung cancer express the most consistent agreement that smoking and lung cancer are causally related. Sixty-three per cent of the thoracic surgeons in the Horn, Cameron, and Kipnis study (1955) expressed agreement, while 83 per cent of the cancer scientists in the present investigation agreed to the proposition. Thus the weight of professional opinion recognizes the carcino- genic potential of heavy cigarette smoking, with those professionally more intimate with the clinical and research material expressing positive convictions with greater frequency.

2. Conviction and Smoking Habit In view of this preponderance of medical and scientific opinion suggesting

the guilt of cigarette smoking as a carcinogen, one is led to ask whether differences in intellectual conviction and involvement in lung cancer research were reflected in differences between the psychologists and cancer scientists in smoking behavior. Analysis of incidence and amount of current smoking suggested that it was: cancer scientists smoke less than psychologists.

However, one must not conclude that these are causally related since it appears that this difference in smoking rate existed at least since 1952, ante- dating the recently intensified research activity in this area and the wide dis- semination of research findings on the controversy. This longstanding differ- ence in smoking rate suggests, then, the possibility that a selective process may operate in determining the kind of personality that is more likely to become involved in lung cancer-smoking research. Studies which link smoking behav- ior to various personality traits and also to vocational choice such as those by Lawton and Phillips (1956), McArthur, Waldron, and Dickinson (1958), and Heath (1958), lend credence to this suggestion.

Conceivably the anxiety provided by lung cancer-smoking research might act as a deterrent to involvement in such work among people who already are heavy smokers.

A test of the relationship between each individual's opinion and his smoking behavior proved significant among the sample of psychologists. That is.

244 JOURNAL OF SOCIAL PSYCHOLOGY

psychologists who felt that smoking was a cause of lung cancer, compared with those who did not express this view were apt to smoke less, and to express greater dissatisfactions with their own smoking.

Corresponding tests of the significance of these relationships for the cancer scientists were indeterminate because of the very low incidence of both smok- ing and of non-agreement that smoking was a cause of lung cancer. The speculation that broader representation of smoking habit and opinion would indicate such a relationship was strengthened by the high concordance of indi- vidual smoking behavior and opinion among cancer scientists as compared to psychologists. In this more sensitive test, the relationship between opinion and habit is closer for the cancer scientists than for the psychologists.

This consonance between opinion and behavior has been amply documented in other studies, as indicated in Table 6. Again, despite differences in phras-

TABLE 6 SUMMARY OF STUDIES RELATING AGREEMENT THAT CIGARETTE SMOKING MAY CAUSE

LUNG CANCER TO CURRENT CIGARETTE SMOKING STATUS

Study

Snegireff and Lombard (1954) Physicians

Horn, Cameron, and Kipnis (1955) Physicians

Cannell and MacDonald (1956) General population

Lawton and Goldman Cancer Scientists Psychologists

ing of the question and differences in populations used, all groups tested showed a relationship between the individual's smoking habit and his attitude toward this controversy. In all studies, nonsmoking was more closely asso- ciated with opinion that smoking causes lung cancer.

At first glance, it would seem that cognitive factors do play a role in determining incidence and extent of smoking habit. However, one must also ask the corollary question, does smoking behavior determine attitude toward causation ? A partial answer may derive from consideration of whether modification of smoking behavior is likewise related to attitude toward causa- tion. Even here, it is logically possible to presume that change in smoking habit may precede or form attitude toward the controversy. However, the more logically expected direction would be one whereby intellectual convic- tion regarding risk of lung cancer would produce a successful effort on the

Per Smokers

56

36

28

82 53

cent agreement Nonsmokers

63

65

54

85 81

M. POWELL LAWTON AND ALFRED E. GOLDMAN 2 4 5

part of the cigarette smoker to quit smoking or to decrease the amount of smoking.

3. Opinion and Modification of Habit

If, as is suggested here, opinion that heavy smoking causes lung cancer is given more frequently in samples closer to professional involvement with lung cancer, and if opinion plays a role in determining smoking behavior, then would it not follow that these cancer specialists should show greater decrease in smoking than those not so involved professionally? From the studies sum- marized in Table 7 it appears that this is a tenable proposition. Thirty and

TABLE 7 SUMMARY OF STUDIES GIVING RATES OF CESSATION OF CIGARETTE SMOKING

Study

Haenzel, Shimkin, and Miller (1956)

Hammond and Percy (1958)

Lawton and Goldman Resident Physician (1957)

Horn, Cameron, and Kipnis (1955)

Lawton and Goldman

three-tenths per cent of the cancer scientists in the present study, and 30 per cent of the thoracic surgeons (Horn, Cameron, and Kipnis, 1955) quit smok- ing while other samples, including physicians and psychologists, showed cessa- tion rates within the range of the general population. The slightly higher rates for the psychologists, pathologists, and resident physicians are consistent with the general cessation rates for professional and technical workers (21.6 per cent, Haenzel, Shimkin, and Miller, 1956). However, in the present study, the difference between psychologists and cancer scientists did not reach statistical significance in number who quit, amount of decrease, or number dissatisfied with the habit.

Information concerning the relationship between an individual's opinion toward the controversy and change in his own smoking behavior is available only from the present investigation.

These results revealed that in the case of the psychologists not only was there a difference in current smoking rate between individuals of differing convictions regarding the controversy, but that some alteration, or motivation

Population

Laborers Clerical workers Professional and technical General population (phone

directory) Experimental psychologists Resident physicians General practitioners Pathologists Cancer research scientists Thoracic surgeons Lung cancer research scientists

Per cent cessation

12 15 20

19 17 20 17 18 22 30 30

2 4 6 JOURNAL OF SOCIAL PSYCHOLOGY

toward alteration, of smoking habit had occurred in individuals of differing convictions. Psychologists who felt that smoking was a cause of lung cancer showed greater decrease in smoking rate and more attempts to quit than those psychologists who did not feel this way. That this correspondence hetween opinion and attempt to change did not ohtain with the cancer scientists seems, as in the case of incidence of smoking, due to the very few cases of cancer scientists who ever smoked, and thus could quit.

An inquiry into the role that cognition plays in establishing or altering the smoking habit must especially consider those in this sample, and in others, who believe that cancer is caused by smoking, are in a position to base this opinion on empirical findings, and still continue to smoke. Even if one is impressed with the fact that 30 per cent of all cancer scientists quit smoking it is difficult to ignore the 45 per cent who did not quit and yet believe in the smoking-cancer link. Perhaps even more striking is that, with the excep- tion of the cancer scientists, amongst those who do quit, relatively few give fear of lung cancer as the primary reason. Cancer scientist-quitters gave this as a reason 50 per cent of the time. However, 18 per cent of the thoracic surgeons, 13 per cent of the American Association for Cancer Research mem- bers, and 11 per cent of the pathologists (Horn, Cameron, and Kipnis, 1955) felt that they gave up smoking because of the cancer-smoking relationship. Among those Massachusetts physicians who altered their smoking habit, though not necessarily quitting, 15.1 per cent gave the recent lung cancer- smoking publicity as a reason (Snegireff and Lombard, 1954). In the gen- eral population the number who quit for this reason is even lower—6.3 per cent (Hammond and Percy, 1958).

Thus, at least in absolute terms, among those who believe in the smoking- lung cancer link, fewer quit than would be expected if opinion were the major factor in the continuation of smoking. In view of the resistant smoking be- havior of the cancer scientists and thoracic surgeons who did not quit despite their opinion that heavy smoking causes lung cancer, it is not surprising that smoking rates in the general public continue to be largely unaffected by recent adverse publicity concerning the smoking-cancer controversy (Hammond and Percy, 1958). It also suggests, rather strongly, the operation of stable psycho- physiological motives which may act to counter the role that intellectual con- viction plays in smoking, even in the intellectually oriented sample investi- gated in this study.

It seems likely that the relative appeal of cognitive and emotional factors differs with different groups depending on their education, vocation, socio- cultural background, and other factors. For example, Cannell and MacDon-

M. POWELL LAWTON AND ALFRED E. GOLDMAN 2 4 7

aid (1956) found that as education increased, the greater the disparity be- tween smokers and nonsmokers in number who believed that smoking causes cancer. Educated nonsmokers were more convinced than less educated non- smokers of the carcinogenic potential of smoking. On the other hand, the better educated smokers were more likely to deny that lung cancer is caused by smoking. Tbis suggests that education may serve to cement prior attitudes more firmly or perhaps to provide smokers and nonsmokers alike with more adequate rationalizations for whatever attitudes they already hold.

An educational campaign by health agencies to alert the public to the po- tential health hazard of heavy smoking should consider the nature of its audience and attempt to appeal to those factors operant in the population to which it is directed. This study and others (McArthur, Waldron, and Dickinson, 1958; Lawton and Phillips, 1956; Lilienfeld, 1959) suggest that emotional factors of which the individual may not be aware are primary in motivating smoking in a large segment of the public. For example, it may be more fruitful in terms of changing smoking behavior, to associate character weakness and smoking behavior, rather than advertise the results of experi- ments linking cancer with smoking.

F. SUMMARY AND CONCLUSIONS

Inquiry into the opinions toward the lung cancer-smoking controversy and personal smoking habit was made of matched pairs of cancer scientists and experimental psychologists.

The results, from a mailed questionnaire confirmed, in part, the hypotheses and appeared to justify the following conclusions:

1. On the basis of the evidence available to them, more lung cancer scien- tists than psychologists felt that cigarette smoking is a cause of lung cancer.

2. Lung cancer scientists smoke less than psychologists now, and apparently have always made more moderate use of cigarettes.

3. A direct test of the relationship between opinion toward the controversy and smoking behavior revealed that the smoking habits of psychologists were likely to be related to their opinions regarding causation. Psychologists who believed that cigarette smoking causes lung cancer were less likely to smoke, and if they did smoke, were more likely to decrease their smoking rate, at- tempt to decrease smoking rate, even if unsuccessful, or at least wish that they could decrease, or quit altogether. These relationships remained unclari- fied among cancer scientists because of grossly skewed data.

4. Nevertheless, comparison of the cancer scientists and the psychologists for the relative degree of correspondence between opinion and incidence of

248 JOURNAL OF SOCIAL PSYCHOLOGY

smoking revealed that this concordance was greater in - the cancer scientist group.

5. Involvement in lung cancer-smoking research per se, and the more in- tense opinions that it generates toward the controversy, did little to modify, or even make one desire to modify, the smoking habit.

6. Despite the greater correspondence between opinion and smoking rate for the cancer scientists, there were insignificant differences between the two groups in number who altered their habit. In addition, there were a large number of subjects who believed in causation and yet did not quit smoking. Finally, considering the small number among those who did quit who gave the smoking-lung cancer link as their primary reason for quitting, it was suggested that the role of cognitive factors in smoking was subordinate to psychophysiological ones.

REFERENCES

1. CANNELL, C. F., & MACDONALD, J. C. The impact of health news on attitudes and behavior. Journalism Quarterly, 1956, 33, 315-323.

2. Doctors have changed their smoking habits. Res. Phys., 1957, 3, 50-63. 3. HAENZEL, W . , SHIMKIN, M . B., & MILLER, H . P. Tobacco smoking patterns in

the United States. Public Health Monograph No. 45. U.S. Department of Health, Education and W^elfare, 1956.

4. HAMMOND, E . C , k PERCY, C. Ex-smokers. N.Y. State J. Med., 1958, 58, 2956- 2959.

5. HEATH, C. W . Differences between smokers and nonsmokers. Arch. Intern. Med., 1958, 101, 377-388.

6. HORN, D., CAMERON, C . S., & KIPNIS, D . Survey of medical opinion towards smoking. The American Cancer Society, 1955.

7. LAWTON, M . P., & PHILLIPS, R. W . The relationship between excessive cigarette smoking and psychological tension. Am. J. Med. Set., 1956, 232, 397-402.

8. LILIENFELD, A. M. Emotional and other selected characteristics of cigarette smok- ers and nonsmokers as related to epidemiological studies of lung cancer and other diseases. J. Nat. Cancer Inst., 1959, 22, 259-282.

9. MCARTHUR, C , WALDRON, E., & DICKINSON, J. The psychology of smoking. J. Abn. ©• Soc. Psychol., 1958, 56, 267-275.

10. SNEGIREFF, L. S., & LOMBARD, O. Survey of smoking habits of Massachusetts physicians. Neijj England J. Med., 1954, 250, 1042-1045.

11. . Comparative study of smoking habits of physicians. Nevi England J. Med., 1955, 252, 619-696.

Norristoivn State Hospital Norristown, Pennsylvania