- Email: [email protected]
Persistent effects of a message counter-marketing light cigarettes
Addictive Behaviors 26 (2001) 447 ± 452
Brief report
Persistent effects of a message counter-marketing light cigarettes Results of a randomized controlled trial Lynn T. Kozlowskia,*, Ray Palmera,1, Michele M. Stinea, Andrew A. Strassera, Berwood A. Yostb a
Department of Biobehavioral Health, The Pennsylvania State University, 315 HHD East, University Park, PA 16802, USA b Center for Opinion Research, Millersville University, Millersville, PA, USA
Abstract In a randomized, controlled trial, a national sample of smokers of Light cigarettes heard by telephone a ``radio message'' counter-marketing Light cigarettes. This message caused immediate changes in beliefs. Follow-up telephone interviews were done about 7 months later. The Message Group (N = 181) was more likely than the Control Group (N = 85) to report that (a) one Light equaled one Regular in tar yield to smokers, (b) Lights did not decrease health risks, and (c) they wanted to give up smoking ( P < .05); they did not report greater quitting or intention to quit, or greater knowledge of filter ventilation. Systematic counter-marketing of Lights is recommended. A telephonebased exposure and follow-up procedure could be a good way to study message effects. D 2001 Elsevier Science Ltd. All rights reserved. Keywords: Tobacco smoking; Nicotine; Marketing; Consumer behavior
The cigarette industry has marketed low-tar cigarettes to try to retain health-concerned smokers as smokers (Glantz, Slade, Bero, Hanauer, & Barnes, 1996). Smokers of Light cigarettes report that Light cigarettes are less dangerous than Regular cigarettes (Kozlowski et al., 1998). Unfortunately, there is little evidence that modern Light cigarettes do reduce disease risks compared to Regular cigarettes (see Kozlowski et al., 1998). Previously, we found that about one-third of smokers of Light cigarettes said that they would consider * Corresponding author. Tel.: +1-814-863-7256; fax: +1-814-863-7525. E-mail address: [email protected] (L.T. Kozlowski). 1 Now at the University of Texas Health Science Center, San Antonio, TX. 0306-4603/01/$ ± see front matter D 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 3 0 6 - 4 6 0 3 ( 0 0 ) 0 0 111 - 8
448
L.T. Kozlowski et al. / Addictive Behaviors 26 (2001) 447±452
quitting smoking if they learned that one Light cigarette gave the same amount of tar to smokers as one Regular cigarette (Kozlowski et al., 1998). In an experimental design, we played a simulated 60-s ``radio message'' during a national, random-digit-dialing telephone survey to self-reported daily smokers of Light cigarettes. The report by Kozlowski et al. (1999) provides the complete text of the message and how it was developed and produced. The key elements involved a smoker saying that: (a) his doctor said Lights were not really Light; (b) though Lights felt lighter, they were just as bad for the health; (c) Lightshadhidden vents onthe filter; (d) oneLight could equal oneRegular intar tothe smoker; and (e) the listener should think about quitting smoking. The results showed that the message increased the likelihood of smokers reporting that Light cigarettes were dangerous to health and that Lights did not decrease tar exposures. Smokers who heard the message were more likely to say that one Light cigarette gave smokers as much tar as one Regular cigarette. They also reported an increased desire to quit smoking and a greater intention to quit smoking. Since smokers were being asked for their reactions immediately after hearing the message, the possibility of so-called ``demand characteristics'' cannot be ruled out (Shimp, Hyatt, & Snyder, 1991). To assess how important this information is to the participants, we conducted a follow-up interview to see if there were any persistent effects of hearing the message. Research on attitude change suggests that it is mainly information that is personally relevant that will be retained (Cook & Flay, 1978; Eagley & Chaiken, 1993; Petty & Krosnick, 1995). If the information is not important to these smokers, there would likely be negligible, if any, residual effects from hearing a brief message twice over the phone. 1. Methods The detailed methods for the exposure to the message can be found in Kozlowski et al. (1999). In brief, random-digit-dialing, computer-assisted telephone interviews were done, screening to find self-reported daily smokers of Light cigarettes. In the follow-up, the budget did not permit elaborate tracking procedures. Basically, if the respondent was not to be found at the prior phone number, we went no further. The follow-up rate was equivalent for the two groups: 181 of 401 (45%) in the Message Group and 85 of 167 (51%) in the No Message Control Group ( P = .46, two-tailed). An additional 13 respondents were not included in the follow-up sample because they were considered unreliable and possibly not the same person talked to at the first interview (they reported different birth years, races, or very different years of education). Key questions are indicated in Table 1. Chi-square analyses, t -tests, and multiple regression analyses were used as appropriate to analyze the results. Linear probability regression models were preferred to logistic models, because the dependent proportions were generally between 0.2 and 0.8. 2. Results The second interview took place 7 months on average (S.D. = 0.5, range = 6±9.5 months) after the first interview. There were no differences in sex, age, or years of education between
L.T. Kozlowski et al. / Addictive Behaviors 26 (2001) 447±452
449
Table 1 Responses (%) to key questions by Message and No Message Control Groups Key questions asked
Message (N = 181)
Control (N = 85)
1.
18 (12,24)
31 * (21,41)
27 (21,34)
40 * (30,50)
30 (23,37) 33 (26,40)
47 * (36,58) 52 * (41,63)
49 (42,56) 67 (60,74)
26 * (17,35) 59 (49,70)
35 (27,43)
47 * * (35,59)
82 (76,88)
71 * (61,81)
75 (68,82)
73 (63,83)
2. 3.
4. 5. 6.
7.
8.
Compared to Regular cigarettes, would smoking Light cigarettes increase, decrease, or have no effect on your risk of having health problems? (% answering decrease) 1A. Those who said ``don't know'' or ``no effect'' were asked, ``If, deep down, they thought there was a chance, even a slim one, that Light cigarettes reduced the risk of having health problems?'' ``Some cigarettes are less risky than others.'' (% agree) Compared to Regular cigarettes, do you think smoking Light cigarettes increases, decreases, or has no effect on your daily intake of tar? (% answering decrease shown) How many Light cigarettes equal one Regular in tar? (% saying 1) Have you seen or heard that one or more rings of small holes are on the filters of some cigarettes? (% yes shown) What is the possibility that you will be smoking 5 years from now? (definitely, probably, probably not, definitely not; % probably, definitely smoking shown) n = 153, 68 How much do you want to give up smoking in the next 6 months? (not at all, only a little, somewhat, very much; % greater than or equal to only a little) n = 164, 73 How likely is it that you will try to give up smoking in the next 6 months? (scored as in 7 above) n = 161, 73 Confidence intervals (95%) are shown below percentage. * P .05. ** P = .10.
the two groups. Overall, 40% were female, average age was 40 (S.D. = 14), average years of education was 13 (S.D. = 2). Younger people were somewhat less likely to be interviewed the second time: mean age at Interview 1 was 37; just 7 months later at Interview 2, mean age was 40 ( P = .005). 2.1. Tar category of usual brand At Interview 1, everyone smoked Lights. Overall, 83% still smoked Light cigarettes, 9% smoked Ultra-lights, and 8% smoked Regulars. There was no statistically significant difference between the two groups.
450
L.T. Kozlowski et al. / Addictive Behaviors 26 (2001) 447±452
2.2. No effects of quitting Eighty-six percent of each group were still daily smokers. 2.3. Effects on desire to quit Those who heard the message still report greater desire to quit in the next 6 months ( P < .05) and show a trend ( P = .10) toward greater perceived likelihood of quitting within the next 5 years (see Table 1). 2.4. Effects on knowledge about Lights The Message Group was more likely to report that (a) one Light equaled one Regular in tar to the smoker, (b) Lights were not less risky than Regular cigarettes, and (c) Lights did not decrease health problems or tar exposure (see Table 1). Question 1A in Table 1 shows the combination of the results in Question 1 with a probing question that asked if ``deep down'' the respondent thought there was still a chance that Light reduced risks. Group differences are maintained with this comparison. 2.5. Knowledge of filter vents There is no difference in knowledge of filter vents between the Message and Control Group (see Table 1). 3. Multivariate effects In regression analyses controlling for age, sex, and level of education, having heard the message was a significant predictor of saying that Light cigarettes were less likely to cause health problems (b = 0.216, P=.03) and that some cigarettes are less risky than others (b = 0.173, P=.006). Similarly, when controlling for the same variables, being in the message group also predicted reporting that one Light cigarette was equal to one Regular cigarette (b = 0.234, P=.0003). Hearing the message was not a significant predictor of intention to quit, however. The number of cigarettes smoked per day significantly predicted how strongly the subject believed he or she would not be smoking in 5 years. 4. Discussion We found that this anti-Light message had effects on knowledge about Light cigarettes and desire to quit smoking over 6±9 months. Although ``demand characteristics'' could explain these results immediately following the playing of the message, it is unlikely that ``demand'' can explain the effects several months later. The most likely mechanism for the long-term effects is that smokers found the information very important to them (Petty & Krosnick, 1995).
L.T. Kozlowski et al. / Addictive Behaviors 26 (2001) 447±452
451
Although there were no effects on quitting smoking or attempts to quit smoking, this is not surprising. This message is an element of a smoking-cessation intervention, but it is not itself a complete-cessation intervention. Though our sample is relatively small, this study has several strengths. The sample derives from a national, random sample of daily smokers of Light cigarettes. We were also able to use a true experimental design. The follow-up rate of about 50% did not differ between experimental and control conditions. There were small, but significant differences in age between the originally interviewed sample and those in the follow-up interviews. Younger people were slightly less likely to be re-interviewed. We believe this small effect is probably due to a higher rate of mobility among younger people. The probing question about ``deep down'' beliefs about Lights reducing the risk of health problems indicates that 40% of the Control respondents harbor beliefs about Lights being somewhat better for health. The perceived ``chance'' for improved health may be sufficient to decrease motivation to quit smoking. Future research should explore the use of telephone surveys to test the impact of messages. Even televised messages can have semantic content that could be tested by phone surveys. In our experience, focus groups can have negativity biases (cf. Kanouse & Hanson, 1972), wherein negative views of one member of the group can be more influential on the group than are the positive views of one member. The individual telephone interviews are not biased by such group effects. Other anti-Light messages need to be developed and tested. Our message focused on Lights not being better than Regulars and on ``one Light equals one Regular.'' The decision was made to de-emphasize a lesson about filter vents. Though there were immediate effects of the message on knowledge of vents, these effects did not persist after 7 months. Long-term knowledge of vents might be achieved with other messages. Though this study provides a kind of simulation of the effects of a counter-marketing campaign, a campaign would involve interlocking messages and much greater repetition.
Acknowledgments This study was supported by subcontract number ATPM/CDC TS257-13/13 under a cooperative agreement between the Association of Teachers of Preventive Medicine and the Centers for Disease Control and Prevention. Christine Sweeney and Janine Pillitteri offered assistance in data management and preliminary work on this project.
References Cook, T. D., & Flay, B. R. (1978). The persistence of experimentally induced attitude change. In: L. Berkowitz (Ed.), Advances in experimental social psychology ( pp. 1 ± 57). New York, NY: Academic Press. Eagley, A., & Chaiken, S. (1993). The psychology of attitudes. New York: Harcourt Brace College. Glantz, S. A., Slade, J., Bero, L. A., Hanauer, P., & Barnes, D. E. (1996). The cigarette papers. Berkeley, CA: Univ. of California Press.
452
L.T. Kozlowski et al. / Addictive Behaviors 26 (2001) 447±452
Kanouse, D. E., & Hanson, R. (1972). Negativity in evaluations. In: E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E. Nisbett, S. Valins, & B. Weiner (Eds.), Attribution: perceiving the causes of behavior ( pp. 47 ± 63). Morristown, NJ: General Learning Press. Kozlowski, L. T., Goldberg, M. E., Sweeney, C. T., Palmer, R. F., Pillitteri, J. L., Yost, B. A., White, E. L., & Stine, M. M. (1999). Smoker reactions to a ``radio message'' that Light cigarettes are as dangerous as Regular cigarettes. Nicotine and Tobacco Research, 1, 67 ± 76. Kozlowski, L. T., Goldberg, M. E., Yost, B. A., White, E. L., Sweeney, C. S., & Pillitteri, J. L. (1998). Smokers' misperceptions of light and ultra-light cigarettes may keep them smoking. American Journal of Preventive Medicine, 15 (1), 9 ± 16. Petty, R. E., & Krosnick, J. A. (Eds.). (1995). Attitude strength: antecedents and consequences. Mahwah, NJ: Lawrence Erlbaum Associates. Shimp, T. A., Hyatt, E. M., & Snyder, D. J. (1991). A critical appraisal of demand artifacts in consumer research. Journal of Consumer Research, 18, 273 ± 283.
Brief report
Persistent effects of a message counter-marketing light cigarettes Results of a randomized controlled trial Lynn T. Kozlowskia,*, Ray Palmera,1, Michele M. Stinea, Andrew A. Strassera, Berwood A. Yostb a
Department of Biobehavioral Health, The Pennsylvania State University, 315 HHD East, University Park, PA 16802, USA b Center for Opinion Research, Millersville University, Millersville, PA, USA
Abstract In a randomized, controlled trial, a national sample of smokers of Light cigarettes heard by telephone a ``radio message'' counter-marketing Light cigarettes. This message caused immediate changes in beliefs. Follow-up telephone interviews were done about 7 months later. The Message Group (N = 181) was more likely than the Control Group (N = 85) to report that (a) one Light equaled one Regular in tar yield to smokers, (b) Lights did not decrease health risks, and (c) they wanted to give up smoking ( P < .05); they did not report greater quitting or intention to quit, or greater knowledge of filter ventilation. Systematic counter-marketing of Lights is recommended. A telephonebased exposure and follow-up procedure could be a good way to study message effects. D 2001 Elsevier Science Ltd. All rights reserved. Keywords: Tobacco smoking; Nicotine; Marketing; Consumer behavior
The cigarette industry has marketed low-tar cigarettes to try to retain health-concerned smokers as smokers (Glantz, Slade, Bero, Hanauer, & Barnes, 1996). Smokers of Light cigarettes report that Light cigarettes are less dangerous than Regular cigarettes (Kozlowski et al., 1998). Unfortunately, there is little evidence that modern Light cigarettes do reduce disease risks compared to Regular cigarettes (see Kozlowski et al., 1998). Previously, we found that about one-third of smokers of Light cigarettes said that they would consider * Corresponding author. Tel.: +1-814-863-7256; fax: +1-814-863-7525. E-mail address: [email protected] (L.T. Kozlowski). 1 Now at the University of Texas Health Science Center, San Antonio, TX. 0306-4603/01/$ ± see front matter D 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 3 0 6 - 4 6 0 3 ( 0 0 ) 0 0 111 - 8
448
L.T. Kozlowski et al. / Addictive Behaviors 26 (2001) 447±452
quitting smoking if they learned that one Light cigarette gave the same amount of tar to smokers as one Regular cigarette (Kozlowski et al., 1998). In an experimental design, we played a simulated 60-s ``radio message'' during a national, random-digit-dialing telephone survey to self-reported daily smokers of Light cigarettes. The report by Kozlowski et al. (1999) provides the complete text of the message and how it was developed and produced. The key elements involved a smoker saying that: (a) his doctor said Lights were not really Light; (b) though Lights felt lighter, they were just as bad for the health; (c) Lightshadhidden vents onthe filter; (d) oneLight could equal oneRegular intar tothe smoker; and (e) the listener should think about quitting smoking. The results showed that the message increased the likelihood of smokers reporting that Light cigarettes were dangerous to health and that Lights did not decrease tar exposures. Smokers who heard the message were more likely to say that one Light cigarette gave smokers as much tar as one Regular cigarette. They also reported an increased desire to quit smoking and a greater intention to quit smoking. Since smokers were being asked for their reactions immediately after hearing the message, the possibility of so-called ``demand characteristics'' cannot be ruled out (Shimp, Hyatt, & Snyder, 1991). To assess how important this information is to the participants, we conducted a follow-up interview to see if there were any persistent effects of hearing the message. Research on attitude change suggests that it is mainly information that is personally relevant that will be retained (Cook & Flay, 1978; Eagley & Chaiken, 1993; Petty & Krosnick, 1995). If the information is not important to these smokers, there would likely be negligible, if any, residual effects from hearing a brief message twice over the phone. 1. Methods The detailed methods for the exposure to the message can be found in Kozlowski et al. (1999). In brief, random-digit-dialing, computer-assisted telephone interviews were done, screening to find self-reported daily smokers of Light cigarettes. In the follow-up, the budget did not permit elaborate tracking procedures. Basically, if the respondent was not to be found at the prior phone number, we went no further. The follow-up rate was equivalent for the two groups: 181 of 401 (45%) in the Message Group and 85 of 167 (51%) in the No Message Control Group ( P = .46, two-tailed). An additional 13 respondents were not included in the follow-up sample because they were considered unreliable and possibly not the same person talked to at the first interview (they reported different birth years, races, or very different years of education). Key questions are indicated in Table 1. Chi-square analyses, t -tests, and multiple regression analyses were used as appropriate to analyze the results. Linear probability regression models were preferred to logistic models, because the dependent proportions were generally between 0.2 and 0.8. 2. Results The second interview took place 7 months on average (S.D. = 0.5, range = 6±9.5 months) after the first interview. There were no differences in sex, age, or years of education between
L.T. Kozlowski et al. / Addictive Behaviors 26 (2001) 447±452
449
Table 1 Responses (%) to key questions by Message and No Message Control Groups Key questions asked
Message (N = 181)
Control (N = 85)
1.
18 (12,24)
31 * (21,41)
27 (21,34)
40 * (30,50)
30 (23,37) 33 (26,40)
47 * (36,58) 52 * (41,63)
49 (42,56) 67 (60,74)
26 * (17,35) 59 (49,70)
35 (27,43)
47 * * (35,59)
82 (76,88)
71 * (61,81)
75 (68,82)
73 (63,83)
2. 3.
4. 5. 6.
7.
8.
Compared to Regular cigarettes, would smoking Light cigarettes increase, decrease, or have no effect on your risk of having health problems? (% answering decrease) 1A. Those who said ``don't know'' or ``no effect'' were asked, ``If, deep down, they thought there was a chance, even a slim one, that Light cigarettes reduced the risk of having health problems?'' ``Some cigarettes are less risky than others.'' (% agree) Compared to Regular cigarettes, do you think smoking Light cigarettes increases, decreases, or has no effect on your daily intake of tar? (% answering decrease shown) How many Light cigarettes equal one Regular in tar? (% saying 1) Have you seen or heard that one or more rings of small holes are on the filters of some cigarettes? (% yes shown) What is the possibility that you will be smoking 5 years from now? (definitely, probably, probably not, definitely not; % probably, definitely smoking shown) n = 153, 68 How much do you want to give up smoking in the next 6 months? (not at all, only a little, somewhat, very much; % greater than or equal to only a little) n = 164, 73 How likely is it that you will try to give up smoking in the next 6 months? (scored as in 7 above) n = 161, 73 Confidence intervals (95%) are shown below percentage. * P .05. ** P = .10.
the two groups. Overall, 40% were female, average age was 40 (S.D. = 14), average years of education was 13 (S.D. = 2). Younger people were somewhat less likely to be interviewed the second time: mean age at Interview 1 was 37; just 7 months later at Interview 2, mean age was 40 ( P = .005). 2.1. Tar category of usual brand At Interview 1, everyone smoked Lights. Overall, 83% still smoked Light cigarettes, 9% smoked Ultra-lights, and 8% smoked Regulars. There was no statistically significant difference between the two groups.
450
L.T. Kozlowski et al. / Addictive Behaviors 26 (2001) 447±452
2.2. No effects of quitting Eighty-six percent of each group were still daily smokers. 2.3. Effects on desire to quit Those who heard the message still report greater desire to quit in the next 6 months ( P < .05) and show a trend ( P = .10) toward greater perceived likelihood of quitting within the next 5 years (see Table 1). 2.4. Effects on knowledge about Lights The Message Group was more likely to report that (a) one Light equaled one Regular in tar to the smoker, (b) Lights were not less risky than Regular cigarettes, and (c) Lights did not decrease health problems or tar exposure (see Table 1). Question 1A in Table 1 shows the combination of the results in Question 1 with a probing question that asked if ``deep down'' the respondent thought there was still a chance that Light reduced risks. Group differences are maintained with this comparison. 2.5. Knowledge of filter vents There is no difference in knowledge of filter vents between the Message and Control Group (see Table 1). 3. Multivariate effects In regression analyses controlling for age, sex, and level of education, having heard the message was a significant predictor of saying that Light cigarettes were less likely to cause health problems (b = 0.216, P=.03) and that some cigarettes are less risky than others (b = 0.173, P=.006). Similarly, when controlling for the same variables, being in the message group also predicted reporting that one Light cigarette was equal to one Regular cigarette (b = 0.234, P=.0003). Hearing the message was not a significant predictor of intention to quit, however. The number of cigarettes smoked per day significantly predicted how strongly the subject believed he or she would not be smoking in 5 years. 4. Discussion We found that this anti-Light message had effects on knowledge about Light cigarettes and desire to quit smoking over 6±9 months. Although ``demand characteristics'' could explain these results immediately following the playing of the message, it is unlikely that ``demand'' can explain the effects several months later. The most likely mechanism for the long-term effects is that smokers found the information very important to them (Petty & Krosnick, 1995).
L.T. Kozlowski et al. / Addictive Behaviors 26 (2001) 447±452
451
Although there were no effects on quitting smoking or attempts to quit smoking, this is not surprising. This message is an element of a smoking-cessation intervention, but it is not itself a complete-cessation intervention. Though our sample is relatively small, this study has several strengths. The sample derives from a national, random sample of daily smokers of Light cigarettes. We were also able to use a true experimental design. The follow-up rate of about 50% did not differ between experimental and control conditions. There were small, but significant differences in age between the originally interviewed sample and those in the follow-up interviews. Younger people were slightly less likely to be re-interviewed. We believe this small effect is probably due to a higher rate of mobility among younger people. The probing question about ``deep down'' beliefs about Lights reducing the risk of health problems indicates that 40% of the Control respondents harbor beliefs about Lights being somewhat better for health. The perceived ``chance'' for improved health may be sufficient to decrease motivation to quit smoking. Future research should explore the use of telephone surveys to test the impact of messages. Even televised messages can have semantic content that could be tested by phone surveys. In our experience, focus groups can have negativity biases (cf. Kanouse & Hanson, 1972), wherein negative views of one member of the group can be more influential on the group than are the positive views of one member. The individual telephone interviews are not biased by such group effects. Other anti-Light messages need to be developed and tested. Our message focused on Lights not being better than Regulars and on ``one Light equals one Regular.'' The decision was made to de-emphasize a lesson about filter vents. Though there were immediate effects of the message on knowledge of vents, these effects did not persist after 7 months. Long-term knowledge of vents might be achieved with other messages. Though this study provides a kind of simulation of the effects of a counter-marketing campaign, a campaign would involve interlocking messages and much greater repetition.
Acknowledgments This study was supported by subcontract number ATPM/CDC TS257-13/13 under a cooperative agreement between the Association of Teachers of Preventive Medicine and the Centers for Disease Control and Prevention. Christine Sweeney and Janine Pillitteri offered assistance in data management and preliminary work on this project.
References Cook, T. D., & Flay, B. R. (1978). The persistence of experimentally induced attitude change. In: L. Berkowitz (Ed.), Advances in experimental social psychology ( pp. 1 ± 57). New York, NY: Academic Press. Eagley, A., & Chaiken, S. (1993). The psychology of attitudes. New York: Harcourt Brace College. Glantz, S. A., Slade, J., Bero, L. A., Hanauer, P., & Barnes, D. E. (1996). The cigarette papers. Berkeley, CA: Univ. of California Press.
452
L.T. Kozlowski et al. / Addictive Behaviors 26 (2001) 447±452
Kanouse, D. E., & Hanson, R. (1972). Negativity in evaluations. In: E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E. Nisbett, S. Valins, & B. Weiner (Eds.), Attribution: perceiving the causes of behavior ( pp. 47 ± 63). Morristown, NJ: General Learning Press. Kozlowski, L. T., Goldberg, M. E., Sweeney, C. T., Palmer, R. F., Pillitteri, J. L., Yost, B. A., White, E. L., & Stine, M. M. (1999). Smoker reactions to a ``radio message'' that Light cigarettes are as dangerous as Regular cigarettes. Nicotine and Tobacco Research, 1, 67 ± 76. Kozlowski, L. T., Goldberg, M. E., Yost, B. A., White, E. L., Sweeney, C. S., & Pillitteri, J. L. (1998). Smokers' misperceptions of light and ultra-light cigarettes may keep them smoking. American Journal of Preventive Medicine, 15 (1), 9 ± 16. Petty, R. E., & Krosnick, J. A. (Eds.). (1995). Attitude strength: antecedents and consequences. Mahwah, NJ: Lawrence Erlbaum Associates. Shimp, T. A., Hyatt, E. M., & Snyder, D. J. (1991). A critical appraisal of demand artifacts in consumer research. Journal of Consumer Research, 18, 273 ± 283.