This article was downloaded by: [York University Libraries] On: 22 November 2014, At: 10:02 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
The Journal of General Psychology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/vgen20
Efficacy of Caffeine Versus Expectancy in Altering Caffeine-Related Symptoms a
a
Larry Christensen , Jeff Miller & Donna Johnson
a
a
Department of Psychology , Texas A&M University , USA Published online: 06 Jul 2010.
To cite this article: Larry Christensen , Jeff Miller & Donna Johnson (1991) Efficacy of Caffeine Versus Expectancy in Altering Caffeine-Related Symptoms, The Journal of General Psychology, 118:1, 5-12, DOI: 10.1080/00221309.1991.9711128 To link to this article: http://dx.doi.org/10.1080/00221309.1991.9711128
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.
Downloaded by [York University Libraries] at 10:02 22 November 2014
This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
The Journal of General Psychology, 118(1),5-12
Downloaded by [York University Libraries] at 10:02 22 November 2014
Efficacy of Caffeine Versus Expectancy in Altehng Caffeine-Related Symptoms LARRY CHRISTENSEN JEFF MILLER DONNA JOHNSON Department of Psychology Texas A&M University
ABSTRACT. The study investigated the independent and interactive effects of caffeine and expectancy on caffeine-relatedsymptoms. High- and low-caffeine consumers were randomly assigned to either an expectancy or nonexpectancy instructional set and one of four caffeine doses. Subjects were administered the State-Trait Anxiety Inventory, (Spielberger & Gorsuch, 1970) and a Symptom Questionnaire(Christensen, White, Krietsch, & Steele, 1990) prior to and 45 min followingconsumption of one of the four caffeine doses. An analysis of covariance identified a significant main effect for the State-Trait Anxiety Inventory scores and significant main and interaction effects for four Symptom Questionnaire items. However, when the alpha levels were corrected for the increased probability of Type I error, using the Bonferroni procedure, these effects failed to achieve significance. These results suggest that previous reports of subjective caffeine effects are also suspect because of their failure to control for the increased probability of 5 p e I error.
CAFFEINE,A CENTRAL NERVOUS SYSTEM STIMULANT,is generally assumed to have the effect of reducing drowsiness and fatigue and of producing a clearer flow of thought (Rall, 1980). At high doses, it also produces restlessness, irritability, agitation, and sensory and sleep disturbances (Edelstein, Keaton-Brasted, & Burg, 1984; James & Stirling, 1983). These effects are assumed to be due to caffeine, with little consideration given to the role of subject’s expectancies about the drug’s effect. It has been well documented (Ross & Olson, 1981) that placebos can produce a variety of effects, such as a reduction in clinical pain, increased arousal, reduced anxiety, reduced depression, feelings of increased alertness, Requests for reprints should be sent to Larry Christensen, Department of Psychology, Texas A&M University, College Station, lX 77843. 5
Downloaded by [York University Libraries] at 10:02 22 November 2014
6
The Journal of General Psychology
tension, relaxation, or drowsiness. These placebo effects generally correspond to one’s knowledge or belief about the type of drug being consumed, thus forming a relationship between expectancy and the placebo reaction. Although such a relationship would also seem to exist with caffeine, little research has been conducted on this issue. The few studies that have investigated the expectancy effects of caffeine have induced them through the use of an instructional set. Ammon, Carlson, Froberg, Karlsson, & Levi (1973) found that subjects who were caffeine consumers were much less likely to report fatigue and more likely to report tension and nervousness when told that they were taking caffeine. Christensen, White, Krietsch, & Steele (1990) revealed that subjects who were informed that the gelatin capsule they consumed, ostensibly containing caffeine, should induce a variety of symptoms, perceived an increase in feelings of alertness, sleepiness, and clarity of thought over the perception of subjects who were not given a specific expectancy. Kirsch and Weixel (1988) also documented an expectancy effect with respect to caffeine but found different effects when using a double-blind challenge design versus deceptive instructions (informing subjects they were getting caffeinated coffee when they really received decaffeinated coffee). The studies, although few in number, have consistently revealed that instructions can induce a variety of symptoms commonly associated with the consumption of caffeine. Their findings are consistent with the results of other drug studies (Hughes, Pickens, Spring, & Keenan, 1985; Kirsch & Weixel, 1988), indicating that instructions can control the occurrence of drug effects. However, little information is provided on the interactiveeffect of instructions and caffeine dose. The present study used a balanced placebo design (Marlatt & Rohensow, 1980) to investigate the independent and interactive effects of caffeine and instructionally induced expectancy on caffeine-related symptoms. This design yields a matrix in which subjects are (a) told they will get a drug and receive it, (b) told they will get a drug but receive placebo, (c) told there will be no drug but they do receive a drug, (d) told there will be no drug and they receive none. In the present study, several drug or caffeine intake levels were used to assess the potential independent and interactive effects of instructionally induced expectancies and increasing levels of actual consumption of caffeine.
Method Subjects and Instruments
The subjects were 120 undergraduate students (39 men and 81 women), enrolled in an introductory psychology class at Texas A&M University. When
Downloaded by [York University Libraries] at 10:02 22 November 2014
Christensen, Miller, & Johnson
7
signing up for the experiment, the subjects were informed that they would be participating in a caffeine experiment. Measures of caffeine-related symptoms included the Strate-Trait Anxiety Inventory (STAI; Spielberger & Gorsuch, 1970) and a Symptom Questionnaire (Christensen et al., 1990). The STAI is a 40-item self-report inventory that requires subjects to respond on a 4-point scale ranging from not at all (1) to very much so (4) for the measure of state anxiety and from almost never (1) to almost always (4) for the trait measure of anxiety. The Symptom Questionnaire is a self-report questionnaire that requires subjects to evaluate themselves on 12 symptoms that can be produced by ingestion of caffeine (Gilliland & Andress, 1984). Each item is evaluated on an 11-point scale ranging from not at all (0) to extremely (10). A food record was constructed that required subjects to record all foods, beverages, and medications consumed as well as the amount consumed to provide an assessment of the amount of ad libitum caffeine consumed. The Social Desirability Scale (Crowne & Marlowe, 1964) was administered to subjects to obtain an assessment of their tendency to evaluate themselves in a socially desirable manner. This scale provided an assessment of the extent to which a socially desirable response style would moderate susceptibility to the instructional set. The Social Desirability Scale is a 33-item inventory with a true-false format. The drug used was anhydrous caffeine. Doses (100 mg, 300 mg, and 500 mg) were prepared in soft gelatin capsules. Placebo capsules containing cellulose were prepared in identical-sized soft gelatin capsules. Procedure Subjects read and signed an informed consent form and were instructed to complete the food record sheets for 3 consecutive days, to avoid caffeine for a 24-hr period before experimental participation, and to provide a saliva sample to measure compliance with this request. (These samples were not analyzed,) Subjects then selected a time between 9 A.M. and 4 P.M. during the next week to participate in the caffeine experiment. Subjects stating that they had avoided caffeine during the prior 24-hr period provided a saliva sample. Subjects who did not avoid caffeine were released from the study. Food records were then collected and inspected to obtain an assessment of ad libitum caffeine consumption. While the food records were being inspected, the subjects completed the STAI, the Symptom Questionnaire, and the Social Desirability Scale. In making the assessment of caffeine consumption, an 8-02 cup of coffee was assumed to contain 100 mg of caffeine and a 12-02 cola beverage or an 8-02 glass of tea was assumed to contain 50 mg of caffeine. The caffeine content of medications was determined by consulting a sheet that listed the
Downloaded by [York University Libraries] at 10:02 22 November 2014
8
The Journal of General Psychology
caffeine content of most prescription and over-the-counter medications. By this assessment procedure, 40 subjects were classified as high- (2200 mg/ day; average daily intake, 262.7 mg.; range, 200-550 mg) and 80 as low( .05. The high- and low-caffeine consumers’ scores were then combined and subjected ‘Huberty and Moms (1989) have also revealed that the MANOVA-ANOVA approach commonly used when analyzing multiple outcome variables does not provide the assumed control for Type I error probabilities.
Downloaded by [York University Libraries] at 10:02 22 November 2014
Christensen, Miller, &Johnson
9
to a three-way classification analysis of covariance (ANCOVA), in which the independent variables were expectancy, amount of caffeine consumed, and high or low social desirability. Categorization into high or low social desirability was determined by a median split of the subjects' scores on this scale. An ANCOVA conducted separately on the STAI posttest scores, using the pretest scores as the covariate, revealed a significant difference for the main effect of expectancy, F( 1 , 101) = 4.99, p < .05, but only for the state anxiety scores. An ANCOVA conducted on each of the Symptom Questionnaire posttest item scores, using the pretest scores as the covariate, revealed a significant main effect of expectancy on the items measuring alertness, F(1,91) = 7.26, p < .05, and rapid heart beat, F(1,91) = 5.36, p < .05; a significant main effect of caffeine on the items measuring anxiety, F(3,91) = 2.93, p < .05, and relaxation, F(3, 91) = 2.97, p < .05; a significant Caffeine x Expectancy interaction on the items measuring headache, F(3,91) = 2.81, p < .05, and clear flow of thought, F(3, 91) = 4.19, p C .05; a significant Caffeine x Sociability interaction on the item measuring anxiety F(3, 91) = 2.93, p < .05; and a significant Caffeine x Expectancy x Sociability interaction on the item measuring alertness, F(3, 91) = 2.96, p < .05. Although a number of main and interaction effects were found to be significant on the basis of the multiple univariate ANCOVAs, this procedure enhances the probability of Qpe I error. To control for Qpe I error, a Bonferroni adjustment of the alpha level was incorporated. With this adjustment, an alpha level of .004 is needed to attain significance that results in none of the previously identified main or interaction effects achieving significance.
Discussion The results of the present study identified a number of significant main and interaction effects when each univariate ANCOVA was considered separately. However, when the probability values obtained from these univariate analyses were corrected for the increased probability of T@e I error, none of the effects achieved significance, which resulted in an inability to support the hypothesis that caffeine has an effect on the subjective mood of college students. This failure to find a significant effect is consistent with the bulk of literature suggesting that caffeine has little effect on performance (Sawyer, Julia, & 'hrin, 1982). The results are, however, inconsistent with several recent studies (Griffiths, Bigelow, & Liebson, 1986; Griffiths & Woodson, 1988; Stem et al., 1989) indicating that caffeine in the absence of expectancy produces's variety of subjective effects and results (Christensen et al., 1990; Kirsch & Weixel, 1988; Mitchell, Ross, & Hurst, 1974), suggesting that the expectancy effect may account for a rather significant portion of the effects generally attributed to caffeine.
Downloaded by [York University Libraries] at 10:02 22 November 2014
10
The Journal of General Psychology
The discrepancy existing between these studies and the present one may be that many of these prior studies did not provide adequate control for Type I error because of the use of multiple univariate analysis or the use of MANOVA as a method for controlling Type I error. In the present study and in most of the prior studies (e.g., Griffiths et al., 1986; Griffiths & Woodson, 1988; Stem et al., 1989), multiple outcome measures were incorporated and each was analyzed by univariate analyses to assess the subjective effects of caffeine. This procedure enhanced the probability of falsely rejecting the null hypothesis. In the present study, several main and interaction effects would have been considered significant if a correction had not been included to control for this enhanced probability of Type I error. When a Bonferroni correction of the alpha level was incorporated to control for Type I error probability, none of the previously identified effects were significant. Stem et al., (1989), Griffiths & Woodson (1988), and Griffiths et al., (1986) analyzed each of their multiple outcome measures using ANOVA without such a correction, which increased the probability of falsely rejecting the null hypothesis. Without including a control for the increased probability of Type I error, it is impossible to determine which of the effects identified in these studies are real and which are due to chance. If Type I error was controlled, the results of these studies (Griffiths et al., 1986; Griffiths & Woodson, 1988; Stem et al., 1989) would possibly support the present study, indicating that caffeine produces few subjective effects. The discrepancy between the results of the present study and those of the Christensen et al. (1990) study is probably due to the method of attempting to control Type I error. The Christensen et al. (1990) study made use of MANOVA, a procedure currently being questioned (Huberty & Morris, 1989) as a method for controlling Type I error. In the present study, use of the Bonferroni correction, was probably the more stringent as well as the more appropriate procedure but resulted in the failure to find a significant expectancy effect. It is also possible that the discrepancy between the present study and many prior studies is due to differences in subject samples. The subject sample in the present study consisted of a volunteer group of college students obtained from introductory psychology classes with an average daily caffeine consumption of 129.4 mg. Other studies (e.g., Griffiths et al., 1986; Griffiths & Woodson, 1988) used an older sample recruited from advertisements and consuming much more caffeine (e.g., mean daily consumption, 361 mg, in the Griffiths & Woodson study). Such differences, particularly in daily caffeine consumption, suggest that many prior studies may have obtained a subject population self-selected in terms of their reactivity to caffeine. Such a subsample may be reactive to caffeine and may experience subjective symptoms. The results of the present study suggest that drawing such a conclusion for the general population, particularly for college students, is tenuous.
Downloaded by [York University Libraries] at 10:02 22 November 2014
Christensen, Miller, Br Johnson
1I
It is also possible that the stimulant effects of caffeine are situation specific. Goldstein, Kaizer, & Warren (1969) have observed that caffeine effectively counteracts the decrement in a variety of types of performance caused by fatigue or sleep deprivation but is less effective in enhancing performance under other conditions. This observation suggests that the stimulant effects of caffeine are most apparent and operate most effectively under fatigue or sleep deprivation conditions. If this is true, then assessment of caffeine effects under rested or nonfatigued conditions would be subject to a ceiling effect and generate minimal effects as observed in this and other (Mitchell et al., 1974) studies. This assessment is consistent with the conditions under which the stimulant effect of caffeine is generally desired: in the morning, afternoon, or at other times when fatigue and lethargy are beginning to develop. Perhaps caffeine effects should be investigated under these conditions; at least, a fatigue or sleep deprivation condition should be included in caffeine studies to identify the differential effects that occur when subjects are rested versus fatigued. REFERENCES Ammon, H. P.T., Carlson, L. A., Froberg, J., Karlsson, C. G., & Levi, L. (1973). Effects of coffee and caffeine on sympathoadrenomedullary activity, blood lipids, psychological ratings and performance (Report No. 31). Stockholm, Sweden: Laboratory for Clinical Stress Research, Karolinska Institute. Christensen, L.. White, B., Krietsch, K., & Steele, G. (1990). Expectancy effects in caffeine research. The International Journal of the Addictions. 25, 27-3 1. Crowne, D., & Marlowe, D. (1964). Social desirability scale in The approval motive. New York: Wiley. Edelstein, B. A., Keaton-Brasted, C., & Burg, M. M. (1984). Effects of caffeine withdrawal on nocturnal enuresis, insomnia, and behavior restraints. Journal of Consulting and Clinical Psychology, 52, 857-862. Gilliland, K., & Andress, D. (1984). Ad lib caffeine consumption, symptoms of caffeinism, and academic performance. American Journal of Psychiatry, 138, 5 12-5 14. Goldstein, A., Kaizer, S.,& Warren, R. (1969). Psychotropic effects of caffeine in man. 11. Alertness, psychomotor coordination, and mood. Journal of Pharmacology and Experimental Therapeutics, 150, 146-1 5 1. Griffiths, R. R., Bigelow, G. E., & Liebson, I. A. (1986). Human coffee drinking: Reinforcing and physical dependence producing effects of caffeine. Journal of Pharmacology and Experimental Therapeutics, 239, 4 16-429. Griffiths, R. R., & Woodson, P. P. (1988). Reinforcing effects of caffeine in humans. Journal of Pharmacology and Experimental Therapeutics, 246, 2 1-29. Huberty, C. J., & Moms, J. D. (1989). Multivariate analysis versus multiple univariate analyses. Psychological Bulletin, 105, 302-308. Hughes, J. R., Pickens, R. W., Spring, W., & Keenan, R. M. (1985). Instructions control whether nicotine will serve as a reinforcer. Journal of Pharmacology and Experimental Therapeutics, 235. 106-1 12. James, F. E., & Stirling, K. P. (1983). Caffeine: A survey of some of the known
Downloaded by [York University Libraries] at 10:02 22 November 2014
12
The Journal of General Psychology
and suspected deleterious effects of habitual use. British Journal of Addiction, 78, 25 1-258. Kirsch, I., & Weixel, L. J. (1988). Double-blind versus deceptive administration of a placebo. Behavioral Neuroscience 102, 319-323. Marlatt, G.A., & Rohensow, D. J. (1980). Cognitive processes in alcohol use: Expectancy and the balanced placebo design. In N. K. Mello (Ed.), Advances in substance abuse: Behavioral and biological research. Greenwich, CT:JAI Press. Mitchell, V. E., Ross, S., & Hurst, P. M. (1974). Drugs and placebos: Effects of caffeine on cognitive performance. Psychological Reports, 35, 875-883. Rall, T. W. (1980). Central nervous system stimulants. In G. A. Goodman & A. Gilman (Eds.), The phrmcological basis of therapeutics. (6th ed., pp. 592607). New York: Macmillan. Ross, M., & Olson, J. M. (1981). An expectancy-attribution model of the effects of placebos. Psychological Review, 88, 408-437. Sawyer, D. A., Julia, H. L., & Turin, A. C. (1982). Caffeine and human behavior: Arousal, anxiety, and performance effects. Journal of Behavioral Medicine, 5 , 4 19-439. Spielberger, C. D., & Gorsuch, R. L. (1970). The State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologist Press. Stem, K. N., Chait, L. D., & Johanson, C. E. (1989). Reinforcing and subjective effects of caffeine in normal human volunteers. Psychopharmacology.98, 8 1-89. Received August 20, 1990
The Journal of General Psychology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/vgen20
Efficacy of Caffeine Versus Expectancy in Altering Caffeine-Related Symptoms a
a
Larry Christensen , Jeff Miller & Donna Johnson
a
a
Department of Psychology , Texas A&M University , USA Published online: 06 Jul 2010.
To cite this article: Larry Christensen , Jeff Miller & Donna Johnson (1991) Efficacy of Caffeine Versus Expectancy in Altering Caffeine-Related Symptoms, The Journal of General Psychology, 118:1, 5-12, DOI: 10.1080/00221309.1991.9711128 To link to this article: http://dx.doi.org/10.1080/00221309.1991.9711128
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.
Downloaded by [York University Libraries] at 10:02 22 November 2014
This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
The Journal of General Psychology, 118(1),5-12
Downloaded by [York University Libraries] at 10:02 22 November 2014
Efficacy of Caffeine Versus Expectancy in Altehng Caffeine-Related Symptoms LARRY CHRISTENSEN JEFF MILLER DONNA JOHNSON Department of Psychology Texas A&M University
ABSTRACT. The study investigated the independent and interactive effects of caffeine and expectancy on caffeine-relatedsymptoms. High- and low-caffeine consumers were randomly assigned to either an expectancy or nonexpectancy instructional set and one of four caffeine doses. Subjects were administered the State-Trait Anxiety Inventory, (Spielberger & Gorsuch, 1970) and a Symptom Questionnaire(Christensen, White, Krietsch, & Steele, 1990) prior to and 45 min followingconsumption of one of the four caffeine doses. An analysis of covariance identified a significant main effect for the State-Trait Anxiety Inventory scores and significant main and interaction effects for four Symptom Questionnaire items. However, when the alpha levels were corrected for the increased probability of Type I error, using the Bonferroni procedure, these effects failed to achieve significance. These results suggest that previous reports of subjective caffeine effects are also suspect because of their failure to control for the increased probability of 5 p e I error.
CAFFEINE,A CENTRAL NERVOUS SYSTEM STIMULANT,is generally assumed to have the effect of reducing drowsiness and fatigue and of producing a clearer flow of thought (Rall, 1980). At high doses, it also produces restlessness, irritability, agitation, and sensory and sleep disturbances (Edelstein, Keaton-Brasted, & Burg, 1984; James & Stirling, 1983). These effects are assumed to be due to caffeine, with little consideration given to the role of subject’s expectancies about the drug’s effect. It has been well documented (Ross & Olson, 1981) that placebos can produce a variety of effects, such as a reduction in clinical pain, increased arousal, reduced anxiety, reduced depression, feelings of increased alertness, Requests for reprints should be sent to Larry Christensen, Department of Psychology, Texas A&M University, College Station, lX 77843. 5
Downloaded by [York University Libraries] at 10:02 22 November 2014
6
The Journal of General Psychology
tension, relaxation, or drowsiness. These placebo effects generally correspond to one’s knowledge or belief about the type of drug being consumed, thus forming a relationship between expectancy and the placebo reaction. Although such a relationship would also seem to exist with caffeine, little research has been conducted on this issue. The few studies that have investigated the expectancy effects of caffeine have induced them through the use of an instructional set. Ammon, Carlson, Froberg, Karlsson, & Levi (1973) found that subjects who were caffeine consumers were much less likely to report fatigue and more likely to report tension and nervousness when told that they were taking caffeine. Christensen, White, Krietsch, & Steele (1990) revealed that subjects who were informed that the gelatin capsule they consumed, ostensibly containing caffeine, should induce a variety of symptoms, perceived an increase in feelings of alertness, sleepiness, and clarity of thought over the perception of subjects who were not given a specific expectancy. Kirsch and Weixel (1988) also documented an expectancy effect with respect to caffeine but found different effects when using a double-blind challenge design versus deceptive instructions (informing subjects they were getting caffeinated coffee when they really received decaffeinated coffee). The studies, although few in number, have consistently revealed that instructions can induce a variety of symptoms commonly associated with the consumption of caffeine. Their findings are consistent with the results of other drug studies (Hughes, Pickens, Spring, & Keenan, 1985; Kirsch & Weixel, 1988), indicating that instructions can control the occurrence of drug effects. However, little information is provided on the interactiveeffect of instructions and caffeine dose. The present study used a balanced placebo design (Marlatt & Rohensow, 1980) to investigate the independent and interactive effects of caffeine and instructionally induced expectancy on caffeine-related symptoms. This design yields a matrix in which subjects are (a) told they will get a drug and receive it, (b) told they will get a drug but receive placebo, (c) told there will be no drug but they do receive a drug, (d) told there will be no drug and they receive none. In the present study, several drug or caffeine intake levels were used to assess the potential independent and interactive effects of instructionally induced expectancies and increasing levels of actual consumption of caffeine.
Method Subjects and Instruments
The subjects were 120 undergraduate students (39 men and 81 women), enrolled in an introductory psychology class at Texas A&M University. When
Downloaded by [York University Libraries] at 10:02 22 November 2014
Christensen, Miller, & Johnson
7
signing up for the experiment, the subjects were informed that they would be participating in a caffeine experiment. Measures of caffeine-related symptoms included the Strate-Trait Anxiety Inventory (STAI; Spielberger & Gorsuch, 1970) and a Symptom Questionnaire (Christensen et al., 1990). The STAI is a 40-item self-report inventory that requires subjects to respond on a 4-point scale ranging from not at all (1) to very much so (4) for the measure of state anxiety and from almost never (1) to almost always (4) for the trait measure of anxiety. The Symptom Questionnaire is a self-report questionnaire that requires subjects to evaluate themselves on 12 symptoms that can be produced by ingestion of caffeine (Gilliland & Andress, 1984). Each item is evaluated on an 11-point scale ranging from not at all (0) to extremely (10). A food record was constructed that required subjects to record all foods, beverages, and medications consumed as well as the amount consumed to provide an assessment of the amount of ad libitum caffeine consumed. The Social Desirability Scale (Crowne & Marlowe, 1964) was administered to subjects to obtain an assessment of their tendency to evaluate themselves in a socially desirable manner. This scale provided an assessment of the extent to which a socially desirable response style would moderate susceptibility to the instructional set. The Social Desirability Scale is a 33-item inventory with a true-false format. The drug used was anhydrous caffeine. Doses (100 mg, 300 mg, and 500 mg) were prepared in soft gelatin capsules. Placebo capsules containing cellulose were prepared in identical-sized soft gelatin capsules. Procedure Subjects read and signed an informed consent form and were instructed to complete the food record sheets for 3 consecutive days, to avoid caffeine for a 24-hr period before experimental participation, and to provide a saliva sample to measure compliance with this request. (These samples were not analyzed,) Subjects then selected a time between 9 A.M. and 4 P.M. during the next week to participate in the caffeine experiment. Subjects stating that they had avoided caffeine during the prior 24-hr period provided a saliva sample. Subjects who did not avoid caffeine were released from the study. Food records were then collected and inspected to obtain an assessment of ad libitum caffeine consumption. While the food records were being inspected, the subjects completed the STAI, the Symptom Questionnaire, and the Social Desirability Scale. In making the assessment of caffeine consumption, an 8-02 cup of coffee was assumed to contain 100 mg of caffeine and a 12-02 cola beverage or an 8-02 glass of tea was assumed to contain 50 mg of caffeine. The caffeine content of medications was determined by consulting a sheet that listed the
Downloaded by [York University Libraries] at 10:02 22 November 2014
8
The Journal of General Psychology
caffeine content of most prescription and over-the-counter medications. By this assessment procedure, 40 subjects were classified as high- (2200 mg/ day; average daily intake, 262.7 mg.; range, 200-550 mg) and 80 as low( .05. The high- and low-caffeine consumers’ scores were then combined and subjected ‘Huberty and Moms (1989) have also revealed that the MANOVA-ANOVA approach commonly used when analyzing multiple outcome variables does not provide the assumed control for Type I error probabilities.
Downloaded by [York University Libraries] at 10:02 22 November 2014
Christensen, Miller, &Johnson
9
to a three-way classification analysis of covariance (ANCOVA), in which the independent variables were expectancy, amount of caffeine consumed, and high or low social desirability. Categorization into high or low social desirability was determined by a median split of the subjects' scores on this scale. An ANCOVA conducted separately on the STAI posttest scores, using the pretest scores as the covariate, revealed a significant difference for the main effect of expectancy, F( 1 , 101) = 4.99, p < .05, but only for the state anxiety scores. An ANCOVA conducted on each of the Symptom Questionnaire posttest item scores, using the pretest scores as the covariate, revealed a significant main effect of expectancy on the items measuring alertness, F(1,91) = 7.26, p < .05, and rapid heart beat, F(1,91) = 5.36, p < .05; a significant main effect of caffeine on the items measuring anxiety, F(3,91) = 2.93, p < .05, and relaxation, F(3, 91) = 2.97, p < .05; a significant Caffeine x Expectancy interaction on the items measuring headache, F(3,91) = 2.81, p < .05, and clear flow of thought, F(3, 91) = 4.19, p C .05; a significant Caffeine x Sociability interaction on the item measuring anxiety F(3, 91) = 2.93, p < .05; and a significant Caffeine x Expectancy x Sociability interaction on the item measuring alertness, F(3, 91) = 2.96, p < .05. Although a number of main and interaction effects were found to be significant on the basis of the multiple univariate ANCOVAs, this procedure enhances the probability of Qpe I error. To control for Qpe I error, a Bonferroni adjustment of the alpha level was incorporated. With this adjustment, an alpha level of .004 is needed to attain significance that results in none of the previously identified main or interaction effects achieving significance.
Discussion The results of the present study identified a number of significant main and interaction effects when each univariate ANCOVA was considered separately. However, when the probability values obtained from these univariate analyses were corrected for the increased probability of T@e I error, none of the effects achieved significance, which resulted in an inability to support the hypothesis that caffeine has an effect on the subjective mood of college students. This failure to find a significant effect is consistent with the bulk of literature suggesting that caffeine has little effect on performance (Sawyer, Julia, & 'hrin, 1982). The results are, however, inconsistent with several recent studies (Griffiths, Bigelow, & Liebson, 1986; Griffiths & Woodson, 1988; Stem et al., 1989) indicating that caffeine in the absence of expectancy produces's variety of subjective effects and results (Christensen et al., 1990; Kirsch & Weixel, 1988; Mitchell, Ross, & Hurst, 1974), suggesting that the expectancy effect may account for a rather significant portion of the effects generally attributed to caffeine.
Downloaded by [York University Libraries] at 10:02 22 November 2014
10
The Journal of General Psychology
The discrepancy existing between these studies and the present one may be that many of these prior studies did not provide adequate control for Type I error because of the use of multiple univariate analysis or the use of MANOVA as a method for controlling Type I error. In the present study and in most of the prior studies (e.g., Griffiths et al., 1986; Griffiths & Woodson, 1988; Stem et al., 1989), multiple outcome measures were incorporated and each was analyzed by univariate analyses to assess the subjective effects of caffeine. This procedure enhanced the probability of falsely rejecting the null hypothesis. In the present study, several main and interaction effects would have been considered significant if a correction had not been included to control for this enhanced probability of Type I error. When a Bonferroni correction of the alpha level was incorporated to control for Type I error probability, none of the previously identified effects were significant. Stem et al., (1989), Griffiths & Woodson (1988), and Griffiths et al., (1986) analyzed each of their multiple outcome measures using ANOVA without such a correction, which increased the probability of falsely rejecting the null hypothesis. Without including a control for the increased probability of Type I error, it is impossible to determine which of the effects identified in these studies are real and which are due to chance. If Type I error was controlled, the results of these studies (Griffiths et al., 1986; Griffiths & Woodson, 1988; Stem et al., 1989) would possibly support the present study, indicating that caffeine produces few subjective effects. The discrepancy between the results of the present study and those of the Christensen et al. (1990) study is probably due to the method of attempting to control Type I error. The Christensen et al. (1990) study made use of MANOVA, a procedure currently being questioned (Huberty & Morris, 1989) as a method for controlling Type I error. In the present study, use of the Bonferroni correction, was probably the more stringent as well as the more appropriate procedure but resulted in the failure to find a significant expectancy effect. It is also possible that the discrepancy between the present study and many prior studies is due to differences in subject samples. The subject sample in the present study consisted of a volunteer group of college students obtained from introductory psychology classes with an average daily caffeine consumption of 129.4 mg. Other studies (e.g., Griffiths et al., 1986; Griffiths & Woodson, 1988) used an older sample recruited from advertisements and consuming much more caffeine (e.g., mean daily consumption, 361 mg, in the Griffiths & Woodson study). Such differences, particularly in daily caffeine consumption, suggest that many prior studies may have obtained a subject population self-selected in terms of their reactivity to caffeine. Such a subsample may be reactive to caffeine and may experience subjective symptoms. The results of the present study suggest that drawing such a conclusion for the general population, particularly for college students, is tenuous.
Downloaded by [York University Libraries] at 10:02 22 November 2014
Christensen, Miller, Br Johnson
1I
It is also possible that the stimulant effects of caffeine are situation specific. Goldstein, Kaizer, & Warren (1969) have observed that caffeine effectively counteracts the decrement in a variety of types of performance caused by fatigue or sleep deprivation but is less effective in enhancing performance under other conditions. This observation suggests that the stimulant effects of caffeine are most apparent and operate most effectively under fatigue or sleep deprivation conditions. If this is true, then assessment of caffeine effects under rested or nonfatigued conditions would be subject to a ceiling effect and generate minimal effects as observed in this and other (Mitchell et al., 1974) studies. This assessment is consistent with the conditions under which the stimulant effect of caffeine is generally desired: in the morning, afternoon, or at other times when fatigue and lethargy are beginning to develop. Perhaps caffeine effects should be investigated under these conditions; at least, a fatigue or sleep deprivation condition should be included in caffeine studies to identify the differential effects that occur when subjects are rested versus fatigued. REFERENCES Ammon, H. P.T., Carlson, L. A., Froberg, J., Karlsson, C. G., & Levi, L. (1973). Effects of coffee and caffeine on sympathoadrenomedullary activity, blood lipids, psychological ratings and performance (Report No. 31). Stockholm, Sweden: Laboratory for Clinical Stress Research, Karolinska Institute. Christensen, L.. White, B., Krietsch, K., & Steele, G. (1990). Expectancy effects in caffeine research. The International Journal of the Addictions. 25, 27-3 1. Crowne, D., & Marlowe, D. (1964). Social desirability scale in The approval motive. New York: Wiley. Edelstein, B. A., Keaton-Brasted, C., & Burg, M. M. (1984). Effects of caffeine withdrawal on nocturnal enuresis, insomnia, and behavior restraints. Journal of Consulting and Clinical Psychology, 52, 857-862. Gilliland, K., & Andress, D. (1984). Ad lib caffeine consumption, symptoms of caffeinism, and academic performance. American Journal of Psychiatry, 138, 5 12-5 14. Goldstein, A., Kaizer, S.,& Warren, R. (1969). Psychotropic effects of caffeine in man. 11. Alertness, psychomotor coordination, and mood. Journal of Pharmacology and Experimental Therapeutics, 150, 146-1 5 1. Griffiths, R. R., Bigelow, G. E., & Liebson, I. A. (1986). Human coffee drinking: Reinforcing and physical dependence producing effects of caffeine. Journal of Pharmacology and Experimental Therapeutics, 239, 4 16-429. Griffiths, R. R., & Woodson, P. P. (1988). Reinforcing effects of caffeine in humans. Journal of Pharmacology and Experimental Therapeutics, 246, 2 1-29. Huberty, C. J., & Moms, J. D. (1989). Multivariate analysis versus multiple univariate analyses. Psychological Bulletin, 105, 302-308. Hughes, J. R., Pickens, R. W., Spring, W., & Keenan, R. M. (1985). Instructions control whether nicotine will serve as a reinforcer. Journal of Pharmacology and Experimental Therapeutics, 235. 106-1 12. James, F. E., & Stirling, K. P. (1983). Caffeine: A survey of some of the known
Downloaded by [York University Libraries] at 10:02 22 November 2014
12
The Journal of General Psychology
and suspected deleterious effects of habitual use. British Journal of Addiction, 78, 25 1-258. Kirsch, I., & Weixel, L. J. (1988). Double-blind versus deceptive administration of a placebo. Behavioral Neuroscience 102, 319-323. Marlatt, G.A., & Rohensow, D. J. (1980). Cognitive processes in alcohol use: Expectancy and the balanced placebo design. In N. K. Mello (Ed.), Advances in substance abuse: Behavioral and biological research. Greenwich, CT:JAI Press. Mitchell, V. E., Ross, S., & Hurst, P. M. (1974). Drugs and placebos: Effects of caffeine on cognitive performance. Psychological Reports, 35, 875-883. Rall, T. W. (1980). Central nervous system stimulants. In G. A. Goodman & A. Gilman (Eds.), The phrmcological basis of therapeutics. (6th ed., pp. 592607). New York: Macmillan. Ross, M., & Olson, J. M. (1981). An expectancy-attribution model of the effects of placebos. Psychological Review, 88, 408-437. Sawyer, D. A., Julia, H. L., & Turin, A. C. (1982). Caffeine and human behavior: Arousal, anxiety, and performance effects. Journal of Behavioral Medicine, 5 , 4 19-439. Spielberger, C. D., & Gorsuch, R. L. (1970). The State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologist Press. Stem, K. N., Chait, L. D., & Johanson, C. E. (1989). Reinforcing and subjective effects of caffeine in normal human volunteers. Psychopharmacology.98, 8 1-89. Received August 20, 1990
