The Effect of Caffeine on Positive Affect

Author’s Note

I wrote this paper for PSC103A: Statistical Analysis of Psychological Data with Rohit Bantra. As an avid coffee connoisseur, I was interested in finding whether or not my caffeine consumption was good for my mental health. I hope this paper interests those who also enjoy their coffee, even if they are a tad disappointed with the results (as I was). This was my first experience coding in R and engaging with quantitative psychology, which both came in handy when I used the same skills/concepts to complete my honors thesis. 

For many people, caffeine is a part of their daily routine. As such, plenty of research has investigated the effects of caffeine on mood. Generally, the National Institutes of Health recommends a dose of 2 to 6 mg of caffeine per kilogram of body weight in order to feel positive effects of the drug (Office of Dietary Supplements, 2017). In a 2005 study, researchers using the caffeine research visual analogue scale found that participants who consumed caffeine reported increased feelings of alertness and decreased feeling of tiredness. The mood measures relating to arousal significantly improved in both habitual and non-habitual caffeine users (Haskell et al., 2005). Another study focusing on college students mirrored these findings about caffeine’s effect on alertness. Participants who consumed caffeine reported less feelings of “sleepiness” than participants who did not consume caffeine. However, the caffeinated group also rated “tense negative mood” higher than the uncaffeinated group (Attwood et al., 2007). This demonstrates how perhaps caffeine influences negative feelings. Researchers using the Profile of Mood States found that participants in the caffeine group reported higher arousal and, in contrast to the above findings, more positive mood than participants in the placebo group (Childs & De Wit, 2006). Prior research tends to agree that caffeine increases feelings of alertness, however, researchers cannot explicitly identify the relationship between caffeine and mood. As such, this paper will offer more clarity about this relationship by investigating the strength of an inverse relationship between caffeine and positive mood. For the purposes of this paper, it is hypothesized that participants who ingested caffeine will, on average, have a lower positive affect score than participants who were given the placebo.

Methods

Participants
For this study, a total of 3032 participants were randomly selected from an Introductory Psychology experimental pool of undergraduate lowerclassmen at Northwestern University over nine years, 1989 to 1998. Students received course credit for participation (Rafaeli & Revelle, 2006). Of these participants, 784 people consumed caffeine and 778 people did not consume caffeine. 

Procedure
The data presented in this paper were collected at the Personality, Motivation, and Cognition Laboratory at Northwestern University. The dataset specifically used for the purposes of this paper was the “msqR” dataset, which included the Motivational State Questionnaire-Revised Form (MSQ-R). The MSQ-R has 72 mood items that participants rate on a four-point scale from 0, not at all, to 3, very much. Certain subsets within the MSQ-R have been shown to demonstrate energetic arousal and tense arousal (Thayer, 1986) as well as positive and negative affect (Watson et al., 1988). Before completing the study, participants signed a consent form. They then consumed 0 or 4 mg/kg of caffeine before taking the MSQ-R (Rafaeli & Revelle, 2006). 

These data were accessed using the “psychTools” package in RStudio. In order to analyze these data, the raw data were made into a sum scale score by summing the scores for positive affect: active, alert, attentive, determined, enthusiastic, excited, inspired, interesting, proud and strong. Only data from Time = 1 were considered for the purpose of this paper. The sum scale scores for each individual was calculated based on what they reported on the previously listed adjectives. Then, the sum scale scores of those who consumed caffeine and those given the placebo were averaged. 

Data Analysis
The following data analysis was completed by using the application R version 4.2.1. An independent samples t-test was appropriate for analyzing these data because the three assumptions necessary to use this test were met. First, the samples were independent of one another, as the selection of participants in each group were not influenced by the selection of participants in the other group. Second, the sum scale scores were normally distributed across both groups. Neither histogram (see Figures 1 and 2) for the non-caffeinated or caffeinated groups indicated that the positive affect sum scores were normally distributed. Many participants reported scores of zero or one and, as a result, both histograms were right skewed. 

The Shapiro-Wilks test was conducted as a secondary check for normality. The results of the Shapiro-Wilks test for the original data failed to indicate that the positive affect sum scores were normally distributed in either group. The Shapiro-Wilks test for the square root transformation also did not indicate normality. The presence of positive affect sum scale scores of zero made it impossible to perform log transformations or reciprocal transformations of the original data. Due to the fact that neither of the Shapiro-Wilks tests completed indicated that the scale sum scores of both groups were normally distributed, the Central Limit Theorem was necessary to determine normality. Both groups in the dataset had sample sizes of above 30, so the Central Limit Theorem applied. Thus, the sampling distributions of the positive affect sum score means were normally distributed. Additionally, the t-test can be used even when there is minor non-normality. 

Third, homogeneity of variances was addressed. For the purposes of this research, it was assumed that the variances of both groups were not the same. In order to account for this, Welch’s t-test was used. For the independent Welch’s t-test, the null hypothesis provided was that the positive affect sum scores for both the caffeine and placebo groups were the same. The alternative hypothesis used was the positive affect sum scores for the caffeine group were greater than the positive affect sum scores for the placebo group.

Results

There was a significant difference between the average positive affect sum scores of participants in the no caffeine condition (M = 11.71, SD = 7.15) and participants in the caffeine condition (M = 8.82, SD = 6.31), t(1532.8) = 8.45, p < 0.001). 

Conclusion

The results of the Welch’s t-test affirmed the original hypothesis that caffeine is associated with lower positive affect scores. These results do not align with the literature stating that caffeine improves mood, but do align with findings that caffeine negatively impacts positive affect. This inverse relationship between caffeine intake and good mood was surprising, considering that one aspect of positive affect evaluated the extent to which participants felt “alert”. This demonstrates that even if caffeine users do consistently report increased alertness as reported by other studies, caffeine does not necessarily correlate with good mood overall (Haskell et al., 2005; Attwood et al., 2007). This would also mean that caffeine has enough of a detrimental effect on other aspects of positive affect, such as enthusiasm, excitement, and inspiration, that it cancels out its effects on alertness. 

Additional research should further investigate the effects of caffeine on mental health outcomes of regular caffeine users. Using the same dataset as was used in this research project, one could consider caffeine’s effect on negative affect as well. If caffeine intake is indicative of lower positive affect scores, would it be associated with higher negative affect scores? Additionally, these data were from college students specifically. As such, it would be interesting to find if caffeine’s inverse relationship with positive affect could be replicated with different age groups. This information would be of interest to a multitude of caffeine consumers, as perhaps caffeine is detrimental enough for mood that it is not worth a boost of energy. 

References

Attwood, A. S., Higgs, S., & Terry, P. (2007). Differential responsiveness to caffeine and perceived effects of caffeine in moderate and high regular caffeine consumers. Psychopharmacology, 190(4), 469–477. https://doi.org/10.1007/s00213-006-0643-5

Childs, E., & De Wit, H. (2006). Subjective, behavioral, and physiological effects of acute caffeine in light, nondependent caffeine users. Psychopharmacology, 185(4), 514–523. https://doi.org/10.1007/s00213-006-0341-3

Haskell, C. F., Kennedy, D. O., Wesnes, K. A., & Scholey, A. B. (2005). Cognitive and mood improvements of caffeine in habitual consumers and habitual non-consumers of caffeine. Psychopharmacology, 179(4), 813–825. https://doi.org/10.1007/s00213-004-2104-3

Office of Dietary Supplements - Dietary Supplements for Exercise and Athletic Performance. (2017). Nih.gov. https://ods.od.nih.gov/factsheets/ExerciseAndAthleticPerformance-Consumer/ 

Rafaeli, E., & Revelle, W. (2006). A premature consensus: Are happiness and sadness truly opposite affects? Motivation and Emotion, 30, 1–12. https://doi.org/10.1007/s11031-006-9004-2

Thayer, R. E. (1986). Activation-Deactivation Adjective Check List: Current Overview and Structural Analysis. Psychological Reports, 58(2), 607–614. https://doi.org/10.2466/pr0.1986.58.2.607

Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070. https://doi.org/10.1037/0022-3514.54.6.1063

Appendix

Figure 1
Histogram of Positive Affect Sum Scores for Participants Who Consumed Caffeine

Figure 2
Histogram of Positive Affect Sum Scores for Participants Who Did Not Consume Caffeine