Previous research has shown a link between low positive affect (PA) and numerous physical and psychological well-being outcomes but, recent research has raised the possibility that this relationship may be driven by physical activity. Thus, we were interested in exploring the PA-exercise connection by examining this relationship across differing levels of body mass and athleticism. We also looked at whether the item “active” that is used in many PA assessments was responsible for this effect.
Participants were part of the Norwegian SHoT2018 national survey of 50,054 young adults (mean age = 23.2, 68.9% women), who completed electronic surveys about their exercise levels (duration, frequency and intensity) and affect.
There was a clear and strong dose-response association between current state PA and the duration, frequency and intensity of exercise. For example, duration, magnitude, and slope effects were strongly driven by regular exercisers who had more than a 20-fold greater likelihood of being in the highest PA deciles compared to the least frequent exercisers. These dose-response connections replicated across both healthy and overweight BMIs, as well as in elite athletes. Removing the word “active” from the PA measure substantially reduced the size of this association, although the dose-response relationship remained.
The observed strong connections have critical implications for health researchers and clinicians, and point to a need to carefully consider what types of activities are most strongly tied to well-being.
There is a strong dose-response connection between exercise and positive emotion. This pattern persists across a range of body types and athlete characteristics. Overlap between adjectives in affect and self-reported activity scales partially but not fully explains the connection. Past work showing benefits of happiness on health may be partially or primarily driven by activity, not emotion.
It has long been established in both research and in common public knowledge that exercise can lead to greater positive affect (PA;
Understanding the nature of this association becomes even more critical given the burgeoning literature connecting PA to better physical health across a wide range of domains (
This work has helped foster a new field of “Positive Health” research (
This problem is compounded by the fact that when utilizing self-report scales, there can be a large overlap between physical health self-reports and PA self-reports. For example, many popular affect measures rely on adjectives like “active” and “energetic” to tap positive affect (
We examined this issue recently in a large sample of over 5000 older adults (
This minimalist approach to assessing physical activity is echoed across the PA-health literature, including in studies showing that it is the high and not low energy components of PA most tied to reduced mortality (no activity control) (
Thus, clearly there is a need to examine the association between these related variables in more detail where PA can be compared to a range of activity markers across a large number of individuals. Furthermore, given the concern about high PA simply being a marker of healthy fitness, this should be tested in those both high and low in fitness. This will enable a deeper understanding of the degree of connection association between fitness and high arousal PA, and more clarity about past research linking PA to health and mortality. To examine the extent to which PA and physical activity are overlapping constructs, we used data from the SHoT2018 study, a sample of over 50,000 Norwegian young adults. We hypothesized that PA would be strongly associated with
The SHoT study (an acronym for the Norwegian name: Studentenes Helse- og Trivselsundersøkelse
The SHoT2018 study was a joint effort between the three largest student welfare organizations in Norway and the Norwegian Institute of Public Health (NIPH). The study was conducted between February 6th and April 5th, 2018, on all full-time Norwegian students taking higher education (both in Norway and abroad). The collection of the health survey was in close collaboration with all the student welfare organizations in Norway. Students were told that participation was completely voluntary, and that there were no penalties for not filling out the survey. Eight percent of the sample were immigrants, defined as either the student or his/her parents being born outside of Norway.
The study protocol was approved by the Regional Committee for Medical and Health Research Ethics of Western Norway (no. 2017/1176/REK vest), whose directives are based on the Declaration of Helsinki. Written electronic consent was obtained from all subjects included in this study.
The PANAS (
The students were first presented with the following brief definition of physical exercise: “With physical exercise, we mean that you, for example, go for a walk, go skiing, swim or take part in a sport.” Physical exercise was assessed using three sets of questions, assessing the average number of times exercising each week, and the average intensity and average hours each time: 1)
IBM SPSS Statistics 25 for Mac (SPSS Inc., Chicago, IL) was used for all analyses. Multinomial logistic regression models were computed to assess the association between levels of physical exercise (independent variable; lowest level of the three physical exercise variables being the reference category) and deciles of PA (dependent variable; first decile being the reference category). Being similar to binary logistic regression, multinomial regression is used when the dependent variable is nominal with more than two levels. Results are presented as odds-ratios (ORs) with 95% confidence intervals (95% CIs). There was very little missing data on the PA items, with missing responses ranging from
In terms of
The response distribution of the 10 PA items for both men and women are presented in
The physical exercise characteristics according to sex-specific tertiles on the PA-scale are presented in
Group Characteristics | Women |
Men |
||||
---|---|---|---|---|---|---|
PA Lower tertile | PA Middle tertile | PA Upper tertile | PA Lower tertile | PA Middle tertile | PA Upper tertile | |
Age group | ||||||
18-20 years | 36.0% | 34.8% | 29.2% | 36.5% | 32.8% | 30.7% |
21-22 years | 33.5% | 34.7% | 31.8% | 31.7% | 34.0% | 34.3% |
23-25 years | 33.7% | 34.0% | 32.3% | 33.8% | 31.8% | 34.3% |
26-28 years | 35.6% | 32.2% | 32.2% | 39.0% | 30.4% | 30.6% |
29-35 years | 30.3% | 32.7% | 37.0% | 38.8% | 31.9% | 29.3% |
Physical exercise (frequency) | ||||||
Never | 59.5% | 27.2% | 13.3% | 62.5% | 24.4% | 13.1% |
Less than once a week | 47.1% | 34.2% | 18.8% | 51.6% | 28.2% | 20.1% |
Once a week | 39.9% | 35.3% | 24.8% | 40.2% | 35.7% | 24.0% |
2–3 times per week | 31.1% | 35.5% | 33.5% | 31.6% | 34.5% | 34.0% |
Almost every day | 24.6% | 31.4% | 44.1% | 22.0% | 31.3% | 46.8% |
Physical exercise (intensity) | ||||||
I take it easy without breaking into a sweat or losing my breath | 43.5% | 34.6% | 21.9% | 48.0% | 29.7% | 22.3% |
I push myself so hard that I lose my breath and break into a sweat | 31.0% | 34.7% | 34.3% | 31.4% | 34.1% | 34.5% |
I push myself to near-exhaustion | 27.0% | 31.0% | 42.1% | 27.4% | 30.4% | 42.2% |
Physical exercise (duration) | ||||||
Less than 15 minutes | 55.0% | 29.0% | 16.0% | 58.5% | 25.4% | 16.2% |
15–29 minutes | 43.5% | 34.8% | 21.7% | 42.8% | 31.7% | 25.5% |
30 minutes to 1 hour | 33.4% | 34.3% | 32.3% | 35.5% | 33.5% | 30.9% |
More than 1 hour | 27.9% | 34.5% | 37.6% | 28.2% | 32.8% | 39.0% |
Top athlete | ||||||
Yes | 18.8% | 26.9% | 54.3% | 15.6% | 23.0% | 61.3% |
No | 24.9% | 31.6% | 43.5% | 22.2% | 31.8% | 46.0% |
A similar pattern was observed for the item assessing the
Finally, students considering themselves to be a top athlete had significantly higher odds of also having a high level of PA. As shown in
As shown in
Removing the adjective “active” from the PANAS cut the association between PA and exercise frequency a great deal. As shown in
Overall, this study replicates past findings indicating a strong association between PA and exercise in a large general population study of over 50,000 Norwegian young adults, but it also adds a great deal of new information. First, we show for the first time a surprisingly clear and strong dose-response relationship between PA and physical exercise across
All together, these results indicate that high PA, as assessed by the PANAS, is in fact picking up on activity to a large extent, especially when assessing regular exercisers. That is, the majority of the
We must also ask the more critical question of how these results impacts our interpretation of the literature connecting PA to better health? The findings clearly cast some doubt on health studies utilizing the PANAS PA or similarly active measures of PA (e.g., vigor), especially among results that don’t account for the effects of physical activity in some way. That said, even if they did, given the typically limited fitness measures used in some studies, more work is probably needed to ensure that it is not simply the most physically fit, active, and healthy people driving these findings or some other related third variable (e.g., cardiovascular health, mitochondrial function) (
In future research exploring the relationship between PA and health we recommend researchers consider taking extra effort to separate the effects of PA from physical activity when exploring health outcomes. This might be done by utilizing objective fitness indicators such as VO2max, accelerometers, extensive exercise and activity self-reports, in addition to covarying perceived health which is likely to relate strongly to fitness. It is only with these deeper and more objective approaches that we will begin to understand when feelings of positivity are promoting health versus activity levels (i.e., healthiness) promoting health.
This study has both strengths and weaknesses. While it is well powered and has an array of physical exercise assessments, it is limited by its cross-sectional design and reliance on self-report. Generalization is also limited to young, healthy, and primarily Caucasian samples. Given the high self-reported exercise levels of this sample, it would be interesting to also contrast these levels against objective activity assessments as well as to look at less active and older samples.
In addition, the use of a state (current) affect scale was weaker than that of a trait (long lasting) affect scale, although the two are known to be highly correlated (
Overall, this study shows strongly that exercise and positive emotions are closely intertwined, especially for the healthiest and most fit individuals. Future work should examine how the same effects are found with objective measures of activity and fitness, and should also further examine the implications for physical health outcomes. That is, when examining PA and health connections, to what extent do these change if we focus on PA measures that do not tap energy, felt vigor, and activity? What happens when we take great efforts to account for activity and fitness? From this data, we might infer that this would not have major implications for sedentary samples, however, for samples that include active individuals, effects may change drastically. It is essential that those of us interested in PA and health start measuring exercise well and that we take the possible different interpretations of high activity/arousal PA effects into account. The extent that we discover that activity levels underlie a large amount of previously observed PA health benefits, it may be the case that activity interventions (with or without PA) may be a more fruitful approach to improving health.
NA was not the focus of the paper given our past work showing that it does not alter PA-health associations (
SHoT2018 has received funding from the Norwegian Ministry of Education and Research (2017), and the Norwegian Ministry of Health and Care Services (2016).
The authors have declared that no competing interests exist.
Twitter Handles for authors: @sarahpressman @KeithPetrie @borgesivertsen
The authors wish to thank all students participating in the study, as well as the three largest student welfare associations in Norway (SiO, Sammen and SiT) who initiated and designed the SHoT study.