Physical activity and nutrition in relation to resilience: a cross-sectional study




The data was collected during the first half of 2022. Participants between the ages of 18 up to about 70 years were recruited from an online-platform for social research and via advertisement in student classes in Germany. The students received course credit for participation, other participants received approximately € 4. Initially, 749 participants responded to the survey, but 54 could not be considered due to questionable validity of the responses (unreasonably fast completion, incomplete or monotonous responding). Ten participants had to be excluded because they were physically disabled or their responses were outliers with extreme z scores ( >|+ − 3|). The final sample consisted of 685 participants aged 24–72 years (M = 48.95, SD = 11.25); 52% of the sample were female; males and females did not differ significantly in age, t(681) = 1.23, p = 0.22. The body mass index was M = 26.58 (SD = 5.51). Seventy-four percent practise at least one sport regularly. Jogging (N = 170) and cycling (N = 144) were the most frequently mentioned activities. Data were checked to ensure the assumptions of normality and outliers. Skew and kurtosis values were below ± 1 for the main variables (see Table 1).

Table 1 Descriptive statistics and bivariate correlations.


Physical activity

Nine items adapted from an instrument developed by Fuchs, Klaperski, Gerber, and Seelig17 were used. Fuchs et al.17 differentiated between sport, occupational or daily activities. We were interested in a more global value and asked participants to think about the frequency of their activities over the past four weeks (1 = very seldom to 7 = very often). They were asked to think about and include all their activitites over the past four weeks, whether at home or at work, as well as sport and leisure activites. Items were physically demanding (house)work, physically demanding caregiving tasks, gardening, intensive exercise in everyday life, moderate execise in everyday life, athletic activities, cycling, walking (Cronbach’s alpha was 0.75).

Less healthy foods/junk food

Due to time constraints, we focused on single items to measure the types of less healthy foods (e.g., fast food, sweets, higher meat consumption) instead of using rather long nutrition inventories. In detail, participants were asked how often they had eaten the following foods in the past four weeks (1 = very seldom to 7 = very often): (1) Fast food (e.g., burger, french fries), (2) Junk food/sweets (e.g., cake, biscuits, soft drinks, iced tea), (3) Junk food/snacks (e.g., crisps, pretzels), (4) Meat. In addition, (5) Adding sugar to foods/drinks was assessed. Scores of these five indicators were entered into a factor analysis. Using principal components, one factor was extracted, accounting for 47% of the variance. The mean value was computed (α = 0.71).

Nutritional awareness

Four items were used to measure the degree to which participants pay attention to their diet. “Do you watch what you eat?” (0 = no, 1 = sometimes, 2 = yes), “Please provide the reasons (e.g., body cult, health, improving physical fitness, maintaining physical fitness)”, the number of reasons was used (0, 1, 2 or more reasons), “Do you follow a dietary concept (e.g., healthy diet, healthy eating, attention to nutritional values)?” (0 = no, 1 = sometimes, 2 = yes), “Do you eat according to a specific nutritional concept (e.g., vegan, low carb, Paleo)?”, the number (0, 1, 2 or more) was used. Scores of these four indicators were entered into a factor analysis. Using principal components, one factor was extracted, accounting for 48% of the variance. We computed the mean value (α = 0.64).

Subjective stress

Subjective stress was measured with a shortened and adapted version of the Trier stress inventory46. The original scale consists of 39 items assessing facets of chronic stress: Worries, intrusive memories, work-related stress (work overload, work discontent), and social stress (social conflicts, lack of social recognition). To make the time window comparable to that of acivity and nutrition, participants were asked to rate how often they have experienced stress episodes during the last four weeks on a scale from 1 (very seldom) to 7 (very often). Example items are “I had intrusive thoughts about an unpleasant experience” or “I had no time for recreation”. We used 18 items from four facets. Because the four subscales were highly intercorrelated (r´s between 0.61 and 0.84), a global stress score was computed (α = 0.93).

Life satisfaction

Life satisfaction was assessed with the Satisfaction With Life Scale (SWLS;47). Participants were asked to rate how well each item applied to them in general (e.g., “I am satisfied with my life”) on a seven-point Likert scale; α = 0.91.

All calculations were conducted using SPSS version 27 and LISREL9.3. We computed bivariate correlations between the main variables. Because age and gender were significantly associated with some of the central constructs, they were used as control variables. To investigate the controlled associations, we used a structural equation model in which we regressed life satisfaction on physical activity, nutritional awareness, subjective stress and control variables. Parcels (test halves) were computed to represent the latent constructs. The model was based on maximum likelihood estimates and predictors were allowed to be correlated.


The descriptive statistics of all relevant variables and bivariate correlations are shown in Table 1. As expected, the amount of physical activity and nutritional awareness were positively correlated with life satisfaction. Subjective stress was negatively associated with life satisfaction, but positively with the consumption of less healthy food.

Because physical activity and nutritional awareness were correlated, and the degree of food consumption was different depending on age and gender, we conducted a controlled analysis. In a structural equation model, we regressed life satisfaction and less healthy food consumption on age, gender, subjective stress, physical activity, and nutritional awareness to show the unique contributions of stress and health-related lifestyle variables after controlling for sociodemographic variables (see Fig. 1). Results showed that the association between physical activity and life satisfaction remained significant after controlling for other variables; the association between nutritional awareness and life satisfaction, however, was not significant. The expected association between subjective stress and the consumption of less healthy foods remained significant.

Figure 1
figure 1

Structural equation model: associations with junk food and life satisfaction after controlling for age and gender. χ2(39) = 102.22; p < 0.01; RMSEA = 0.049; NNFI = 0.97; CFI = 0.98; AGFI = 0.95; ***p < 0.001. Depicted are standardized path coefficients.

In the third step, we tested the adaptive function of physical activity and nutritional awareness. Because we were interested in the moderating, stress buffering-effects of both variables, we used z-transformed scores to compute the interaction terms Subjective Stress x Physical Activity and Subjective Stress × Nutritional Awareness. Then we computed two moderated regression analyses (e.g., Subjective Stress, Physical Activity, Subjective Stress × Physical Activity → Life Satisfaction). As expected, the predictive value of subjective stress on life satisfaction was moderated by physical activity (βSubjective Stress × Physical Activity = 0.10, p < 0.01) and nutritional awareness (βSubjective Stress × Nutritional Awareness = 0.09, p = 0.015). Figure 2a,b depict the interaction effects. The negative associations between subjective stress and life satisfaction were less when physical activity and nutritional awareness were high.

Figure 2
figure 2

Moderating effects. Life satisfaction as a function of subjective stress and (a) physical activity, (b) nutritional awareness. Unhealthy nutrition as a function of subjective stress and (c) physical activity, (d) nutritional awareness.

To test our final hypothesis, we computed the interaction between subjective stress and nutritional awareness using a regression model Subjective Stress × Nutritional Awareness → Less Healthy Food. The interaction was significant (β Subjective Stress × Nutritional Awareness = 0.07, p < 0.05). As expected, the negative association between nutritional awareness and less healthy food consumption was absent when the degree of subjective stress was high (see Fig. 2c).

Because the correlation between physical activity and less healthy food was, in contrast to our expectations, significant and positive, we also examined the interaction effect with subjective stress (Subjective Stress × Physical Activity → Less Healthy Food). The interaction effect was significant (β = 0.09; p < 0.014), showing that in times of less stress, the relationship between physical activity and junk food was low, however, in times of high subjective stress, the correlation was high (see Fig. 2d).

In sum, results of Study 1 showed that the amounts of physical activity and nutritional awareness are correlated with life satisfaction; the association with nutritional awareness was small and failed to remain significant in a controlled analysis. Physical activity and nutritional awareness moderated the relationship between stress appraisals and general life satisfaction (stress-buffer effect). The relationship between nutritional awareness and less healthy food consumption was moderated by the degree of subjective stress. Interestingly enough, in contrast to the hypothesis, physical activity is correlated with less healthy food consumption. Because we did not differentiate between the situations in which people exercise (i.e. whether they do sport or have an active occupation), we conducted Study 2 to differentiate this more precisely. We also wanted to test whether the interactions can be replicated.


Leave a Reply

Your email address will not be published. Required fields are marked *