2.1 Executive Functions and Demographic Characteristics of College Students with Different Levels of Trait Anxiety
Based on previous studies, we divided trait anxiety into three groups[37], with high trait anxiety being one standard deviation above the mean score (39.931+9.161≈49); low trait anxiety being one standard deviation below the mean score (39.931-9.161≈31), and high, medium, and low trait anxiety scores were 52.805±3.132, 40.795±5.230, and 26.680±2.559, respectively.
Pearson's bivariate correlation analysis showed that college students' trait anxiety scores were significantly correlated with shifting function (r=0.182, P=0.004) and inhibition function (r=0.163, P=0.010) and not with working memory (r=0.056, P=0.385), as shown in Figure 2. One-way ANOVA showed that the inhibition function of the low trait anxiety group was significantly better than that of the high trait anxiety group (P=0.006), while the shifting function of the low trait anxiety group was significantly better than that of the medium trait anxiety group (P=0.007) and the high trait anxiety group (P=0.003), with no statistically significant difference in working memory between the three groups (P=0.278). There were no significant differences in BMI, age, gender, and percentage of only children between the three groups (all P>0.05), as detailed in Table 1.
Table 1 Executive Functions and Demographic Characteristics of College Students with Different Levels of Trait Anxiety
Variables
|
Grouping of trait anxiety
|
Comparison between groups
|
LSD post hoc multiple comparisons
|
Low (n=50)
|
Medium (n=156)
|
High (n=41)
|
F (χ2)
|
P
|
Low VS Medium
|
Low VS High
|
Medium VS High
|
Trait anxiety (score)
|
26.680±2.559
|
40.795±5.230
|
52.805±3.132
|
386.514
|
0.000***
|
0.000***
|
0.000***
|
0.000***
|
Inhibition function (ms)
|
95.916±41.805
|
108.033±42.689
|
121.742±51.588
|
3.866
|
0.022*
|
0.092
|
0.006**
|
0.078
|
Working memory (ms)
|
581.295±94.451
|
601.124±88.066
|
609.172±87.997
|
1.286
|
0.278
|
|
|
|
Shifting function (ms)
|
308.741±106.073
|
366.268±135.237
|
390.742±130.149
|
5.279
|
0.006**
|
0.007**
|
0.003**
|
0.281
|
BMI (kg/m2)
|
21.362±3.700
|
20.920±3.550
|
21.344±3.945
|
0.322
|
0.725
|
|
|
|
Age (Years)
|
20.083±1.820
|
19.791±1.830
|
19.564±1.569
|
0.938
|
0.393
|
|
|
|
Gender (Male%)
|
0.533
|
0.423
|
0.455
|
1.642
|
0.440
|
|
|
|
Only child (Yes%)
|
0.644
|
0.556
|
0.563
|
1.618
|
0.806
|
|
|
|
2.2 Effect of Physical Activity Levels on Trait Anxiety
Linear regression analysis was performed with physical activity level and different intensity levels of physical activity as independent variables and trait anxiety as the dependent variable, respectively. The results showed that physical activity level (Beta=-0.195, P=0.002, R2=0.038), high-intensity physical activity level (Beta=-0.185, P=0.004, R2=0.034), and medium-intensity physical activity level (Beta=-0.129, P=0.042, R2=0.017) all negatively predicted college students' trait anxiety levels, while the regression coefficients for low-intensity physical activity level (Beta=-0.047, P=0.463, R2=0.002) were not significant, as shown in Figure 3.
2.3 Effect of Physical Activity Level on Executive Function
Linear regression analysis was performed using the physical activity level and different intensity physical activity levels as independent variables and each subcomponent of executive functions as dependent variables, in separate cases. The results demonstrated that physical activity level (Beta=-0.285, P<0.001, R2=0.081), high intensity physical activity level (Beta=-0.208, P=0.001, R2=0.043), medium intensity physical activity level (Beta=-0.170, P=0.007, R2=0.029), low intensity physical activity level (Beta=-0.187, P=0.003, R2=0.035) all negatively predicted working memory response time in college students. Physical activity level (Beta=-0.234, P<0.001, R2=0.055), medium intensity physical activity level (Beta=-0.159, P=0.012, R2=0.025), and low intensity physical activity level (Beta=-0.211, P=0.001, R2=0.044) all negatively predicted college students' shifting function effects, while the regression coefficient for the high-intensity physical activity level (Beta=-0.123, P=0.053, R2=0.015) was not significant. Physical activity level (Beta=-0.205, P=0.001, R2=0.042), high intensity physical activity level (Beta=-0.144, P=0.024, R2=0.021), medium intensity physical activity level (Beta=-0.133, P=0.037, R2=0.018), and low intensity physical activity level (Beta=-0.131, P=0.039, R2=0.017) all negatively predicted the amount of inhibition function effects in college students, as presented in Figure 4.
2.4 Construction and Validation of a Structural Relationship Model of Physical Activity Level, Executive Function, and Trait Anxiety in College Students
To examine the degree of influence of physical activity level and executive function on trait anxiety and to explore the feasibility of a structural relationship model, a multiple linear regression analysis (entry method) was conducted with trait anxiety score as the dependent variable and physical activity level, inhibition function, working memory, and shifting function as independent variables. The results indicated that the regression model passed the significance test (F(4, 243)=4.882, p=0.001 , R2=0.075), and physical activity level (Beta=-0.156, p=0.018), inhibition function (Beta=0.147, p=0.038), and shifting function (Beta=0.135, p=0.038) for the trait anxiety in college students were all significant predictors, while the regression coefficient for working memory (Beta=-0.081, P=0.261) was not significant. The VIF values of each independent variable were all <5, and the effect of multicollinearity could be largely excluded from the results of this study. As detailed in Table 2.
Table 2 Multiple Linear Regression Analysis of Influencing Factors of Trait Anxiety among College Students
Dependent variable
|
Independent variables
|
B
|
SE
|
95%CI
|
Beta
|
t-value
|
VIF value
|
F-value
|
Model Abstract
|
Lower
|
Upper
|
Trait anxiety
|
Constant
|
40.264
|
4.520
|
31.450
|
49.129
|
|
8.909***
|
|
4.882**
|
R2=0.075
|
|
Physical activity level
|
-0.001
|
0.000
|
-0.002
|
0.000
|
-0.156
|
-2.373*
|
1.138
|
|
adR2=0.059
|
|
Inhibition function
|
0.030
|
0.014
|
0.003
|
0.059
|
0.147
|
2.090*
|
1.296
|
|
|
|
Working memory
|
-0.008
|
0.007
|
-0.023
|
0.007
|
-0.081
|
-1.126
|
1.349
|
|
|
|
Shifting function
|
0.009
|
0.004
|
0.001
|
0.018
|
0.135
|
2.106*
|
1.078
|
|
|
To test for bias in common methods, a confirmatory factor analysis of the original entries and executive function performance of the International Physical Activity Questionnaire and Trait Anxiety Inventory was conducted using the Harman one-way test. Eight factors with characteristic roots greater than 1 were obtained, and the variance explained by the first factor was 23.11%, which was much less than the critical value of 40%. Therefore, the effect of common method bias could be largely excluded from the results of this study.
The research hypotheses were tested based on the interrelationship of physical activity level, executive function, and trait anxiety in college students. Model 1 was established with physical activity level as the independent variable, executive function as the mediating variable, and trait anxiety as the dependent variable, followed by three sub-models with high, medium, and low intensity physical activity levels as the independent variables, respectively. The paths with insignificant coefficients were removed one by one, and the path coefficients were recalculated until all passed the Bootstrap significance test. CMIN/df = 1.985 for model 1, indicating a good model fit, fully meeting the reference criteria of RMR < 0.05, RMSEA < 0.08, and GFI, NFI, and CFI values > 0.9, indicating a reasonable and reliable structural equation model. The CMIN/df of the three sub-models were all <3, and each goodness-of-fit index basically met the standard. The path analysis is shown in Figure 5, and the results of the mediating effect test are shown in Table 3.
Physical activity level had a 72.31% direct effect on reducing trait anxiety (B=-0.195, 95% bootstrap CI: -0.313, -0.082), with the mediating effect of inhibition function accounting for 11.79% (B=-0.023, 95% bootstrap CI: -0.067, -0.001) and the mediating effect of shifting function accounting for 15.90% (B=-0.031, 95% bootstrap CI: -0.076, -0.004). High intensity physical activity level had a direct effect on the reduction of trait anxiety of 81.08% (B=-0.150, 95% bootstrap CI: -0.290, -0.040), with the mediating effect of inhibition function accounting for 9.19% (B=-0.017, 95% bootstrap CI: -0.058,0) and the mediating effect of shifting function of 9.73% (B=-0.018, 95% bootstrap CI: -0.054, -0.002). The effect of medium-intensity physical activity level on trait anxiety was fully mediated by executive function, with the mediating effect of inhibitory function accounting for 40.91% (B=-0.018, 95% bootstrap CI: -0.054, -0.001) and the mediating effect of shifting function accounting for 56.82% (B=0.025, 95% bootstrap CI: - 0.064, -0.005). The effect of low-intensity physical activity level on trait anxiety was exclusively mediated by executive function, with 34.62% of the mediating effect mediated by inhibitory function (B=-0.018, 95% bootstrap CI: -0.052, -0.002) and 65.38% of the mediating effect mediated by shifting function (B=-0.034, 95% bootstrap CI: -0.076, -0.008). Different intensity levels of physical activity effectively improved working memory (B=-0.187 to -0.208) but working memory did not mediate between physical activity level and trait anxiety.
Table 3 Results of Bootstrap Test for Mediating Effects
Model
|
Types of effects
|
Effect value B
|
Bootstrap SE
|
Bias-Corrected95%CI
|
Percentile95%CI
|
Percentage of effects
|
Goodness of fit of the model
|
Model 1
|
Total effect
|
-0.195**
|
0.059
|
[-0.313,-0.082]
|
[-0.314,-0.083]
|
100%
|
CMIN/df=1.985
RMR=0.042
RMSEA=0.063
GFI=0.990
NFI=0.951
CFI=0.973
|
|
Direct effect
|
-0.141*
|
0.062
|
[-0.263,-0.019]
|
[-0.264,-0.020]
|
72.31%
|
|
Mediating effect of inhibition function
|
-0.023*
|
0.016
|
[-0.067,-0.001]
|
[-0.060,-0.001]
|
11.79%
|
|
Mediating effect of shifting function
|
-0.031*
|
0.017
|
[-0.076,-0.004]
|
[-0.069,-0.001]
|
15.90%
|
Model 2
|
Total effect
|
-0.185**
|
0.066
|
[-0.326,-0.070]
|
[-0.332,-0.073]
|
100%
|
CMIN/df=2.892
RMR=0.054
RMSEA=0.088
GFI=0.986
NFI=0.916
CFI=0.939
|
|
Direct effect
|
-0.150**
|
0.064
|
[-0.290,-0.040]
|
[-0.290,-0.040]
|
81.08%
|
|
Mediating effect of inhibition function
|
-0.017*
|
0.014
|
[-0.058,0]
|
[-0.051,0.002]
|
9.19%
|
|
Mediating effect of shifting function
|
-0.018*
|
0.013
|
[-0.054,-0.002]
|
[-0.048,0]
|
9.73%
|
Model 3
|
Total effect
|
-0.044**
|
0.019
|
[-0.091,-0.014]
|
[-0.087,-0.012]
|
100%
|
CMIN/df=2.496
RMR=0.057
RMSEA=0.078
GFI=0.984
NFI=0.898
CFI=0.932
|
|
Direct effect
|
|
|
|
|
|
|
Mediating effect of inhibition function
|
-0.018*
|
0.013
|
[-0.054,-0.001]
|
[-0.049,0]
|
40.91%
|
|
Mediating effect of shifting function
|
-0.025*
|
0.014
|
[-0.064,-0.005]
|
[-0.059,-0.002]
|
56.82%
|
Model 4
|
Total effect
|
-0.052***
|
0.021
|
[-0.105,-0.019]
|
[-0.097,-0.017]
|
100%
|
CMIN/df=1.741
RMR=0.049
RMSEA=0.055
GFI=0.989
NFI=0.931
CFI=0.967
|
|
Direct effect
|
|
|
|
|
|
|
Mediating effect of inhibition function
|
-0.018*
|
0.012
|
[-0.052,-0.002]
|
[-0.047,0]
|
34.62%
|
|
Mediating effect of shifting function
|
-0.034**
|
0.017
|
[-0.076,-0.008]
|
[-0.071,-0.006]
|
65.38%
|