SPSS Data Analysis

We used logistics regression analysis to analyse our research question to see if there is a relationship
between the independent variable and the dependent variable. From our table, well-being and health
perception is indicated by the ‘distress’ variable which captures the overall distress levels of each
participant. Note that for this analysis, we converted the height recorded for each participant to a value in
meters. Since Body Mass Index (BMI) is one of the required covariant for our analysis, we calculated this
value with the formula:
weight (kg)
BMI = 2
height ( m )
Research Question 1: "Is walking associated with well-being and health perception?"
Descriptive Statistics
To understand the relationship between well-being and health perception and walking in addition to other
covariates such as age, gender, Race/ethnicity, education, BMI, sleep duration and comorbidities, we
carry out the logistics regression analysis. For this specific analysis, 88.1% of cases were included, and
11.9% were missing.
Table 1: Frequency Table

Frequency
EDUCATION Less than high school 9
High school or equivalent 12
Some college or trade school 40
Bachelor's degree 30
Master's degree or higher 25
Associate degree 17
RACE_ETHNICITY Asian 9
Black/African American 23
Hispanic or Latino 33
White 57
2+ races 9
Other 2
WALKING No days 2
Less than 1 day 24
1 to 2 days 28
2 to 4 days 24
 4 to 6 days 55
MARITAL_STATUS Single 21
Married 98
Divorced 6
Widowed 5
Separated 3
BMI Underweight 27
Healthy weight 21
Overweight 19
Obese 66
SLEEP_DURATION 3 or less hours of sleep per night. 11

4-6 hours of sleep per night. 75

7-8 hours of sleep per night. 44
9 hours 3
COMORBIDITIES NO 24
YES 109
GENDER Female 108
Male 25

Result
The omnibus tests in Table 2 assess the overall significance of the model coefficients. In this analysis, the
chi-square statistic for the model was 30.800 with 28 degrees of freedom, resulting in a p-value of 0.326.
Since the p-value > α, this suggests that the model as a whole, is not statistically significant at the 0.05
significance level.
Table 2: Omnibus Tests of Model Coefficients

Chi-square df Sig.

Step 1 Step 30.800 28 .326

Block 30.800 28 .326

Model 30.800 28 .326

Table 3 presents the model summary, and from this table the Cox & Snell R Square was found to be
0.207, indicating that approximately 20.7% of the variance in the dependent variable was explained by
the model. The Nagelkerke R Square is more accurate in presenting the model summary since it adjusts
for sample size. There are 37.8% of the variations in the output.
Table 3: Model Summary

Step -2 Log likelihood Cox & Snell R Square Nagelkerke R Square

1 74.645a .207 .378

The classification table in table 4 shows what we observed and what would be predicted using our model
for the Distress variable. This model correctly predicts that 5 participants who were not distressed were
actually not distressed. But it incorrectly predicts that 13 participants (who were actually not distressed)
were distressed. It got about 27.8% of this prediction correctly. Likewise, our model correctly predicted
that 114 of the participants was distressed, but it was incorrect in only one case, thus giving it a correct
percentage score of 99.1% Overall, 89.5% of cases were correctly classified.

Table 4: Classification Table

Predicted

DISTRESS
Percentage
Observed Not in distress Distress Correct

Step 1 DISTRESS Not in distress 5 13 27.8

Distress 1 114 99.1

Overall Percentage 89.5

Table 5 shows us which variable has a significant effect on the model and which one doesn’t. We can see
that significant variables have coefficients with 95% confidence intervals that do not include zero and P-
values > α, and this suggests that their impact on the well-being and health perception of each participant
is not statistically significant. The table shows that there is no association between walking and well-
being and health perception.
The sign of the coefficient (B) indicates the direction of the relationship, and the magnitude indicates the
strength. For example, being divorced, having comorbidities, and having fewer hours of sleep per night
are associated with higher odds of distress. The negative coefficient in Age suggests a decrease in the
odds of distress with increasing age. Again, from the table, we can see that being female is associated
with significantly higher odds of distress compared to males.
Table 5: Variation In the Equation
Frequency 95% C.I.for EXP(B)
B S.E. Sig. Lower Upper
WALKING No days .697
Less than 1 day -.607 29079.142 1.000 .000 .
 1 to 2 days 1.564 1.068 .143 .589 38.776
2 to 4 days .418 .920 .649 .250 9.218
4 to 6 days .581 1.130 .607 .195 16.373
AGE -.729 .517 .158 .175 1.328
GENDER 1.589 .754 .035 1.117 21.496
RACE_ETHNICIT Asian .891
Y
Black/African American -22.969 26437.801 .999 .000 .
Hispanic or Latino -23.014 26437.801 .999 .000 .
White -23.572 26437.801 .999 .000 .
2+ races -22.198 26437.801 .999 .000 .
Other -3.485 29577.896 1.000 .000 .
EDUCATION Less than high school .498
High school or
-1.261 1.835 .492 .008 10.331
equivalent
Some college or trade
1.473 1.937 .447 .098 194.078
school
Bachelor's degree -.547 1.443 .705 .034 9.788
-.771 1.447 .594 .027 7.883
Associate degree -1.325 1.419 .350 .016 4.290
MARITAL_STAT Single .827
US
Married -.423 1.965 .830 .014 30.850
Divorced .338 1.658 .838 .054 36.125
Widowed -1.411 2.110 .504 .004 15.238
Separated .032 2.183 .988 .014 74.431
BMI Underweight .363
Healthy weight .481 .964 .618 .245 10.702
Overweight 2.230 1.386 .108 .615 140.799
Obese -.335 1.054 .751 .091 5.647
COMORBIDITIES NO .953
YES -18.141 40192.669 1.000 .000 .
 SLEEP_DURATIO 3 or less hours of sleep per
.354
N night.
4-6 hours of sleep per
-.695 21542.285 1.000 .000 .
night.
7-8 hours of sleep per
-20.625 18903.824 .999 .000 .
night.
9 hours -21.938 18903.824 .999 .000 .
Constant 65.871 51689.052 .999
Conclusion
A total of 151 cases were initially included in the analysis. After accounting for missing data, 133 cases
(88.1%) were included in the final analysis. The non-significant omnibus test suggests that the dependent
variables such as age, gender, Race/ethnicity, education, BMI, sleep duration and comorbidities did not
significantly predict the well-being and health perception of the participants. Table 5 suggests that there is
no relationship between walking and well-being and health perception.
Research Question 2: " Is sedentariness associated with well-being and health perception?"
Descriptive Statistics
The frequency distribution in Table 6 provides an overview of the participants' reported sedentary
behavior in terms of the hours spent sitting during the last 7 days. This distribution allows us to observe
the variability in sedentary behavior among the study participants. these categories will be explored to
determine if there is a significant association between the reported sedentary behavior and the
participants' overall well-being and health perception scores.
Table 6: Summary Statistics

Frequency
SEDENTARY 1 hour or less 25
1 to 2 hours 21
2 to 4 hours 32
4 to 6 hours 23
6 + hours 20
Result
From Table 7, it is evident that both the Step Chi-square and Block Chi-square tests yield a significance
level of 0.191, suggesting that, collectively, the predictor variables in the model may not be statistically
significant in explaining the variability in well-being and health perception scores. The -2 Log likelihood
in Table 8 represents the model's fit, with lower values indicating better fit. The model accounts for 24%
of the variance according to Cox & Snell and 43.1% according to Nagelkerke.
Table 7: Omnibus Tests of Model Coefficients

Chi-square df Sig.

Step 1 Step 33.179 27 .191

Block 33.179 27 .191

Model 33.179 27 .191

Table 8: Model Summary

Step -2 Log likelihood Cox & Snell R Square Nagelkerke R Square

1 65.040a .240 .431

Table 9 indicates that the logistic regression model correctly predicted the Distress status for 88.4% of
cases. However, it's important to note that there were misclassifications, especially for participants who
were not in distress, where 29.4% of predictions were correct. The model performed exceptionally well in
predicting participants experiencing distress, with a correct prediction rate of 98.1%. These results
highlight the trade-off between sensitivity and specificity in logistic regression models. Sensitivity refers
to the ability to correctly identify true positives (participants in distress), while specificity refers to the
ability to correctly identify true negatives (participants not in distress).
Table 9: Classification Table

Predicted

DISTRESS
Percentage
Observed Not in distress Distress Correct

Step 1 DISTRESS Not in distress 5 12 29.4

Distress 2 102 98.1

Overall Percentage 88.4

Table 10 shows us which variable has a significant effect on the model and which one doesn’t. The
overall variable for sedentariness and for sleep duration is not statistically significant, suggesting that
sedentariness and sleep duration does not significantly predict distress. The coefficient for age is -0.451,
which is not statistically significant (p = 0.416). This suggests that age does not significantly predict the
likelihood of experiencing distress. The coefficient for gender (1 = Female) is 1.880, with a significant p-
value of 0.026. This indicates that being female is associated with higher odds of distress compared to
males. The overall race and ethnicity variable is not statistically significant (p = 0.890), suggesting that,
collectively, these categories do not significantly predict distress.
Table 10: Variations
95% C.I.for EXP(B)
B Sig. Exp(B) Lower Upper
AGE -.451 .416 .637 .215 1.887
GENDER 1.880 .026 6.557 1.252 34.347
RACE_ETHNICIT Asian .890
Y
Black/African American -21.526 .999 .000 .000 .
Hispanic or Latino -22.277 .999 .000 .000 .
White -22.435 .999 .000 .000 .
2+ races -21.129 .999 .000 .000 .
Other -2.962 1.000 .052 .000 .
EDUCATION Less than high school .659
High school or equivalent -2.258 .269 .105 .002 5.751
Some college or trade 1.252 .630 3.497 .021 570.303
school
 Bachelor's degree -.096 .952 .909 .040 20.720
Master's degree or higher -.332 .838 .718 .030 17.271
Associate degree -1.139 .480 .320 .014 7.524
MARITAL_STAT Single .815
US
Married -.017 .993 .983 .027 36.015
Divorced .872 .570 2.391 .118 48.534
Widowed -.547 .793 .579 .010 34.141
Separated -.965 .716 .381 .002 68.771
BMI Underweight .983
Healthy weight .322 .742 1.380 .203 9.385
Overweight 1038432899.25
20.761 .998 .000 .
3
Obese -.159 .880 .853 .110 6.639
COMORBIDITIES
.327 .747 1.386 .190 10.103
(1)
SLEEP_DURATI 3 or less hours of sleep per
.531
ON night.
4-6 hours of sleep per
.765 1.000 2.148 .000 .
night.
7-8 hours of sleep per
-19.199 .999 .000 .000 .
night.
9 hours -20.399 .999 .000 .000 .
SEDENTARY 1 hour or less .656
1 to 2 hours 2.512 .158 12.333 .377 403.987
2 to 4 hours .401 .732 1.493 .151 14.747
4 to 6 hours .246 .822 1.279 .151 10.841
6 + hours .061 .958 1.063 .109 10.342
Constant 5642895223902
43.177 .999
121000.000
Conclusion
The model's omnibus test indicates that, collectively, the predictors do not significantly predict distress (p
= 0.191). The interpretation of the results suggests that sedentariness and other factors such as age,
education, marital status, BMI, comorbidities, and sleep duration do not significantly predict the well-
being and health perception of the participants. Hence, we can conclude that there is no association
between sedentariness and well-being and health perception.
Research Question 3: "Is vigorous physical activity associated with well-being and health
perception?"
Descriptive Statistics
For this logistic regression analysis, we are investigating the association between vigorous physical
activity (Vigorous PA) and the well-being and health perception of participants. The descriptive statistics
in Table 11 provide a breakdown of participants based on their reported engagement in vigorous physical
activities. We can see from the table that most participants engaged in vigorous physical activities for less
than 1 day, while 8 reported not engaging in such activities at all.
Table 11: Summary Statistics

Frequency
VIGOROUS No days 8
Less than 1 day 78
1 to 2 days 21
2 to 4 days 8
4 to 6 days 19
Result
From Table 12, the omnibus tests suggest a lack of collective significance of predictor variables since p-
value > α (i.e 0.05). Table 13 assesses the goodness-of-fit of the logistic regression model, and the results
suggest that the logistic regression model, including the predictor variable Vigorous PA, shows some
degree of significance in predicting the well-being and health perception of the participants.
Approximately 44.2% of the variability in the outcome variable (well-being/health perception in your
case) is explained by the combination of predictor variables included in the model.
Table 12: Omnibus Tests of Model Coefficients

Chi-square df Sig.

Step 1 Step 37.002 27 .095

Block 37.002 27 .095

Model 37.002 27 .095

Table 13: Model Summary

Step -2 Log likelihood Cox & Snell R Square Nagelkerke R Square

1 68.732a .241 .442

For participants categorized as "Not in distress" in Table 14, the model correctly predicted 7 out of 18
cases, achieving a correct prediction percentage of 38.9%. For participants categorized as "Distress," the
model correctly predicted 114 out of 116 cases, achieving a very high correct prediction percentage of
98.3%. Overall, the model achieved an accuracy rate of 90.3%, indicating the percentage of correctly
classified cases across both distress categories.
Table 14: Classification Table

Predicted

DISTRESS
Percentage
Observed Not in distress Distress Correct

Step 1 DISTRESS Not in distress 7 11 38.9

Distress 2 114 98.3

Overall Percentage 90.3

Table 15 suggests that Vigorous PA and other factors such as Age, Gender, Education, BMI, and Sleep
Duration have no significant effects on predicting well-being and health perception, and this conclusion is
based on the p-values (which all have p-value > 0.05). Positive coefficients such as Vigorous PA and
Gender suggest a positive association with the likelihood of distress, while the negative coefficients (e.g.,
Age, Education, Sleep Duration) suggest a negative association with the likelihood of distress.
Table 15: Variations
95% C.I.for EXP(B)
B Sig. Exp(B) Lower Upper
AGE -.835 .112 .434 .155 1.216
GENDER 1.407 .075 4.084 .866 19.261
RACE_ETHNICIT Asian .914
Y
Black/African American -20.729 .999 .000 .000 .
Hispanic or Latino -20.958 .999 .000 .000 .
White -21.901 .999 .000 .000 .
2+ races -20.755 .999 .000 .000 .
Other -2.264 1.000 .104 .000 .
EDUCATION Less than high school .753
High school or equivalent -.288 .888 .750 .014 41.327
Some college or trade
.568 .769 1.764 .040 77.976
school
Bachelor's degree -.519 .707 .595 .040 8.927
-1.071 .443 .343 .022 5.278
Associate degree -1.473 .294 .229 .015 3.585
MARITAL_STAT Single .874
US
Married -1.017 .576 .362 .010 12.712
Divorced .167 .913 1.181 .059 23.703
Widowed -1.057 .609 .348 .006 20.028
Separated -.290 .894 .748 .011 52.932
BMI Underweight .232
Healthy weight -.174 .856 .840 .127 5.536
Overweight 2.779 .045 16.103 1.069 242.524
Obese .811 .477 2.251 .241 21.025
COMORBIDITIES
-1.022 .275 .360 .058 2.251
(1)
SLEEP_DURATI 3 or less hours of sleep per
.379
ON night.
4-6 hours of sleep per
-.034 1.000 .967 .000 .
night.
7-8 hours of sleep per
-19.061 .999 .000 .000 .
night.
9 hours -20.393 .999 .000 .000 .
VIGOROUS No days .172
Less than 1 day 20.344 .999 684403326.665 .000 .
1 to 2 days 2.552 .017 12.833 1.566 105.182
2 to 4 days 2.641 .051 14.031 .984 200.108
4 to 6 days .922 .568 2.514 .106 59.573
Constant 1679542334601
44.268 .999
3325000.000

Conclusion
The logistic regression results with p-value = 0.095 suggest that the relationship between vigorous
physical activity and well-being/health perception is statistically not significant. A Nagelkerke R Square
of 0.442 was obtained in the result, which suggests that the model, including age, gender, race/ethnicity,
education, marital status, BMI, comorbidities, sleep duration, and vigorous physical activity, explains a
substantial portion of the variation in distress.