Image 1: data results part 1
Image 2: data results part 2
Image 3: data results part 3
Image 4: variable view
Image 5: Value Labels for Gender
1 = “male”
2 = “female”
Image 6: Value Labels for if respondent has been exercising regularly in the last 4 weeks (Frequency of exercise)
1 = “yes”
2 = “no”
Image 7: Value labels for the form of exercise most engaged in by the respondent (Type of exercise)
1 = “Aerobic”
2 = “Anaerobic”
3 = “Mixed”
Image 8: Value labels for if respondent exercises with company
1 = “alone”
2 = “group”
Image 9: Value labels for respondent’s Reason/ Motivation to exercise
1 = “appearance”
2 = “health”
3 = “peers”
4 = “others”
Image 10: Value labels if respondent is considered to have “exercised” based on our multivariate conceptualization of “Exercise”
1 = “yes”
2 = “no”
Because our question is a correlation question and the datatype of our dependent variable (Rosenberg Self-Esteem Scale Score) is scale, we propose Pearson’s product moment correlation coefficient as the appropriate statistical technique to test our H0.
Having chosen the appropriate statistical test to analyse our data, we then verify its assumptions.
Assumption 1: All observations must be independent of each other.
We assure respondents of their confidentiality so that they do not feel like they have to provide a set of data that “impresses” us the interviewers. As such, they should not feel a need to alter the information they are providing even if they interacting with another respondent with close physical proximity whilst either are doing the survey.
Assumption 2: The dependent variable should be normally distributed at each value of the independent variable.
Image 11: Summary Statistics for Duration of exercise (“mins in one week”) and Rosenberg Self-Esteem Scale Scores
Note that for both variables, the statistical value of Skewness is less than twice the standard error value for Skewness, indicating that both variables are normally distributed and thus, there is no violation of normality assumption.
Assumption 3: The dependent variable should have the same variability at each value of the independent variable.
Assumption 4: The relationship between the dependent and independent variables should be linear.
Assumption 3 and 4 are inspected by examining the scatter plot.
Image 12: Scatter Plot between the Duration of exercise (Total mins in one week) and the Rosenberg Self-Esteem Scale Score with the best fit line and r² of 0.046
As seen in Image 12, the distribution of the data is too widespread to observe any conspicuous form/ pattern. Hence, it is concluded that the scatter does not appear to follow any general trend and there the linearity assumption (Assumption 4) is violated.
It can also be observed that the variability or spread of the dependent variable (ie. the Rosenberg Self-Esteem Scale Scores) for each corresponding value of the independent variable (ie. the Total mins in one week) from the proposed best fit line is not remotely similar. Therefore, Assumption 3 has also violated.
Since the assumptions of the Pearson’s product moment correlation coefficient have been violated, we will then have to use a non-parametric test, the Spearman’s Rank Correlation in particular, to analyze our data.
Image 13: Results of the Spearman’s Rank Correlation
There is a positive, weak and insignificant association between exercise and self-esteem. (r = 0.258, p = 0.104 [> 0.05], N = 41)
Based on the Spearman’s correlation, p=0.104 and α=0.05. Since p>α the study fails to reject H0 and there is insufficient evidence to suggest that H1 is correct at the 5% error level and conclude that there is no significant relationship between the duration of moderate- to vigorous-intensity exercise and a person’s self-esteem.
Further Analysis : Extraneous variables
- Alone vs Group
Although our results were not significant, we can analyze from this scatterplot that most of the people who exercise alone fall on the left half of the graph whereas those who exercise in a group are more spread out across the X-axis(minutes in one week), to the right side of the graph, which can imply that social factors can affect the duration of exercise. Those who exercise with others, exercise for longer duration in one week as compared to those who ran alone. This is in line with our literature review (Carnes, 2014) that states the presence of company increases the amount of time that participants run. Thus, this shows that social factors do play a part in the duration of exercise. However, we are unable to see the trend from this graph that exercising with others increases self-esteem. This is shown in the graph as the strength of both of the best fit lines (alone and group) were weak, alone r^2 linear =0.005, group r^2 linear =0.198.
- Gender
From this graph, we can see that females generally exercise for a shorter duration as compared to males. Generally, most of the females exercise less than 300 minutes in one week. Whereas for males, there is a wider range from 90 to 600 minutes per week. Generally, the females also have a lower Rosenberg scale as compared to the males. This led to a positive relationship for females and a negative relationship for males. Research has shown that gender have different responses to exercise. (McKelvie, 2005), (Ivey, et al., 2000). This may have an impact on the self esteem of the individual, leading to the differences in the relationship in males and females. Gender is hence an extraneous variable and may have possibly affected our overall results. More research needs to be done on our part on the difference of the level of self-esteem on gender.
- Motivations for Exercise
Out of the 4 options given for motivations to exercise, approximately 46.3% of respondents that were exercising put “health” as their primary reason. From the distribution, it can be observed that most of the data points cluster in the lower left corner, specifically between the Rosenberg self-esteem scale score between 15 and 20 (the normal range is between 15 and 25) and less than a total of 200 mins of exercise in week. This is surprising, as the results seem to suggest that these respondents only “felt average” about themselves even though there can be considered to be exercising (44.6% of the respondents were classified as not exercising). This indicates that at least for this sample, exercise is regarded as a daily activity and is not considered to add “worth” to one’s value. Taking into consideration that these respondents are exercising for the sake of health, it could be assumed that these respondents feel that exercise is a necessity in maintaining health and they are not seeking other benefits to exercise other than to simply maintain health/fitness, which accounts for the cap at a total of 200 mins of exercise in a week. As for the other 3 reasons, the data points are too few and widely distributed to be subjected to interpretation and generation of a trend.
- Types of Exercise
It can be observed that there is an almost equal proportion of people involved in aerobic (n = 15), anaerobic (n = 12) and mixed (n = 14). Of the three types of exercise, “aerobic” has the most conspicuous distribution, particularly in the Rosenberg self-esteem scale scores between 15 and 25 (the normal range) and less than a total of 200 mins of exercise in one week. In addition, when compared to the other types of exercise, the data points for “aerobic” exercise tended to cluster in the normal range of the Rosenberg self-esteem scale and towards the leftmost side, the lower end of the spectrum for total minutes spent exercising in a week. These results seem to suggest that compared to the “anaerobic” and “mixed” types of exercise, doing “aerobic” exercise had the highest likelihood in making respondents “feel adequate” and for lesser duration invested as well.
Exercise and Self-Esteem 