Critically evaluate the statistical material contained in this paper in relation to items 10-12-17 of the STROBE check-list.
The following research question can be addressed using the data set that you have been assigned for your Assignments:
Can the logarithm from MVPA predict GPA after correcting the overweight population of Australian university students
Please give your opinion on the strengths and weakness of the presentation in Fox et.al. 2010, against items 10-12-17 from STROBE
Researchers used 4746 students from middle and high schools as their sample.
The method of sampling isn’t clear.
Researchers did not mention the methods they used for sampling, such as simple random, cluster sampling, systematic sampling, and stratified sampling.
The methods of sampling should be clearly explained to enable respondents to determine if the results can be generalized.
I think the researchers did a great job in providing demographic information about the respondents.
For example, they gave information about the proportion of males and females in each age group, as well as the ethnicity and socio-economic status for the students.
The researchers did not mention how they control for any confounding that might have arisen during the course the study.
To avoid biased results, the reader should assume that there weren’t any confounders that could have been controlled.
Researchers didn’t mention or highlight whether they were having issues with missing data.
Missing data is also called missing values. It occurs when data for one or many variables in an observation are not stored.
Missing data in research can have a major impact on the results and conclusions.
The researchers performed parametric tests to perform inferential analyses when it came time for analysis.
The missing element is the test of the necessary assumptions for the parametric testing.
Parametric tests must be distribution-free. This means that they must adhere to some assumptions. Any violation of any assumption could lead to biased results.
Parametric tests can be used to test for linearity, normality, homogeneity, independence, and equal variances assumptions.
Present the results of your descriptive analysis
This question was asked to determine if the logarithms of MVPA can predict GPA after correcting overweight among Australian university students.
The summary statistics of both variables was first looked at.
The logMVPA was found to have a mean of 0.44 and a median of 0.52, which is slightly higher than its average.
The minimum value was 0.52, while the maximum was 1.22. There was also a range of 1.75.
The skewness and kurtosis values are both negative (-0.49, -0.46), respectively. This could be interpreted as negatively skewed data.
Refer to figure 1.
Figure 1: LogMVPA Histogram
Average GPA scores for students were 4.76, with a median score at 4.7.
The highest GPA score of this group was 6.9, while the lowest was 2.4.
The summaries also included the values of skewness, kurtosis and -0.08 respectively.
This suggests that data was likely to have been taken from a normally distributed collection, as the skewness is close to zero.
Figure 2 shows this idea.
Below is the histogram for GPA.
It is clear from the figure that the GPA variable data is normal distributed (bell-shaped curvature).
Figure 2: GPA histogram
Present the relevant regression models and inferential analysis findings (approximately 150-200 words, 10 marks).
We used linear regression models and Pearson correlation tests to determine if logMVPA accurately predicts GPA.
A Pearson correlation test was performed and the coefficient between GPA (GPA) and logMVPA was 0.6648. This means that there is a moderately positive relationship between GPA/logMVPA.
A scatter plot of GPA and logMVPA is shown to account for the overweight
Figure 3: Scatter plot for GPA and logMVPA
The graph clearly illustrates the linear positive relationship between GPAs and logMVPAs.
To predict the GPA score, a regression analysis was conducted using logMVPA (logMVPA).
Two key elements of the model’s fitness were taken into consideration: the coefficient of determination and the significance value.
We first looked at the model’s goodness of fit. The model can accurately predict GPA based upon logMVPA at the 5% level (p > 0.05).
The coefficient of determination (Rsquared) of the model is 0.4242. This means that 44.2% variance in the dependent variable (GPA), is explained by logMVPA in the model.
The model also showed significant logMVPA (p 0.050). Its coefficient is 1.6256. This means that one unit of logMVPA will result in a change of the GPA to 1.6256.
LogMVPA would increase by one unit. This would result in a rise in the GPA by 1.6256.
The GPA would decline by 1.6256 if logMVPA falls by one unit.
The constant intercept was 4.03387
Your answer to the research question
This study examined whether logMVPA (logMVPA) can predict GPA by correcting for overweight in Australian university students.
This question was answered by regression models.
When overweight is controlled, the logMVPA predicts GPA.