For reliable papers only

HW #11 – Correlation Name_____________________________

Open the HW#11 data set (BodyFat.sav) found in the Week by Week folder

The Scenario is as follows:

Direct measurement of percent body fat is not a simple task. It has been proposed that there is a correlation between subcutaneous fat and total percent body fat. A group of researchers decided to test this theory in a pilot study with the ultimate goal of developing a new predictor model that would estimate body fat based upon subcutaneous fat readings or limb circumferences. (Please excuse these researchers. They have been living under a rock and haven’t kept up on developments in the world of body fat assessment.)

Twenty females ranging in age between 25 and 30 years old were randomly selected as subjects for this study. All subjects underwent direct measurement of percent body fat. Additionally, all were measured for circumference or subcutaneous fat deposits at three locations; triceps, thigh, and mid-arm. Measurements were recorded in mm.

The researchers wanted to assess how well each fat measurement correlated with the actual percent body fat.

The researchers then wanted to select two of the circumference or subcutaneous fat measurements to include in the model that they hoped to construct.

Part 1: Determine Linearity, Normality, and presence/absence of Outliers.

Cut and paste your graphs and or tests into a Word document in the order above.

Answer each of the following:

1. Were the variables all linear?

2. If they were, not, which one(s) were not linear?

3. Were the variables all normally distributed?

4. If they were, not, which one(s) were not normally distributed?

5. Were there any outliers present?

6. If outliers were present what would you do to modify them?

*To assess data for Linearity and Monotonicity make a scatter dot plot for each variable pair and visually assess.

**To assess data for normality, do so as always with tests and graphs.

***To assess for outliers do so with the box plots in the Normality test output

Part 2: Using unmodified data, run a PPM and a Spearman Correlation on the following variable pairs:

a. % Body fat and Thigh circumference

b. % Body fat and Triceps measurement

c. % Body fat and Mid-arm measurement

(Your output will include more than one Correlation coefficient.)

Cut and Paste the Correlation Tables from your SPSS Output here.

1. Fill in the following table with the associated correlation coefficients:

	%Body Fat vs Tricepts	% Body Fat vs Thigh	%Body Fat vs Midarm
Pearson
Spearman

2. Compare the Pearson vs Spearmen correlation coefficients obtained for each circumference location.

a. Why were the coefficients different even though you used the same data?

Part 3: Compose a COMPLETE write-up paragraph of the research using the Pearson Correlation as the main Test.

Include all of the statistical information obtained including assumption tests.

Part 4: Answer the following questions:

1. What 2 variables would be the best to include in a model to predict % body fat?

2. Why did you select these as the 2 variables to include in the model?

SPSS Directions:

Pearson Product Moment and Spearman correlation:

Be sure to check for linearity, normality, and outliers before proceeding

1. Click Analyze > Correlate > Bivariate... on the main menu

2. You will be presented with the Bivariate Correlations dialogue box:

3. Highlight all variables at the same time by clicking on each while holding down the shift-key.

4. Transfer these variables into the Variables: box by clicking on the arrow.

5. Click on Pearson and Spearman in the Correlation Coefficients box

6. Click on Two-tailed in the Test of Significance box

7. Click on the box for Flag significant correlations

8. Click on Options

9. Click on Means and Standard Deviations

10. Click on Exclude cases pairwise

11. Click on Continue

12. Click on OK