Using Minitab to Construct Linear Regression Models

Scatter Plot

To obtain a simple scatter plot using Minitab:

One of the first steps in the analysis of the study data is to create a scatter plot that compares the quantitative variables. Create each of the following scatter plots, find Pearson's correlation coefficient, and perform the corresponding linear regression analysis in each case. Detailed instructions follow on the next page.

1. Create a scatter plot, find the r value, and perform the regression analysis that compares the patients' average daily blood sugar level to their Hemoglobin A1C level.

2. Create a scatter plot, find the r value, and perform the regression analysis that compares the patients' carbohydrate intake to average glucose levels.

3. Create a scatter plot, find the r value, and perform the regression analysis that compares the patients' carbohydrate intake to Hemoglobin A1C levels.

To Obtain a Sample Scatter Plot Using Minitab
1. Open the HealthCareData.mpj file using Minitab.
2. From the menus, select Graph, Scatterplot

3. Choose Simple and then click OK.

4. Select the following variable pairs for the scatter plots
a. y-variables: Hemoglobin x-variables: Glucose
b. y-variables: Glucose x-variables: carb_intake
c. y-variables: Hemoglobin x-variables: carb_intake

5. Generate all three scatter plots by clicking OK and examine the results. Copy and paste each graph below.

To Obtain Linear Regression Using Minitab
6. From the menus, select Stat, Regression and then General Regression

7. Select the following variable pairs for the regression analysis (you will need to do this 3 separate times).
Response = y-axis andModel = x-axis
a. Response: Hemoglobin Model: Glucose
b. Response: Glucose Model: carb_intake
c. Response: Hemoglobin Model: carb_intake

8. Then click Graphs, select Standardized and select Histogram and Normal probability plot and click OK.

9. Click OK in the Linear Regression window to perform the analysis.

10. Think about it: Were there any strong relationships indicated? Were there any extreme values that might skew results? How would you use the regression equations generated by the software? What preliminary conclusions would be supported and what further study indicated?

Attachment:- instructions.rar