Advance Biostats SPSS Assignment: Multiple Logistic Regression in Action

PART2

Step-by-Step Guide to Assignment 7.2

Logistic Regression

Problem 2. Simple Binary Logistic Regression

a. Use Hypertension as the dependent variable and Chole_Cat as the independent variable in the first model. Report the Odds Ratio and significance of the Odds Ratio for the relationship between the dependent and independent variables.

Step 1: Open the saved SPSS data set from 7.1.

Step 2. Go to Analyze ( Regression ( Binary Logistic

Step 3. The Logistic Regression window will appear. Move the Hypertension variable to the Dependent box and move the Cholesterol Category (Chole_Cat) variable to the Covariates box using the arrows. Click on Categorical

Step 4. In the Logistic Regression: Define Categorical Variables window, move Chole_Cat to the Categorical Covariates box

Step 5. In the Change Contrast area, make sure the Indicator default is selected for Contrast. Select First for Reference Category and click Change. This tells SPSS to use the lowest Chole_Cat level (Level 1), which are the normal range values, as the reference for comparing means in Levels 2 and 3.

Step 6. “First” should appear next to Chole_Cat in the Categorical Covariates box. Click Continue.

Step 7. In the Logistic Regression window, click Options.

Step 8. The Logistic Regression: Options window will open. Check the boxes for Classification plots, Hosmer-Lemeshow goodness-of-fit, Casewise listing of residuals, and CI for exp(B) 95%. Click Continue

Step 9. In the Logistic Regression window, click OK

The SPSS output:

Variables in the Equation
	B	S.E.	Wald	df	Sig.	Exp(B)	95% C.I.for EXP(B)
							Lower	Upper
Step 1a	Chole_Cat			10.989	2	.004
	Chole_Cat(1)	1.294	.537	5.816	1	.016	3.648	1.274	10.443
	Chole_Cat(2)	2.667	.828	10.369	1	.001	14.400	2.840	73.018
	Constant	-1.569	.492	10.182	1	.001	.208
a. Variable(s) entered on step 1: Chole_Cat.

Results:

Exp(B)(exponentialtion of the B coefficients) gives the odds ratio.

In your response, be sure to also discuss what the odds ratios mean in this model.

Problem 2b:

b. Use Hypertension as the dependent variable and Serum Cholesterol (the original variable) as the independent variable in the second model. Report the Odds Ratio and significance of the Odds Ratio for the relationship between the dependent and independent variables.

Note: Before moving on to the next problem, save your Output from Problem 2a above and store it in an Output folder as Problem 2a_Hypertension-Chole_Cat. This will help if you need to go back and review the Output. Once it is saved, you can close out the Output window. After you complete the below steps for Problem 2b, another Output window will open containing only the output for Problem 2b.

Step 1. Go to Analyze ( Regression ( Binary Logistic.

Step 2. In the Logistic Regression window, click Reset to clear the selections used in the previous problem.

Step 3. Place Hypertension in the Dependent box and place Serum Cholesterol [scl] in the Covariates box.

Note: The scl variable contains data that are continuous, so we will not be using the categorical button for this regression.

Step 4. Select Options and check the boxes as below. Click Continue.

The SPSS output:

Variables in the Equation
	B	S.E.	Wald	df	Sig.	Exp(B)	95% C.I.for EXP(B)
							Lower	Upper
Step 1a	scl	.013	.004	8.673	1	.003	1.013	1.004	1.022
	Constant	-3.416	1.045	10.694	1	.001	.033
a. Variable(s) entered on step 1: scl.

Results:

In your response, be sure to comment on how an increase in serum cholesterol relates to hypertension based on this model, as well as whether this is significant and why.

Problem 2c:

c. How does the level of measurement for the independent variable affect the outcome (include the OR and its significance in your response)? How does the level of measurement of the independent variable change your interpretation of the Odds Ratio?

In your response, discuss how the OR changed and whether this would effect the interpretation of the relationship.

Why would you chose to use categorized data on cholesterol rather than continuous?