Understanding Logistic Regression Analysis Using SPSS

Logistic Regression in SPSS Directions: Follow the example to find unadjusted and adjusted odds ratios in SPSS. Then do the same for the questions. Use the format of the example interpretation for an adjusted odds ratio to interpret the odds ratios in the following questions. Example. We have seen previously that there is a relationship between child abuse exposure and experiencing pain limiting activity, as seen through the odds ratio and 95% confidence interval below. Table 1: Logistic regression results for any CPA or CSA and pain Interpretation: The odds of pain are 2.043 times higher (95% CI: 1.894-2.204) for those who experienced any CPA or CSA compared with those who did not. We know that this relationship is statistically significant because 1 is not in the 95% confidence interval. To get this output in SPSS, we go to Analyze  Regression  Binary Logistic, as shown in Figure 1. Since pain is the outcome, we move it into the box labeled Dependent. Since Any CPA or CSA (ch_abuse) is the exposure, we move it into the box labeled Covariates. Figure 1: Logistic regression menu and categorical menu in SPSS

Logistic Regression in SPSS We also need to tell SPSS that any CPA or CSA is a categorical variable. To do this, we use the following steps: 1. Select the “Categorical” option from the main logistic regression menu, which opens the Define Categorical Variables menu box (shown in Figure 1). 2. Move ch_abuse from Covariates into Categorical Covariates. 3. Under “Reference Category”, select First. Then select “Change” (immediately above First). You should see first in parentheses after the variable name, just like it is in Figure 1. 4. Click Continue. We must also tell SPSS that we want the 95% confidence interval for the odds ratio. From the main Logistic Regression menu, we take the following steps: 1. Select “Options”, which opens the Logistic Regression Options menu (Figure 2). 2. In the second column of options, check “CI for exp(B)”. It will automatically set the level to 95%. 3. Click Continue. Figure 2: Logistic regression menu and options menu in SPSS Once you have completed these steps, you should get the output seen in Table 1. However, we know from previous assignments that there are other variables that may be confounding variables, or ones that are related to both exposure and outcome, distorting the relationship between any CPA or CSA and pain. Therefore, we should add those to the logistic regression model. To add additional explanatory variables, simply move them into the Covariates box. If they are categorical, follow the same procedures as above to make sure that SPSS treats them as categorical variables. We will look at the rural/urban variable as an example.

Logistic Regression in SPSS Figure 3: Logistic regression and categorical menus for including rural/urban as a confounding variable There are three main tables we want to look at in the output: Case Processing Summary (first table), Categorical Variable Codings (third table), and Variables in the Equation (last table). Table 2: Case Processing Summary table for logistic regression Table 2 shows us that there were 726 missing cases. Therefore, the final analytic sample size for this analysis was 15,367 (included in analysis row). Table 3 (next page) shows us the variable codings for the exposure and confounding variables. As before, any CPA or CSA is the exposure, so “yes” versus “no” is what is shown in the output. For rural/urban, living in an urban environment is the numerator (“yes”), while living in a rural environment is the denominator (“no”).

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

Logistic Regression in SPSS Table 3: Variable codings for any CPA or CSA and rural/urban Table 4 shows the final table of the SPSS output, which includes lines for the main exposure (any CPA or CSA), the confounding variable (rural/urban), and the constant (“y-intercept”, which we can safely ignore for these purposes). Table 4: Logistic regression results for any CPA or CSA and pain, adjusting for rural/urban living In the output from Table 4 , there are estimates of odds ratios and 95% confidence intervals for all the variables we included in the regression. However, we are really interested in the relationship between the exposure and outcome – any CPA or CSA and pain – while adjusting for all of these other variables. Therefore, we only interpret the adjusted odds ratio for any CPA or CSA. We know from Table 3 that no CPA or CSA is our reference group. Therefore: The odds of pain are 2.033 times higher (95% CI: 1.881-2.197) for those who experienced any CPA or CSA compared with those who did not, adjusting for rural or urban living. This interpretation is nearly identical to the unadjusted one. Since the odds ratio was calculated with other variables in the regression, we must include what those variables are in the interpretation. Our result is still statistically significant, since 1 is not in the confidence interval. In addition, we know that since the odds ratio (and confidence interval) is greater than 1, the odds of pain is higher among those who experienced any CPA or CSA than those who did not.

Logistic Regression in SPSS Question 1. Using the example as a guide, find and interpret the unadjusted and adjusted odds ratios for the relationship indicated. a. Use SPSS Binary Logistic Regression to find the odds ratio and 95% confidence interval for the relationship between any CPA or CSA ( ch_abuse , exposure variable) and any gastrointestinal (GI) problem ( anygi , outcome variable). Copy/paste or screenshot/paste the Variables in the Equation table below. b. What is the interpretation for the odds ratio and 95% confidence interval? The odds of any gastrointestinal problem are 1.868 times higher (95% CI: 1.732-2.013) for those who experienced any CPA or CSA compared with those who did not. c. Use SPSS Binary Logistic Regression to find the odds ratio and 95% confidence interval for the relationship between any CPA or CSA ( ch_abuse , exposure variable) and any gastrointestinal (GI) problem ( anygi , outcome variable), adjusting for ever problem alcohol use ( probalc ). Copy/paste or screenshot/paste the following 3 tables: Case Processing Summary, Categorical Variable Codings, and Variables in the Equation.

Logistic Regression in SPSS d. Are there any missing cases? How do you know? Yes, the Case Processing Summary table states that there are 49 missing cases. e. What is the interpretation for the odds ratio and 95% confidence interval for the relationship between any CPA or CSA and any GI problem, adjusting for ever problem alcohol use?

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

Logistic Regression in SPSS The odds of any gastrointestinal problem are 1.850 times higher (95% CI: 1.715-1.995) for those who experienced any CPA or CSA compared with those who did not, adjusting for ever problem alcohol use. f. Did the odds ratio change from part (a) to part (c)? If so, is it higher or lower in part (c)? The odds ratio without adjusting for alcohol use was 1.868 whereas when adjusting for ever problem alcohol use, it dropped to 1.850. Therefore, the odds were 0.018 lower. Question 2. Using the example as a guide, find and interpret the unadjusted and adjusted odds ratios for the relationship indicated. a. Use SPSS Binary Logistic Regression to find the odds ratio and 95% confidence interval for the relationship between any CPA or CSA ( ch_abuse , exposure variable) and ever cardiovascular disease ( evcvd , outcome variable). Copy/paste or screenshot/paste the Variables in the Equation table below. b. What is the interpretation for the odds ratio and 95% confidence interval? The odds of ever cardiovascular disease are 1.913 times higher (95% CI: 1.715-2.133) for those who experienced any CPA or CSA compared with those who did not. c. Use SPSS Binary Logistic Regression to find the odds ratio and 95% confidence interval for the relationship between any CPA or CSA ( ch_abuse , exposure variable) and ever cardiovascular disease ( evcvd , outcome variable), adjusting for race ( race ) and age group ( ageg2 ). Copy/paste or screenshot/paste the following 3 tables: Case Processing Summary, Categorical Variable Codings, and Variables in the Equation.

Logistic Regression in SPSS

Logistic Regression in SPSS d. Are there any missing cases? How do you know? Yes, the Case Processing Summary table states that there are 118 missing cases. e. What is the interpretation for the odds ratio and 95% confidence interval for the relationship between any CPA or CSA and ever having cardiovascular disease, adjusting for race and age group? The odds of ever cardiovascular disease are 1.883 times higher (95% CI: 1.685-2.105) for those who experienced any CPA or CSA compared with those who did not, adjusting for age and race f. Did the odds ratio change from part (a) to part (c)? If so, is it higher or lower in part (c)? The odds ratio without adjusting for age and race was 1.913 where as when adjusting for age and race, it dropped to 1.883. Therefore, the odds were 0.030 lower.

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

logistic_regression_spss

Related Documents