Binary logistic regression spss categorical variables
We may want to fit a logistic regression model using neighpol1 as our dependent variable and remploy binary logistic regression spss categorical variables, respondent employment status, as our independent variable to see if we can find a significant relationship between these two variables.
Just as we did binary logistic regression spss categorical variables the beginning of our logistic regression investigation of neighpol1 and agewe should run some exploratory analysis to determine if a relationship between these variables exists. When our independent variable age was continuous, we used a t test to compare means. Select AnalyzeDescriptive Statisticsand Crosstabs. Move neighpol1 into the Column s box and remploy into the Row s box. Click the Statistics button and select Chi-Square.
Click on Cellsand then under the Binary logistic regression spss categorical variables header, select Row. Then, click OK to run the crosstabulation. Your output should look like the one on the right. Is there a significant relationship between neighpol1 and remploy? How can you tell? Now we can fit our logistic regression model using neighpol1 as the dependent variable and remploy as the independent variable.
Select AnalyzeRegressionand then Binary Logistic. Move binary logistic regression spss categorical variables to the Dependent text box. Move remploy to the Covariates text box.
Because remploy is a categorical variable, we have binary logistic regression spss categorical variables tell SPSS to create dummy variables for each of the categories. SPSS will do this for us in logistic regression — unlike in linear binary logistic regression spss categorical variables, when we had to create the dummies ourselves.
Move remploy from the Covariates text box on the left to the Categorical Covariates text box on the right. The original Logistic Regression dialogue box should now have remploy Cat in the Covariates text box.
We also want SPSS calculate confidence intervals for remploy for us. In the Logistic Regression dialogue box you should have open, click Options. Now we can examine the output. Again, just like in the simple logistic regression we performed on the previous page, we will be predicting the odds of being unaware of neighbourhood policing in this logistic regression.
The Categorical Variables Codings table shows us the frequencies of respondent employment. In addition, it also tells us that the three categories of remploy have been recoded in our logistic regression as dummy variables. In logistic regression, just as in linear regression, we are comparing groups to each other. In order to make a comparison, one group has to be omitted from the comparison to serve as the baseline.
You can change the category to be used as the baseline to either the first or last categories — this is done where you specify that the variable is categorical. Remember that the Omibus Tests of Model Coefficients output table shows the results of a chi-square test to determine whether or not employment has a significant influence on neighbourhood policing awareness.
The Chi-square has produced a p-value of. Take a look at the Variables in the Equation output table below. If we were to fit this model again, and wanted to use remploywe may be tempted to remove remploy 2 from the model, as it is not significant.
Because remploy binary logistic regression spss categorical variables with a p-value of. This means that the employed are more likely than the economically inactive to know about neighbourhood policing. An odds ratio less than 1 means that the odds of an event occurring are lower in that category than the odds of the event occurring in the baseline comparison variable. An odds ratio more than 1 means that the odds of an event occurring are higher in that category than the odds of the event occurring in the baseline comparison variable.
In addition, SPSS has calculated confidence intervals for us. Remember that confidence intervals allow us to extend out analyses from the sample in our data to the population as a whole. First, you used a chi square test test to determine whether or not a statistically significant relationship existed between our categorical independent variable remploy and our categorical dependent variable neighpol1. Then, using simple logistic regression, you predicted the odds of a survey respondent being unaware of neighbourhood policing with regard to their employment status.
Finally, using the odds ratios provided by SPSS in the Exp B column of the Variables in the Equation output table, you were able to interpret the odds of employed respondents being unaware of neighbourhood policing.