Week 5: Multiple Logistic Regression
Last week you explored categorical relationships including conditional probabilities, conditional odds, logits, odds ratios, relative risk, and slopes in the discussion. In the SPSS statistics assignment, you conducted a simple logistic regression analysis (one independent variable and one dependent variable) to begin to explore this statistical test and odds ratios. This week you start to fit models using multiple predictor variables that are continuous and categorical. As you learned with regression, you often need to construct models that move beyond bivariate analysis to control for other variables. In this sense, logistic regression is very similar except that you use a dichotomous outcome variable and interpret odds ratios rather than unstandardized coefficients.
Countless times, you have probably heard someone ask, “What are the odds?” At the conclusion of this week, you will be able to act as a statistical guru, educating others about what they are actually asking. Moreover, given data, you will be able to model these odds and provide meaningful results.
This week you will build on the simple logistic regression analysis did last week. You will use the same two variables (one independent variable and one dependent variable) you used in your SPSS analysis last week and add a second independent variable to the analysis. How has your statistical significance and odds ratio changed by the introduction of the second independent variable?
Learning Objectives
Students will:
- Apply multiple logistic regression tests in a dataset
- Analyze multiple logistic regression tests in a dataset
- Apply statistical software to analyze data
- Critique research studies that apply multiple logistic regression
Learning Resources
Required Readings
Warner, R. M. (2013). Applied statistics: From bivariate through multivariate techniques (2nd ed.). Thousand Oaks, CA: SAGE Publications.
- Chapter 23, “Binary Logistic Regression” (pp. 1007–1050)
This chapter presents several simple examples of binary logistic regression. It also explains why researchers need to include probabilities along with odds when they report results.
Datasets
In this course, your instructor will provide you with one dataset (Guided Sample Dataset: Statistics Anxiety) that will be used in several of the tutorial videos. You are encouraged to reproduce the analyses performed in the video using this dataset so you can compare your answers and check your understanding of how to do each analysis in SPSS. You will also be provided with three large datasets for you to use in application of the statistics you’re learning each week. The Statistics Anxiety dataset is not for use in your weekly Assignments. Your instructor may also recommend using a different dataset from the ones we have provided here, which is fine. Your instructor will share these datasets in the Doc Sharing section of the classroom and as an Announcement.
In this course, your instructor will provide you with one dataset (Guided Sample Dataset: Statistics Anxiety) that will be used in several of the tutorial videos. You are encouraged to reproduce the analyses performed in the video using this dataset so you can compare your answers and check your understanding of how to do each analysis in SPSS. You will also be provided with three large datasets for you to use in application of the statistics you’re learning each week. The Statistics Anxiety dataset is not for use in your weekly Assignments. Your instructor may also recommend using a different dataset from the ones we have provided here, which is fine. Your instructor will share these datasets in the Doc Sharing section of the classroom and as an Announcement.
Required Media
Laureate Education (Producer). (2017j). Introduction to binary logistic regression [Video file]. Baltimore, MD: Author.
Note: The approximate length of this media piece is 1 minute 30 seconds.
Begin each week of the course by viewing the Weekly Introduction video, in which experienced statistics Instructors, Dr. Matt Jones and Dr. Annie Pezalla, provide context for new learning, explain statistical methods in easy-to-understand language, and describe the learning activities you will need to complete.
Accessible player –Downloads– Download Video w/CC Download Audio Download Transcript
Laureate Education (Producer). (2017d). Binary logistic regression [Video file]. Baltimore, MD: Author.
Note: The approximate length of this media piece is 12 minute 45 seconds.
This video provides step-by-step instructions that explain how to perform the statistical test required in Assignment 1.
Accessible player –Downloads– Download Video w/CC Download Audio Download Transcript
Collaboration Lab
Collaboration Labs are offered throughout the course as places of academic refuge, where you can ask questions, offer insights, and interact with the Instructor as they provide global feedback to the group based on trends, common problems, and common strengths in student posts.
As a peer, you are also encouraged to provide constructive, helpful feedback to your peers. Data analysts always benefit from the feedback of others. Your Collaboration Labs posts may be procedural (“How do I get the answer using the SPSS software?”), conceptual (“How does this relate to the other tests we have learned?”), or analytical (“What do my calculations actually mean in the context of my research question?”).
Although not mandatory, this is an opportunity to interact and practice as you navigate the assignments, so you are highly encouraged to take part in this activity. Full participation in activities like these is a statistically significant predictor of success.
Note that each week contains its own unique Collaboration Lab, so the Collaboration Lab should contain only posts related to this week’s topic and this week’s Assignments.
To participate in this Collaboration Lab:
Week 5 Collaboration Lab
Assignment 1: Binary Logistic Regression in SPSS
This week you will build on the simple logistic regression analysis did last week. You will use the same two variables (one independent variable and one dependent variable) you used in your SPSS analysis last week and add a second independent variable to the analysis. How has your statistical significance and odds ratio changed by the introduction of the second independent variable?
To prepare
- Use the one independent variable and one dependent variable you used to conduct your simple logistic regression analysis in Week 4.
- Add a second independent variable to your analysis (multiple logistic regression).
- Remember that your dependent variable must be dichotomous/binary.
- Think about how you might use the odds ratio in your analysis to simplify the interpretation of your results.
- How has your statistical significance and odds ratio changed by the introduction of the second independent variable?
By Day 7
The Assignment
Use SPSS to answer the research question you constructed. Write an analysis in APA format, including title page, references, and an appendix, that includes your data output and addresses each of the tasks listed above. The content should be 2–3 pages, including setup of the assignment and a discussion of whether the predictive relationship is statistically significant as well as the odds ratio and what it means. Your SPSS output should be included as an appendix.
Early in your Assignment, when you relate which dataset you analyzed, please include the mean of the following variables. If you are using the Afrobarometer Dataset, report the mean of Q1 (Age). If you are using the General Social Survey Dataset, report the mean of Age. If you are using the HS Long Survey Dataset, report the mean of X1SES. See page 1032 in your Warner textbook for an excellent APA-compliant write-up of a binary logistic regression analysis.