Cronbach alpha for non-Likert & Likert survey

Question

Hi!

I have a thesis ongoing and I have a survey that needs for reliability testing. I was advised to do a Cronbach's alpha test. My survey consists of two parts:

1. Part 1: All questions have multiple choices with "Others:__" as a choice. Each question has a unique number of choices. Some questions have "Select all that apply".
2. Part 2: Likert scale questions. All are to be answered with a 4-rating key.

My question: I am aware that the Cronbach's alpha for the Likert scale will be coming from the numerical responses (from the 4-rating key) of the respondents. How would I do the Cronbach for Part 1? Also, would there be a different approach for questions with "Select all that apply"?

I tried searching information on how I will do a Cronbach's alpha test but all I'm seeing are "how to compute" for the test, not the "how to do Cronbach for non-Likert multiple choice questions" for the test. I'm aware that the Part 1 and Part 2 will be having a different set of Cronbach's alpha test due to it's different nature of question.

Another question: Part 1 and 2 has five sections each (let's name them A, B, C, D, and E). These sections are connected to answer one whole theme/objective. Should the Cronbach alpha test be separated per section or I can just test them all in one go?

Please do enlighten me regarding this. Also, if you have any references (websites, free books, etc.) that could be of help regarding my inquiry, please do share it with me. Thank you!

Answer 1

Hi Jared!

Conducting Cronbach's alpha for different types of questions in your survey requires a thoughtful approach. Here's guidance on how to handle Cronbach's alpha for each part of your survey:

Part 1 - Multiple Choice with "Others" and "Select All That Apply"

1. Multiple Choice with "Others":
2. For questions with multiple-choice options, including "Others," you can calculate Cronbach's alpha using the numerical responses assigned to each choice. Treat each response as a separate item in the analysis.
3. If there are multiple items within a question (e.g., checkboxes for "Select all that apply"), treat each selected option as a separate item for Cronbach's alpha calculation.
4. Handling "Select All That Apply" Questions:
5. For questions with "Select all that apply," consider treating each selected option as a separate item for Cronbach's alpha. However, keep in mind that this approach assumes independence between selected options. If options within a question are expected to be correlated, other reliability measures or factor analysis might be more appropriate.

Part 2 - Likert Scale Questions:

1. For Likert scale questions, Cronbach's alpha is calculated based on the numerical responses provided by respondents. Assign numerical values to each response option (e.g., 1 for Strongly Disagree, 2 for Disagree, and so on).
2. Use these numerical responses to calculate Cronbach's alpha for the Likert scale items collectively.

Testing Sections A, B, C, D, and E:

1. You can calculate Cronbach's alpha for each section separately and for the entire survey as a whole. This provides insights into the reliability of each section and the overall survey. It's advisable to test both individual sections and the entire survey to identify specific areas that may need improvement.

References:

1. Books:
2. "Psychological Testing: Principles, Applications, and Issues" by Robert M. Kaplan and Dennis P. Saccuzzo.
3. "Research Methods in Psychology" by John J. Shaughnessy, Eugene B. Zechmeister, and Jeanne S. Zechmeister.