Use of code as an analysis unit

Add-on: I think I have found a solution, although perhaps not the simplest one. If I activate code A and all relevant documents, the n = 1086 codes of code A are displayed in the list of coded segments. I can then export this as an Excel file and select to save additional codes assigned to the segment as supplementary information in the Excel table. I can then use Excel to count the occurrence of code B, C, D, etc. within code A, even if not directly in MAXQDA.

Dear Khaled,
Thank you very much for your feedback. I assume your explanation refers to the Visual Tools > Code Relations Browser function.
I have tried the following setting:

• Rows: Code A
• Columns: Code B
• Type of analysis: Intersection of codes in a segment
• Tested both with and without selecting code A as “only for segments of the following codes”.

Due to the high number of individual codes, I have set up a fictitious code B, which I have assigned a total of four times within three segments of code A. So I now have three times code A in which code B occurs one or more times, even if the total number of code B assignments is four. Now I have used the Code Relations Browser with the above settings. Regardless of whether I selected Code A as “only for segments of the following codes” or not, I was always shown four overlaps as a result, although the result I was looking for would be three. So the suggested function still does not fulfill my requirements.
Do you have any other suggestions or alternative approaches to achieve this “if-then” counting without multiple counting for code B within the same code A segment?

Best regards,
Birte

Dear Birte, 

To achieve your objective of determining the percentage of Code A segments that include at least one instance of Code B, you can use the following approach in MAXQDA:

1. Use the Code Co-Occurrence Analysis: Instead of relying solely on the Code Patterns feature, consider using the Code Co-Occurrence analysis. This allows you to analyze how many segments of Code A also have Code B assigned to them.

2. Select the Right Unit of Analysis: When setting up your analysis, ensure that you select "Coded segments" as the unit of analysis. This way, you can focus on the segments rather than documents.

3. Count Hits for Code A: In your analysis, you can count how many segments of Code A have been assigned. This will give you the total number of Code A segments (n = 1068).

4. Analyze Co-Occurrences: Then, analyze how many of those Code A segments also have Code B assigned. This will allow you to calculate the percentage of Code A segments that include at least one instance of Code B.

5. Calculate the Percentage: Finally, you can calculate the percentage by dividing the number of Code A segments that co-occur with Code B by the total number of Code A segments (1068) and multiplying by 100.

This method will provide you with the desired analysis based on the total number of Code A segments rather than the number of documents.

If you need further assistance with specific steps in MAXQDA, feel free to ask!

Best regards,
Khaled

Dear Khaled,

Thank you for your explanation regarding the use of the Code Patterns feature and the “Count hits only once per document” option. This approach indeed prevents multiple counts of Code B within a single document.

However, I’m facing an additional challenge. In my dataset, Code A segments (n = 1068) appear multiple times across a total of 162 documents. If I limit the count to one hit per document, the resulting relative frequency of Code B is calculated based on the number of documents rather than on the number of Code A segments. For example, if code A was assigned 10 times within a document and an overlap with code B was identified 7 times, the co-occurrence would still only be counted once in this case (for the entire document), right?This does not align with my objective, which is to determine the percentage of Code A segments that include at least one instance of Code B.

Is there any alternative method or adjustment within MAXQDA Analytics Pro that allows the analysis to be based on the total number of Code A segments (n = 1068) rather than on the total number of documents? Or is my assumption about the function just wrong? I’d appreciate any insights or suggestions on how to resolve this issue.

Best regards,
Birte

Dear Birte, 

In MAXQDA Analytics Pro, you can achieve this type of analysis by using the "Code Patterns" feature. Here's how you can proceed:

1. Select Code Patterns: Use the Code Patterns tool to analyze the co-occurrence of codes. This tool allows you to analyze how many segments contain both Code A and Code B.

2. Count Hits Only Once: When evaluating segments, enable the option "Count hits only once per document." This ensures that each segment of Code A is counted only once, even if Code B appears multiple times within it.

3. Analyze Results: The results table will show you the number of segments where Code A and Code B co-occur, without multiple counts for repeated occurrences of Code B within the same segment. 

This approach will help you treat Code A as the primary unit of analysis and provide the "if-then" count you need. 

Hope this answers your questions. 

Best regards,
Khaled