I have a questionnaire with 2 sets of 4 main questions, with confidence questions following depending on their answer. Participants are randomly allocated to one condition and answered minimum 4 questions and maximum 8. Each question has information displayed in a different order. For example, one question would be to choose between brand A, brand B, or prefer not to choose. If they chose brand A or B, a follow-up question would appear asking their confidence on a scale of 1-5 (20%, 40%, 60%, 80%, 100%). We take "prefer not to choose" as 0% confidence. (So choosing "prefer not to choose" means they answer 4 questions only, and choosing a brand means they get a confidence question each time so will answer up to 8 questions - 4 choosing a brand and 4 confidence).
I need to calculate the mean confidence levels for each participant for each data order (the 4 main questions) but participants answered anywhere from 4-8 questions each. Currently I have written this command in SPSS which seemed to work:
compute Mean1 = mean.1(v1, v2, v3, v4, v5, v6, v7, v8).
Where v1/v2 are data order 1 (v1 is confidence in brand A, v2 is confidence in brand B), v3/v4 is data order 2, v5/v6 is data order 3, v7/v8 is data order 4.
From my understanding of this syntax, SPSS will only calculate the mean if the participant answered 1 or more confidence questions. What I am confused at is, whether SPSS will calculate the mean based on the number of questions a participant answered? For example:
Participant A answers 8 questions (4 brands chosen, 4 confidence questions)
Participant B answers 4 questions (2 brands chosen, 2 confidence questions)
Participant C answers 3 questions (1 prefer not to choose, 1 brand chosen, 1 confidence question)
Since they have all answered at least 1 confidence question, will SPSS do:
Total confidence divided by 4 questions (Participant A)
Total confidence divided by 2 questions (Participant B)
Total confidence divided by 1 question (Participant C)
Or just Total confidence divided by 8 each time (since there are 8 variables specified in the syntax)? Ideally I want SPSS to do the former (divide by the number of questions answered).
I appreciate this was very long and hard to visualise but please if you need any more details I am happy to provide.
Hello Miguel,
The spss syntax:
compute mean1 = mean(v1, v2, v3, v4) .
will compute a mean score as long as there is at least one non-missing value for a case in the set of scores v1-v4. Only non-missing values are included in the computation (both for numerator and denominator). Assuming that's what you want, that's the ticket.
It wasn't clear to me why you'd want to combine confidence scores across products, but perhaps that makes sense for your research aims.
Good luck with your work.
David Morse thank you for your answer, that's exactly what I was looking for.
I'm struggling to understand what you are doing and why, Miguel Benitez
First, think about how your data are scaled. If a question asks respondents to choose among A or B or nothing, then how is that recorded? If it's something like A=1, B=2, None=3, then what is an average? You have categories, not numbers, and any numbers are simply symbols that identify the categories. Computers generally, and statistics packages in particular, like to see numbers. Also, computers are stupid - they will do exactly what you tell them to do, if it makes sense or not. If you tell SPSS to find the average of A, or B or Nothing then it will do so. So, how might you interpret an average of 1.3?
Second, does it make sense to find the average of a three-level scale (choices) and a five-level scale (confidence)? In this context I would scale the choice question as: A=1, B=0, None=MissingValue. With that, an average score gives you the proportion of those respondents, who made a choice, who selected A.
Third, does it make sense to find the average of two completely different concepts: Choice and Confidence? I think not.
Or perhaps I have interpreted your question wrongly. Can you give us more detail on what the questions are (or something similar) and how you have scaled them?
- Badges
- Science topic
- Similar topics
- Statistics
- Survey