Anyone who has entered KAP survey data knows the moment things get confusing. It usually happens at a question with more than one answer. The trick to remember is simple. How many variables a question needs depends on how many responses one person can give, not on how the questionnaire looks.
In short
- Single answer question → one variable (q1): Yes/No coded 1/2.
- One choice from many options → still one variable (q7): each option gets its own code (1–4).
- Select-all-that-apply → one Yes/No variable per option (q9a, q9b, q9c…).
- Score afterwards from these variables — never collapse a multi-select into one column at entry.
KAP studies throw three kinds of questions at you. Each needs its own coding approach in SPSS, and each section below comes with a live form so you can see what the respondent actually experiences.
1. Single response, one variable
Take the question Can you identify an Aedes mosquito? with a yes or no answer. One answer means one variable. Call it q1, enter 1 for Yes and 2 for No, and you are done.
Here is how that looks in the SPSS Variable View.
| Name | Type | Width | Decimals | Label | Values | Missing | Measure |
|---|---|---|---|---|---|---|---|
| q1 | Numeric | 8 | 0 | Can identify Aedes mosquito | {1 = Yes, 2 = No} | None | Nominal |
No 43 (36%)
2. One choice from many options is still one variable
Now ask people how often they wash their hands, with answers ranging from never to always. There are four options here, but a respondent still picks just one of them. That makes it a single variable job. Call it q7 and give every option its own number.
- 1 = Never
- 2 = Sometimes
- 3 = Often
- 4 = Always
| Name | Type | Width | Decimals | Label | Values | Missing | Measure |
|---|---|---|---|---|---|---|---|
| q7 | Numeric | 8 | 0 | Handwashing frequency | {1 = Never, 2 = Sometimes, 3 = Often, 4 = Always} | None | Ordinal |
This ordered coding is exactly what you want when responses need to be scored or compared, since a bigger number simply means better practice.
3. Select all that apply needs one variable per option
The third kind asks about sources of information on dengue and lets people tick as many as apply. Television, radio, social media, health workers, anything else. A respondent may tick three boxes while another ticks none, but a single SPSS cell can only hold one value. So we split the question up and give each option its own little yes or no column.
- q9a for TV, where 0 means unticked and 1 means ticked
- q9b for Radio
- q9c for Social media
- q9d for Health worker
| Name | Type | Width | Decimals | Label | Values | Missing | Measure |
|---|---|---|---|---|---|---|---|
| q9a | Numeric | 8 | 0 | Source of info – TV | {0 = No, 1 = Yes} | None | Nominal |
| q9b | Numeric | 8 | 0 | Source of info – Radio | {0 = No, 1 = Yes} | None | Nominal |
| q9c | Numeric | 8 | 0 | Source of info – Social media | {0 = No, 1 = Yes} | None | Nominal |
| q9d | Numeric | 8 | 0 | Source of info – Health worker | {0 = No, 1 = Yes} | None | Nominal |
Multiple responses, so percentages add up to more than 100. Each person could tick several sources.
Once every option sits in its own column, finding out how many people learned about dengue from radio is a single Frequencies run away. You find it under Analyze, then Descriptive Statistics. Defining a Multiple Response set works nicely too if you want the totals side by side. One thing to expect with these tables: the percentages will add up to well over 100 percent. That is not a mistake in your data. Since each respondent could tick several sources, every option is counted out of the full sample independently. A total of 216 percent simply means people picked about two sources on average.
4. Scoring comes later, and good coding makes it easy
Why bother coding so carefully at entry? Because scoring becomes almost effortless afterwards, using nothing fancier than simple COMPUTE statements in SPSS.
- Knowledge items, like our yes or no question, get 1 for correct and 0 otherwise. Add them up across items and you have a knowledge score.
- Attitude and practice items already carry ordered codes from 1 to 4, so the sum itself becomes a practice score. A higher total simply reflects better practice.
- Multiple-select items are already stored as 0s and 1s, which leaves scoring entirely up to whatever your analysis needs. Count how many sources someone ticked by summing q9a to q9d. Turn at least one tick into a plain yes or no indicator if that suits your question better. Or weight the options first, say counting advice from a health worker double, before adding them up.
This flexibility is the real reason we keep one column per option instead of squeezing everything into a single code during entry. The raw detail stays intact. Frequencies stay one click away. And if your supervisor changes the scoring rule two weeks before the report is due, nobody has to go back to the paper questionnaires.
The rule of thumb
One possible answer per person gets one variable. Many possible answers get one variable per option. Do all your scoring afterwards from those variables, and never collapse anything at entry.
