How to analyze multiple response questions in SPSS

March 25, 2021

Multiple response questions are very common in research. When working with multiple responses in SPSS, there are specialized procedures that will help you better summarize a combination of variables resulting from multiple response questions. In this tutorial, you will learn multiple response analysis using frequencies, crosstabulations and charts in SPSS.

Video transcript

Multiple response questions like these ones are very common in research.

So instead of allowing a respondent to pick just one option, with a multiple response, the respondent can choose multiple options from a list provided.

When entering the data in software like SPSS or Excel, this is the way you set it all up

Each of the items in your list options becomes a variable on its own.

In many cases, it is recommended to use 0 to mean the option was not selected and 1 to mean that the option was selected.

0 – no it wasn’t selected, 1 it was selected

If you used data collection software like KoBo Toolbox, or Open data Kit, your multiple response questions will already be exported with each option being represented by a variable and zeros and ones being used for not selected and selected.

Now let’s look at how you would summarize such variables in SPSS

The first option for summarizing multiple response questions is to run frequency tables for all the variables. Let’s try that

So you go to Analyze…

Descriptive statistics

Frequencies

I will grab all the variables except the ID into the Variables box on the right

Click OK

The output is OK. At least from here we can tell that for example 70.5% have farming as one of their household sources of income

For  non-farm business, that’s 51.5% and so on

The problem with this however is that there are way too many tables

In most cases, we are only interested in the frequency of those who chose each of the options we have. That means we are only interested in counting the ones and not the zeros

This is where you need to use the multiple responses option in SPSS

First, you need to define each multiple response set

Let’s minimize the output

You will find that there are several places for defining multiple response sets in SPSS – for reasons best known to the developers at IBM

And unfortunately, you need all of them

Let’s start with going to Analyze

Multiple response

Define variable set

We will drag all the variables in our multiple response set to the right hand side

Then, we will specify which value we are going to be counting here where it says counted value.

Remember that we are only interested in counting the number of people who actually selected the response. And when they did so, we recorded 1

So we will type 1 as the counted value

Then, we need to give a name to the variable that will hold our multiple responses.

I will say IncomeSources, without a space – remember this is a variable name

Ten we can specify the label for our variable

I will say Household sources of income…

Then click Add

If you have more multiple response sets, you can continue in the same manner, adding the variables to the variables in set box and the specifying the value you are counting, then finally the variable name and label

Let’s close this

So now we have our variable, we can use it to create one table of frequencies. But we can only do that using the Multiple response option on the menu

So click Analyze again

Multiple response

Frequencies

Drag the multiple response variable to the right hand side…

Click OK

This looks much neater

The percent value is expressed out of the total N at the bottom.

The total N here is not the number of cases in the data set, but rather the total number of times all the options were selected.

Since these are multiple responses, the Total N is bigger than the actual number of cases.

So the Percent should be construed as a frequency relative to the total of the number of times all the options were selected.

While the percent of cases here is for each of the option out of the total number of cases we have in the data set.

It’s exactly the frequencies we had earlier on when we used the Frequency procedure to summarize each of the variables individually.

So the Percent should tell you which option was selected the most out of all the options, but Percent of cases should now tell you the real representation of how many times the option was selected by the total number of cases.

You may have noticed that we can also produce a crosstabulation using out multiple response variable.

Let’s try that

Go to Analyze…

Multiple response

Crosstabs…

Let’s drag the household sources of income variables here, to the rows…

Then let’s use the gender variable in the columns

Notice we have question marks inside the gender brackets.

That’s because we need to specify the values of gender

So click define ranges

We are using 0 for female and 1 for male so we will type those values here

Click continue

It’s best to turn on column percentages so we can see the relative frequencies by gender

To do that, click options

Under cell percentages, click column

Click continue

Finally, click OK

Nice, so now each of the options has been cross tabulated by gender, without having to worry about those who did not select the option

Now what happens when we want to produce some charts with the multiple response variable

Let’s go to Graphs

Chart builder

Well, as you can see from the variables list here, the multiple response variable we just defined from the Analyze menu is nowhere to be seen

That variable does not appear here.

The one that will appear here is the one you will have to define using the Data menu

So lets close this for now

Go to Data…

Define multiple response sets…

The way we define the multiple response set here is exactly the same

So drag our multiple response variables from farming to remittances into the variables in set box

In the counted value box, we type 1

Then we go ahead and define the set name here.

Once again, I will call it IncomeSources, again without a space

This time, no label. Click Add

Click OK

Minimize the output

Now this time when you go to graphs

Chart builder

Perfect, we have our multiple response variable right here.

Let’s try a bar chart with it

Drag and drop the first bar chart type[e from the gallery

Drag income sources and drop it onto the x axis

Let’s change the statistic using the element properties

So here, instead of count, let’s change to percentage

Click OK

Awesome

Now we have a bar chart using our multiple response variable

That’s it about summarizing multiple response variable

We have even more cool SPSS tutorials coming your way. So before you go, click the subscribe button and hit that bell icon so you don’t miss anything

Till next time

Learn SPSS, STATA, KoboToolbox, Power BI and more

Join DATAFORDEV for self-paced courses, live online workshops, and a private community forum for social impact professionals.

Join now