In our last post, we talked about how to ask people about their own behavior, like how many times they’ve gone to the gym this week. Now we’ll show you how to ask people to tell you predictions of their own behavior, like how many times they plan on going to the gym next week.
There are two important obstacles in getting accurate predictions. In order to show you what we mean by each, let’s dive into two different situations.
Sometimes people just don’t know the truth
Katie wants the perfect prom dress, but more importantly, she wants to make sure none of her friends wear the same one. Katie gets a little carried away. On the first day of school in September, she surveys all of her friends to find out what color dress they’re buying for prom in May. Flash-forward to prom night, Katie shows up wearing a magenta dress only to see that Stacey is wearing the same exact one. Boiling with rage, she reminds Stacey that on the September survey she had said she’d be wearing yellow. Stacey shrugs and says, “Well, yellow would clash with my new blonde highlights. I didn’t have highlights in September.”
What happened? People aren’t very good at making long-term predictions. Circumstances can change in ways that we can’t anticipate and that impacts the way that we behave.
How could you avoid this problem?
- Repeat it: It might be that Stacey–or any of Katie’s other friends–changes her opinions about dress color a lot. Sending out a survey more than once will capture these shifts and allow Katie to intervene before prom night and remind her friends how she has that magenta dress on lockdown!
- Keep it current: Survey people as close to the event as possible. In Katie’s case, if she had surveyed her friends two weeks before prom, she might have discovered Stacey’s change of heart before prom night.
Sometimes people just don’t want to tell you the truth
In 1982, Tom Bradley ran for governor of California against George Deukmejian. Polls leading up to the election showed an easy win for Bradley. The San Francisco Chronicle was so sure about this that it declared: “Bradley Win Projected.” This became a little embarrassing for them when Bradley lost the election to Deukmejian. Why? Racial dynamics was said to have played a role. Tom Bradley is African-American and it seemed as if the white voters who were polled over the phone were uncomfortable admitting out loud that they didn’t want to vote for an African-American candidate.
What happened? People don’t like admitting behaviors–in surveys or otherwise–that might make them look bad. When asking for predictions of what people will do, watch out for social desirability bias just as we discussed earlier.
How could you avoid this problem?
- Keep it confidential: Write a sentence or two reminding your respondents that no one is reading their answers and that the responses will be analyzed in aggregate only—that is, you’re looking at averages and trends, not individual responses. People will be more truthful if they think no one will be able to trace their answer back to them.
- Let it out: Allow respondents to explain why if they want to. This will allow them to justify the circumstances that led them to pick a certain answer. It doesn’t matter if the explanation they provide is truthful or not, it’s there as a safety blanket for them.
There you have it–all the tips you need to start predicting behaviors without having to drag around a pesky crystal ball!
Don’t be shy! What kinds of predictions will you ask for in your next survey?
Oh, and what color shirt are you planning on wearing two years from now–we want to make sure it doesn’t clash with the blog post we’ll be posting that day.
Image courtesy of BigStock photo