What Science Says About Designing Better HR Surveys

Every year, human resource professionals distribute the dreaded employee satisfaction survey. Oftentimes, it's the same questionnaire emailed through SurveyMonkey, year after year. But HR workers should scrutinize their surveys, improving questions to make the ordeal worth everyone's while.

Usually, employees worry that their tell-all responses will be read by the boss, so they share a sanitized version of the truth. Even if they do divulge their real feelings, workers are then dismayed when no real change happens after the survey—it feels like their detailed concerns weren't even heard.

There is a better way. Social scientists study how to make the most scientifically accurate surveys, and HR professionals can put their findings to good use. I spoke with Frauke Kreuter, a sociologist and statistician at the University of Maryland, to learn a few important tips that HR workers can act on to improve their surveys immediately.

We ask questions on a daily basis, and as a result, some HR workers think surveys are easier to write than they truly are, Kreuter says.

"It took scientists 300 years to measure longitude—it does take time to develop a good measurement instrument," she says. "To measure satisfaction or any other concept in HR, thinking that you can do this in an afternoon is a misconception of how difficult measurement can be."

Instead, slow down and do pre-testing before you distribute your survey. Gather a small focus group of workers to test questions in person. When you ask them, "Do you feel engaged at work?" you'll notice that each person has a different definition of engagement, including time spent at work and job satisfaction. Once you gather sample responses, you'll be able to refine your questions to get more specific responses that will be useful in the workplace.

"Taking that time with focus groups, qualitative interviews and cognitive pre-testing to see if people understand the questions... that doesn't often happen," Kreuter says.

"On the web, with all the tracking possibilities, people might not always feel as safe as with a paper questionnaire," Kreuter says.

Instead of letting employees' imaginations run wild with fear over who might read their qualitative responses, spell out what will happen with their answers and why you're conducting the survey. Simply sharing who will be handling the data and how they will do it will lead to more honest answers.

To keep your survey unbiased, tailor your questions to the diversity of your audience. Kreuter has noticed that university-wide surveys often don't have responses that suit the different departments. After all, questions appropriate for the science department might not apply to the art faculty. Make sure you have questions and answers available for each type of worker. Then, make sure you poll them all.

"If you have a hypothesis on what types of people are more likely to differ in their answers, you want to make sure you represent all of these different types in the survey," she says. "If you only get the dissatisfied employees, that causes biases. If you only have those with secure employment and not those on contracts, that causes a problem. Think about your composition of your sample, it's key."

The biggest challenge of survey writing comes as a surprise: "You will always get an answer," says Kreuter. "By age three, you learned that you have to give answers—only small children can ignore a question. As adults, we give answers even if we don't have one. It's a mistake on the part of survey issuers [to assume] that just because they got answers, the survey results are meaningful."

To ensure that your survey responses are significant, ask specific, behavioral questions. " With those questions, you might be more likely to get actionable items," says Kreuter.

Instead of asking how to improve office morale, give your employees a multiple-choice question about how they would distribute a budget of $10,000 dollars across departments. These answers will be more concrete and actionable.

Take the time to improve your surveys, and the results might not be such a hassle to sift through. A well-designed employee satisfaction could even end up boosting office morale itself.

Photo: Twenty20