Nothing can ruin an employee communication survey like poorly written questions. Bad questions decrease the likelihood that employees will complete your survey, as well as impact the validity of your data. Here are the most common problems I’ve seen over the years:

Problem: demographics overload

This example was actually a series of questions to allow segmentation of results by work group. But the company was so complex that employees couldn’t accurately identify their group. Some groups ended up with more people than the actual population, rendering the segmented data invalid.

My advice: Less is more, especially when it comes to demographic questions. Make it easy for employees to see their level or role among the options.

Problem: complex questions

I’m part of an insiders’ group for a clothing retailer. (Basically, the retailer bribes me to participate in their research with chances to win gift cards.) I was asked for my opinion on ten different promotional offers. Each was long, complicated and hard to differentiate from the next. I was so frustrated that I gave up and quit the survey.

My advice: Simple is always better. A survey shouldn’t feel like work.

Problem: two-in-one questions

It’s a classic mistake, but one I see often: two questions in one. For example: Was the message clear and compelling? Is your manager honest and trustworthy? The results are inherently flawed, since a respondent may feel differently about each characteristic.

My advice: Split these questions into two for more valid responses.

Problem: vague questions

Every survey has at least one question that’s open to interpretation. For example: Did you enjoy the Town Hall meeting? Employees might respond favorably because the content was compelling or because they got to nap in the auditorium for an hour.

My advice: You can learn more by making the question concrete. For example: Was the Town Hall meeting a good use of your time?

Problem: questions with a biased rating scale

Unbalanced rating scales are a problem. Here’s an example that used a five-point scale, but the options were weighted towards the positive: Poor, Fair, Satisfactory, Good, Excellent. I’m glad I didn’t have to present those results!

My advice: Ensure your rating scale is balanced with an equal number of positive and negative responses. Note: There are scales that focus on one dimension (called unipolar) rather than opposite dimensions (bipolar), but that’s a conversation for another day.