Previously, we published a blog post that explained the practical and well established ways that response bias can interfere with getting quality data. Response bias comes down to the fact that we all struggle at times to answer survey questions accurately for a variety of reasons.

This blog post intends to follow up with some practical best practices for avoiding response bias in your next survey or questionnaire. The fact is that response bias comes down to subtle human error, so it can't be totally eradicated. However, there are a few things that you can do to keep it to a minimum so that your data is accurate.

‍

1.Make sure that your language is appropriate for your audience.

If you’re creating a science survey for the general public, don’t use overly scientific or technical terms reserved for the academic community. Make sure that your questions are clear and that your language reflects the kind of language your target group uses.

‍

2. Don’t make the mistake of asking two questions at once

It’s hard to answer “How did you like the food and the entertainment at the annual gala on a scale of 1-5?” when you thought the singing and dancing was awesome, but the chicken was raw in the middle and gave you food poisoning.

“How was the food at the gala?” and “How was the entertainment at the gala?” are two separate questions. You can see why combining them won’t give you accurate answers to either.

‍

3. Avoid inherent bias in your questions

Make sure that your questions don’t already have implications or bias due to their wording. A leading question sways the respondent to answer one way or another because the wording isn’t neutral.

‍

Let’s say that your firm manages a large apartment building in New York City, and there is a safe (but ancient) elevator that you are considering replacing. You aren’t sure if it’s budget-wise to replace it now or wait a few years. It’s time to ask the tenants what they think about the elevator to decide if it’s necessary to invest in a replacement at this moment.

Don’t ask: How bad is the old elevator in the F wing?

Ask: How would you rate the elevator in the F wing? (scale of very poor to excellent)

‍

If you ask how bad the old thing is right off the bat, then you automatically plant the idea that it is both old and bad. Try to use neutral words and avoid judgemental adjectives in your wording so that people have a chance to give answers that are as fresh and authentic as possible.

‍

4. Do your research and provide enough options

Not providing enough options is a great way to get skewed results. People will quickly abandon surveys that they can’t answer. If they don’t simply abandon, they might also provide false answers just to skip questions. There are ways to avoid this.

For one, make sure that you do your homework before you create your survey. Gather all of the information and opinions on the topic that you can, so that the options that you provide are as relevant as possible.

Know that no matter how much research you do and how superb your survey may be, there will always be questions that some respondents can’t or don’t want to answer. Respect and understand that, and you are on you way to collecting more accurate data.

Always make sure that respondents have the option to skip a question or to answer “I don’t know”, or “this doesn’t apply to me”. You are guaranteed response bias if you force people to answer questions that they can’t or don't wish to answer.

Remember that people are usually doing you a favor just by taking the time to take your survey. Make it easy for them.

‍

5. Make sure you target the right audience

This is like the basic concept of quality over quantity, and it ties back into doing your research. The more personalized and targeted your survey is, the more likely people will be to complete it accurately.

Let’s say you are looking to improve your senior living facility, and your target group is people 80 years and older. If you have proudly advertised your survey all over social media and gotten 10,000 responses, then think again. While 10,000 people may have answered your survey, half of your target group doesn’t ever use the internet. There is a good chance that the people answering your survey aren’t really the people you need to be hearing from.

Research your target group and understand what channels they are hanging out in. Then target the right people in the right places, and you’re on your way to getting a data pool from the people that actually matter for your results.

‍

‍