Creating an online survey today is simple: google a free survey tool, type up the questions you would like to ask and distribute the link. In essence it is simple, but all too often simple mistakes can discredit your conclusions from the survey.
To get to really understand what your data says and to gain real and credible insight from your survey, you first have to make sure that the data itself is credible. After collecting credible data you need to analyze it correctly to get to really understand what the collected data means.
In previous blog posts we have discussed the analysis part. In this blog post, I’ll try to cover a few key points regarding data collection and specifically – how to avoid some key mistakes in phrasing your survey questions.
It is important to keep the following in mind for each and every question you create:
Ambiguous questions/answers
How often do you use your bike?
-
Regularly
-
Sometimes
-
Occasionally
-
Never
In this example no real value can be extracted from respondent’s answers as each choice is subjective. To avoid this - make sure the question incorporates a standard, be it time or usage.
How often do you use your bike?
-
Use almost every day
-
About once per week
-
About once per month
-
Never
Incomprehensible/Unanswerable questions
Incomprehensible: What is your attitude about the increase in gas prices by oil companies due to legislation passed by the government to increase the number of jobs?
Unanswerable: What was the average price you paid for a gallon of gas in October of 2009?
In both situations there may be a reason for asking the question, but if the respondent can not accurately answer, the responses will be invalid and of no use. To avoid this – use short and clear sentences, avoid double negatives and make sure that you are asking questions that can be answered during the survey without deep research.
Leading questions
The government has done nothing to protect your online privacy and your personal information is legally being sold to many large corporations to make money off of you. Do you personally believe this is a violation of your personal rights?
Leading a respondent to your “answer of choice” may be self gratifying when looking over the results or proving your hypothesis, but keep in mind this will completely discredit your results and negate the purpose of the survey in the first place. To avoid this – do not use phrases or statements that might lead the user to an answer.
Loaded questions
You care about your health, don’t you?
A loaded question carries unintended connotations. In the example above, respondents will almost always respond yes due to the negative social stigma of not taking care of oneself and that fact that everyone case about their health in one facet or another.
Double-barreled questions
What is your assessment of the price and convenience offered by Strabucks?
-
Excellent
-
Average
-
Below Average
-
Poor
In this example, two unique and different attributes in regards to Strabucks are asked in a single question. A respondent is forced to answer both attributes as one providing no distinction between the respondent’s assessment of price and convenience. For this question to be valid it would need to be separated into two distinct questions.
Next week we at the groketeer blog will look at how additional subtleties in how you ask questions will affect the credibility and integrity of your online survey data.
