Plan what you are trying to find out – think about the answers you want. Keep this plan in mind throughout the whole survey.
Starting with the answers you want, determine question types and write questions that will get you that information. Break down large, complicated answers into smaller questions.
See Purdue OWL: Conducting Primary Research
Question types:
Looking for a count or percentages? Use yes/no, multiple choice, checkbox, drop-down (really just another multiple choice), or ranking (what’s your favorite food? Rank in order of preference).
Want to find averages or test correlations? Use ranking scale (on a scale of 1-5 how clean is our school), matrix (the same but they all use the same scale), textbox (how many times do you eat lunch at school in a week).
Write questions that are:
- Simple and easy to understand
- Only ask ONE thing
- Thorough and give all possible answers
- Unbiased and don’t lead the answerer
- Balanced with positive and negative answer options
What Makes A Good Survey Question
10 Examples of Biased Survey Questions
Create the survey:
Write simple instructions, be clear about the intent of the survey and how the information will be used, consider letting answerers stay anonymous.
Google Forms or Survey Monkey are good options for online survey creation.
Distribution:
- In person, or online?
- Who are you surveying?
- What is the timeframe?
Analyze data:
Try putting raw data into Excel, Numbers or Sheets. If your answers were open-ended try to put results into groups. Try to support your survey findings with other published survey data or interview information.
Don’t confuse Correlation with Causation:
Correlation = a relationship between two things
Causation = when one thing causes another
These are easily confused and misused – be careful! Have another person go over your analysis and pay careful attention to this point.
Classic example = the rooster is convinced his crowing makes the sun rise
Article with more details: Confusing correlation with causation