Suggestion: Modelling survey result


Hi, there! 


When I build the survey I often try to imagine what kind of result I’ll get. Actually, I just download empty survey list with only qestions and fill it random answers. After filling I build the plots and try to evaluate either it’s usefull or not. 

Of course I can evaluate without such doing, just in my mind, but it is not convenient.

Maybe someone else will find it as usefull features and surveygizmo will consider it for a developing plan. 


seanmcnally380846 answered
    Lola Gill 6.99K Rep.

    Maybe I’m misunderstanding, but are you looking for testing options for your survey?

    SurveyGizmo does have an option to quickly and easily Generate Test Responses:

    I use this all the time to get a sense of how my survey data will report – there are some limitations, but I think it’s a great option for previewing your survey results.

    Test data can also be deleted or filtered out of any reports once you have real data coming in.

    Not sure if this helps, but I just wanted to mention it in case this is what you are looking for.


    pushvintsev commented
      • Yes, sorry, I just didn’t notice it. Before it I thought tap “Test” may only tell you about time respondent spends on your poll. Now I see this feature.

        It’s quite handy, but it gives me uniform distribution, but talking about polls, normal distribution you can meet more often. So, to continue conversation of suggestion theme, It’s could be better to choose or only to get normal distribution on getting test responses.


      This is when piloting and pre-testing come in handy.  For all studies I pilot with at n=25 actual respondents before going to full launch.  For more complicated and longer term studies n=100 is more of a minimum.

      If you work with others, then have them pre-test and pilot the survey with you.  


      Simulation of data are possible but they are dependent on you knowing the likely distribution of responses.  When logic skips are involved and survey timing is critical, this is a big assumption.  

      To just get data to test reporting, exporting, coding, etc, then the SurveyGizmo Generate Test Responses is of value.

      pushvintsev commented
        • Thank you for attention to my question.

          I try to test it on real respondent if I’ve already constructed the survey. But this feature more usefull while building survey, just to see shape of result.

          And yes, actually you’re right. That’s true that generator doesn’t know about distribution and also gives you uniform distribution that unusuall for social surveys — it’s more likely to get normal distribution. Anyway, it’s better if we can choose the distribution.



        Question stats

        • Active
        • Views7551 times
        • Answers2 answers
        • Followers1 follower