Online Surveys = Ongoing Insights

Hide and Go Seek (a Construct)

 → 
 → 
Hide and Go Seek (a Construct)

A question such as “How happy were you with the quality of service at our restaurant?” seems reasonable on its face.  But does the researcher mean happy?  Or satisfied?  And service, in terms of what?  Why not ask several questions for each dimension of quality service.  For instance; “How clean was our restaurant?” and “How prompt was the server in re-filling drinks?”  See how “clean” and “prompt” are two different ideas and, more importantly, how one customer may think about cleanliness while another thinks about drink refills when trying to answer the original question?

So let’s assume you’ve settled on speed of service as what you’d like to gather customer feedback on.  Since I’ve already written in this space that agree/disagree response options are not ideal, you may be tempted to ask something like “How satisfied were you with the promptness of the wait staff?”  But this phrasing has induced an extra step, translating promptness into a more abstract feeling of satisfaction.  A better approach is to rate the speediness of food and drink delivery straight away.  In this way, we might ask “How prompt were the wait staff serving you?”

Check back here for a post on the various types of constructs there are in the world.

Tags: , , , , ,

  • Peter Smetaniuk

    A more precise way I would address the “promptness” statement is have respondents check off “categories” of promptness with the notion of “speed.” In other words, use a likert scale (1 thru 5), where 1 designates “‘less than 2 minutes’; 2 designates ‘within 5 minutes’; 3 designates ‘It doesn’t matter to me”; 4 designates ‘over 15 minutes, but less than 30 min’; and 5 designates ‘over 30 minutes.” I would change the variable statement also, making it more “specific.” “In terms of ‘time,’ how fast was your waiter at serving you at your last restaurant experience?” This variable is now easier to measure the “idea” (dimension) of promptness for statistical analysis. Hence, “The speed of waiter service is a function of promptness.” Plus, when put into your statistical package software (Systat or SPSS), your dependent variable for “promptness” is numeric. See yas.

    • Phil G

      Hi Peter,
      Thank you for your comment on the post. There is a problem, unfortunately, with giving people prefabricated scales as you propose. That is, respondents use the scale to get a sense of “average” and assume that the middle choice is the naturally occurring average. So the respondent first looks to the middle and says to himself “if that is average, am I above it or below it,” in which case an open ended question for time is best. For example “how long did it take…” On the other hand, by asking “how fast…” you’ve eliminated the ability to express “slow” since, it seems, speed expressed this way is a bipolar construct. Lastly, the five and seven point scales I’ve described in the blog post can certainly be used as ordinal variables in statistical analysis software. Let me know if you have any questions.

  • Pingback: Yes, There is a Right and Wrong Way to Number Rating Scales | The SurveyMonkey Blog()

  • Pingback: 質問の隠れた構成要素を探してみよう | The SurveyMonkey Japan Blog()

  • calvin sims

    cleanliness as opposed to promptness.

  • Ievers

    Go get ’em Tiger! :)

  • Peter Smetaniuk

    Thanks for the added format Phil G. No doubt, designing items is an art! Also, what’s your take on including the “middle” or “average” response in a 5, 7, or 10-point scale. I have always had a concern about this response option as giving very little significance to the variable being ex[plored. In other words, when a respondent chooses the “average” it expresses an avoidance of the “extreme” responses–which is not what a survey is hoping to achieve: The authentic response, that is. The big question: Should the average be included or excluded as an indice response within a scale. Thanks, love your feedback.

  • vinay

    Thanks for providing information regarding the question as it will help us to know how rightly we can ask Q.

  • ihsan

    I think the usefulness of “averages/numbers” is dependent on the size/scale of the survey.
    For the “restaurant” example, I think the “restaurant manager” wants to know intimate information for their regular customers (famous and successful restaurants have surveys done on the spot by talking directly to their customers and accordingly tailoring services”. i.e. need to invite contributors to make statements about the various things as they will tell you what is good and bad, in numbers.

    • Phil G

      Hi Hsan,

      Numbers are, as you suggest, only as good as their context. The context of the scale matters, because too many scale points can sap any individual scale point’s meaning, while too few points doesn’t leave room for mapping the true extent of our attitudes and behaviors. The context of the research question matters as well. It is up to the researcher—in this case a restaurant owner—to decide what numbers matter and, more importantly, how much movement in the numbers is tolerable month to month or week to week.

      Phil

  • It is a very good content and learning lesson for fresher to understand the tips and technique about the survey and what questions need to ask.

  • kamol

    very happy

  • THOMAS MOMOH

    Very impressive

  • thnx verry much

Inspired? Create your own survey.

Inspired? Create your own survey.

PRO Sign Up Sign Up FREE or Sign in
Write Surveys Like a Pro

Ever wonder what SurveyMonkey’s really made of?

Ever wonder what SurveyMonkey's really made of?

Read our engineering blog »