4 Responses to “Um…I Really Can’t Say”

1. DS Says:
   March 3rd, 2006 at 14:16

   If the undecideds don’t vote, the sample taking the poll is artifi­cially compressed, and the percent­ages of yeses and nos end up being more impres­sive than they have a right to be. “Can’t say” *is* useful infor­ma­tion from the poll-taker’s point of view.

2. Ramnath R Iyer Says:
   March 4th, 2006 at 00:25

   That presumes that the popula­tion partic­i­pating in the poll is a fair sample of the electorate in the first place. This is unlikely to be true if the poll is on the Internet (which is avail­able to a specific stratum of society) hosted on a partic­ular site (which restricts it to a popula­tion with specific inter­ests). A smart poll-taker would not rely on the results of this poll for mission-critical decisions.

   If we assume that the sample is a fair distri­b­u­tion, then I opine that the ratio x = YES:NO is a better indicator of opinion that YES:(YES + NO + CANT SAY), because it indicates direct opposi­tion. The question of artifi­cial compres­sion arises only in the second case, when the results are indicated in percent­ages. Instead, a graph­ical repre­sen­ta­tion could be a single bar with YES at y = –1 and NO at y = +1. The bar should be divided into two parts, based on the ratio x. If x > 1 then choose y, the point of division, based on a bounded function f(x) that maps [1, inf) to [0, 1), and if x < 1 then f(1/x) that maps [1, inf) to [0, –1).

   Also, a poll that doesn’t contain a “Can’t Say” forces the user to take a stand, rather than take the easy way out. I might be wrong, but I don’t think not having a “Can’t Say” option would dissuade the user from taking the poll. So this might actually suggest to the poll-taker which option has an edge over the other, in terms of support.

3. Sridevi Says:
   March 4th, 2006 at 06:02

   As you say, the sample is already restricted to that section of the popula­tion that has access to that website, and accesses that website. Of the sample which does consider replying to the poll, there are people who are inter­ested in the topic, but not sure of the answer. If you restrict this section of the popula­tion as well, then you have really narrowed down to only those people who have an opinion. Marking the “can’t say” option could also be an indicator of the wording of the question. Too many “can’t say” answers could imply that the question is too ambiguous or not very precisely framed, or maybe it is too precisely framed.

   Some examples of polls:
   1. Do you think right to privacy should always be protected? Yes/ No/ Can’t say. In this case, i’d mark “can’t say”, because, I’d say it depends on the situa­tion. For instance, if an actor comes up with a contro­ver­sial state­ment in his house criti­cising some politi­cian, then it’s private opinion. The person he made the comment to can object to it. But, if the same actor makes the same state­ment on the Oscar red carpet, then it’s not a private comment anymore, since he has done it on a public stage. Hence, I cannot mark a definite yes or no answer.

   2. Is Kareena Kapoor more popular than Karishma Kapoor?
   Ans “can’t say” Who is keeping track of the popularity meter? How is the popularity to be measured?

   3. In a study on the soft drinks market, Would you rather drink water or Pepsi?
   Ans “can’t say” because, in some situa­tions like after some heavy work, i’d rather drink water. But if I’m sitting in front of the TV, I’d rather have a Pepsi.

   So, you see, when a poll is conducted, it is impor­tant to do what is called a ‘pilot study” to smoothen out these imper­fec­tions in the question­naire. There can be lots of cases when the question seems obvious to the qt’naire designer, but not to the respon­dent. This is something you’ll under­stand better, when you actually do some surveys, for e.g in a marketing research class.

4. Ramnath R Iyer Says:
   March 4th, 2006 at 06:48

   If I restrict the discus­sion to polls in which the poll-taker hopes to gain some useful infor­ma­tion, then a “Can’t Say” option only indicates that the question has not been framed properly — because it does *not* provide any useful information.

   Let’s say you are conducting a survey. If you think that a partic­ular question may need to be answered with “Can’t Say” then reframe the question.

   (1) Do you think privacy should always be protected? ‘Always’ is a strong term to use because you are clubbing together distinct opinions. “Sometimes” implies “Not always”. “Never” also implies “Not always”. Instead the question could be: do you think privacy should be protected? (Always/Depends/Never)

   (2) Is Kareena Kapoor more popular than Karishma Kapoor? This couldn’t possibly provide any useful infor­ma­tion because instead of giving you an opinion, the voter can only tell you what she thinks *might* be the situa­tion. Wouldn’t it be better to ask: whom do *you* like better? Many answers to this question can be used to figure out what general opinion is. There’s no point in answering “Can’t Say” to the second question.

   (3) Would you rather have water or Pepsi? You haven’t defined your target audience! The people whom you are asking, are they Pepsi drinkers in the first place? What is the situa­tion, summer or winter? Why not first ask them how often they drink Pepsi, and then ask them if they drink it because of thirst or taste? (Mostly Thirst/Mostly Taste/Either)

   In some cases, if it doesn’t seem possible to supply distinct options, then the second best thing would be to give slightly “fuzzier” options, like “Always/Most of the time/Some of the time/Never”.

   I am aware that there is a science to the whole business of taking polls. Writing a good question requires a good under­standing of the infor­ma­tion that you stand to gain by having the question answered, as well as an under­standing of how much infor­ma­tion the audience is in a position to give to you. Not having a “Can’t Say” does not mean that the question should be permitted to be unanswerable.

Leave a Reply

Name (required)

Mail (will not be published) (required)

Website

Notify me of followup comments via e-mail