Find us on Google+

Monday, 17 December 2007

Polling the obvious?

In case you have missed the news - The Post is now running "opinion polls" via SMS. I applaud the Post for attempting to gauge public opinion on various issues, but unfortunately the method chosen (SMS) is likely to lead to "self selection bias". Its my view that The Post's chosen method is most likely to elicit responses from those that feel negatively by government - or are seeking some form of change. I mean seriously, how many people would spend time and money to say something positive about any government?

10 comments:

  1. That's why I wouldn't take any opinion poll seriously. And while SMS polling may be 'ingenious', considering the disposable income factor in the Zambian environment, the poll is already skewed.

    ReplyDelete
  2. Even if the response is only from people who take issue with government policy, the strength of the reaction is also very useful to know.

    ReplyDelete
  3. Well, the people who text and the people who vote are not necessarily the same (in Zambia), as we observed from the past general elections where the President was voted in primarily by the rural people who are highly unlikely to participate in any SMS poll!

    ReplyDelete
  4. I think if, and its a big if, we accept that SMS was a valid approach, the income issues raised by Zedian can be overcome by providing the results by province. By that I mean, we can view the rural results simply as results that rely on a low sample...

    On MrK's point about "strength of the reaction", I think that would only be true if a larger proportion of mobile users took part!

    ReplyDelete
  5. ...we can view the rural results simply as results that rely on a low sample...

    Cho,
    That's an acceptable approach. The important thing is to take samples from various regions, not simple figures.

    ReplyDelete
  6. ...the income issues raised by Zedian can be overcome by providing the results by province...

    If the Post is actually thinking of doing it this way, how will they implement it? Two methods come to mind:

    1) Provide different access numbers for different regions, which is the simpler one, though does not guarantee that the SMS came from the purported region.

    2) Collaborate with the network providers to identify texts from different regions; more complicated but accuracy is guaranteed.

    ReplyDelete
  7. Perhaps a telephone poll?

    The Post can call people, that way they the can easily control the "geography".

    I suspect the Post are trying to do this on the cheap...

    ReplyDelete
  8. Yes, the Post probably wouldn't want to make this a major cost centre, hence unlikely to be the ones making the call to people and footing the bill, however much they want the data.

    Another issue is that landlines are almost non existent in people's homes in the rural areas.

    ReplyDelete
  9. Yes, the landline question is the key.

    There's no substitute for asking questions in person.... I actually happen to think its cheaper than other alternatives....there's money to be made for the person who sets up the Zambian version of MORI!!! Its the cheapest business idea ever...

    ReplyDelete

All contributors should follow the basic principles of a productive dialogue: communicate their perspective, ask, comment, respond,and share information and knowledge, but do all this with a positive approach.

This is a friendly website. However, if you feel compelled to comment 'anonymously', you are strongly encouraged to state your location / adopt a unique nick name so that other commentators/readers do not confuse your comments with other individuals also commenting anonymously.