Wed, Jul 22, 2015 page8

• Keeping opinion surveys accurate

  • By Jessica Chou 周韻采
  • • 
    • 
    • 

  Election fever has recently started heating up. We are hearing of the results of opinion polls from this organization or the other, all trying to predict how much of the electoral pie any given candidate is likely to get. The trouble is, the support they are reporting is all over the place, as if the electorate has no idea what it really thinks.

  All of this creates the suspicion that the polls might not be all that accurate.

  It does seem that polls are becoming less able to accurately predict the outcome of elections.

  The biggest problem is that the information in the databases pollsters have access to is increasingly at odds with the real situation on the ground. Polling companies get their base samples using landline telephone numbers, but fewer people, especially the young, still use a landline number. This means that a whole section of the population is not reachable for purposes of survey sampling. This results in sampling error, and the situation is only going to get more serious.

  Is there any way to resolve this problem of accessible information becoming increasingly out of touch with the population? Of course; we can look for a new source of information that can more adequately represent the population — cellphone numbers.

  Not only is the take-up rate of cellphone numbers in Taiwan incredibly high, the numbers are owned by individuals, making them a much more suitable source of data for opinion polls than landline numbers.

  However, academics and polling companies attempting to conduct surveys using cellphone numbers have encountered all kinds of problems because privacy laws prohibit telecommunications providers from giving out phone numbers without good reason. This makes it very difficult for researchers to create databases from which to take samples of the population.

  Given these restrictions, the only answer would be for some kind of cooperation between telecom providers and polling companies, in which polling would become part of the former’s remit. Telecom companies could first obtain the permission of their users before collecting their data, which would then be used for survey sample purposes.

  Many users would probably not be willing to make their data readily and openly available. This means that telecoms would have to provide some form of incentive, such as reduced fees, which would also enable the users to share in some of the profits from the polling side of a telecom company’s business.

  Then there is the question of whether the telecoms would be able to guarantee that the users’ personal data would only be used for the purposes that the user had authorized. It is likely that privacy protection advocates would shake their heads at this scheme, saying that it violates users’ privacy rights.

  However, personal data are a kind of personal property, and it ought to be up to the individual concerned to decide what happens to their personal property, and to get something back from authorizing a third party to use that personal property. At the same time, a company paying to use this information would also be making commercial profit, which would be of benefit to society.

  I would tend to trust telecoms with my personal data more than I would Internet companies or social media: After all, they have been subject to many years of oversight and there is a high degree of consensus within the industry about following the rules.

  The same problem with data being non-representative of the population has also emerged with regard to television ratings. The Nielson Company has often come under criticism for its sample set being out of touch with the actual make-up of the TV audience in Taiwan.

  There is an answer to this problem now that TV has “gone digital.” Since the viewing behavior of the digital TV audience is recorded in the digital set-top boxes, all that is needed is for cable channels or Internet Protocol TV companies to publish viewing figures for different channels on a regular basis, and we would be able to see which channels are the most popular and which ones are just taking up space, with nobody watching them. This information would be invaluable to the National Communications Commission’s efforts to improve the quality of TV programs and to restructure the platform.

  Any objections to digital TV operators disclosing channel viewing figures on the basis that such a move would violate individual or corporate property rights is misleading, as these ratings are the total viewing figures and have little to do with the individual or corporate confidentiality mentioned in the privacy laws. Therefore, the commission could require that operators provide on a regular basis — perhaps monthly — a report on viewing figures, which could then be published on the commission’s Web site for the public to access.

  Now that we are in the digital age, providers have to move with the times and meet the challenges of the new media and new platforms. There is a whole new world out there.

  Jessica Chou is a professor of Information Management at Yuan Ze University.

  Translated by Paul Cooper

  Comments will be moderated. Keep comments relevant to the article. Remarks containing abusive and obscene language, personal attacks of any kind or promotion will be removed and the user banned. Final decision will be at the discretion of the Taipei Times.
• Most Popular

  • 1

    German warships to pass through Strait

  • 2

    Sino-Russian air patrol designed to counter US

  • 3

    Court orders Ko Wen-je’s detention

  • 4

    Silicon photonics alliance formed

  • 5

    Ko Wen-je release quashed