The Times-Gazette web site has a poll feature; in our current layout, there is a poll on the bottom left hand corner of the front page of our web site. A smaller version of the poll, without the option of leaving a comment, appears on the news headline page.

There is also a sports-themed poll on the main sports page; that one is maintained by our sports editor, Danny Parker.

I'm the person who generally maintains the poll on the news pages, and it's not always easy. We try to alternate the poll between serious and not-so-serious topics.

A month or two back, I had two polls in a row on relatively light-hearted topics, because I hadn't planned far enough ahead and the second one was time-sensitive and needed to go right away. Some anonymous voter didn't like this, and left a comment to the effect of "put something serious back on here."

Our poll, like any such web site poll, is entirely non-scientific. A scientific poll, of the kind conducted by, say, Zogby, carefully selects the participants to make sure that they represent a true cross-section of the public (or whatever sub-group the poll is supposed to be measuring). In an Internet poll, by contrast, the voters choose themselves. You have no idea who's voting, or whether or not they represent a true cross section of the public. Most web site polls have some sort of feature designed to make sure that people don't vote multiple times from the same computer, but like anything related to computers and security, there's probably a way around it if you want to try hard enough.

While the votes from our polls are often revealing and interesting to talk about (we sometimes use them as the basis of stories), we have to be careful not to rely on them too heavily.

I try to word the questions as neutrally as I can -- for controversial issues, I word each of the options the way I think the people who support that particular option would like it to be worded. But every now and then, people will object to the wording of a question, and claim that it is intended to draw a given response. I'm only human, and so I've certainly made mistakes in how I've worded poll questions or in what I've chosen to be a poll question. I can honestly say it's not been intentional.

I normally leave the polls up until they have at least 200 votes, although I'll let them run longer in some cases. Usually, though, once you get to 70 or 80 votes, you will find that the percentages voting for each option will stay pretty much the same the rest of the way. If a poll starts to take a sharp turn past that point, I start to get suspicious that someone is cooking the results somehow.

The T-G poll, like the comments which people can leave under stories on our web site, and like our community blogs, is a way to let people have their say. We're pleased to be able to offer polls, but at the same time it's best not to take them as the final word on any topic.

John I. Carney is city editor of the Times-Gazette and covers county government and other topics. His home page is lakeneuron.com.