Polls and E-mail Surveys
Some years ago, the Myrtle Beach Sun News was looking for a way to tap what was on the minds of residents in that fast-growing resort community but couldn't afford a formal community survey.
The Internet, with its survey tools and polling software, was only just beginning to make inroads into newsrooms. So, the newspaper went with another simple, low cost tool: It distributed neon yellow postcards with six questions and asked people to write in and mail the cards back to the paper.
One of the questions: "What really makes you mad right now?"
The responses were not scientific, but they were passionate and generated a useful roadmap for reporting on the community. A large number of the respondents agreed on an answer that surprised some editors: "Tacky beachwear stores," they proclaimed.
The Sun News' query, while successful in its way, had the same shortcomings as most web polls, It was not sent to a scientifically representative pool of possible respondents; the respondents were self-selected; and they could have sent in more than one post card, if anyone felt like it.
And like most web polls, it worked best as a starting point for a story about mounting tensions in the community — not the end story in itself.
For community websites, some of the best uses for online polls and e-mail surveys are to get an early sense of issues that are surfacing on your residents' radar screens. That sensibility can then be used as a fulcrum for reporting a full-fledged story, to take the community's pulse on a particular issue, or to leaven discussions in your forums.
Various local news sites have used polling techniques to ask users to:
- Participate in a contest; say, to name an initiative.
- Brainstorm solutions to a community problem.
Others take their "polling" offline and build a database of user contacts and information so they can survey people informally via e-mail.
"We use our reader e-mail database all the time to survey readers informally for anecdotal information to help flesh out a story," says Ken Sands, online publisher of the SpokesmanReview.com. "In fact, I've now run a number of nationwide surveys using a network I've built up of newsrooms around the country that can send out the same e-mail to readers, directing them to a central website for results to be collected."
"We've purposely asked for essay responses so that we get quotes for stories. But we've also found that we need to have some quantitative 'yes-no' types of questions to get some sense of proportion without having to read every single essay question and figure out how it should be tallied."
"No, these surveys are not scientific, so they are not statistically valid," he said. "But for purposes of getting a pulse of the community, they're very valid."
One tip: If you are a small community site and want to do a formal poll in your community, think about building some useful partnerships. You can reach out to such partners as a local journalism school, the survey research department of a local university or the marketing department of a nearby business school. They may well collaborate on a community/class project.
Of Quizzes and Polls
From a programming point of view, a quiz is a just poll with special "correct" answers, though there can also be differences in the way you compare answers, keep track of responses, etc.
One other way that quizzes and polls might vary is that you might want to structure a quiz question by question, so that a user answers one question and sees the result before they proceed to the next question.
And of course, you should note that people LOVE to "cheat" on quizzes, taking them multiple times or trying to figure out how to get a perfect score, so treat the results with at least twice the skepticism you do your poll results.
A 2003 poll by the Pew Internet & American Live Project found that 44 percent of adult Internet users had contributed content online in some way — to respond to something, to publish their thoughts, to post pictures or share files. Another Pew poll in 2004 found that 26 percent of adult Internet users had rated a product, a service or a person using an online rating system. So a poll may be one way to capture some of this willingness to contribute.
Polls are a good trigger for interactivity. Asking a question and presenting a simple button to click "yea" or "nay" prompts a response far more effectively than a simple link that says "Participate in our site forums." Providing a link from poll results to a discussion forum might also convince a few extra participants to join in.
Polls are graphically interesting. On a site that's text heavy, polls can provide a compelling visual element and an entry point to a story or a page for people who are just scanning your site. See the simple poll The Forum, of Deerfield, N.H., used for the opening of its new Farmer's Market. http://www.forumhome.org/
Polls can be good for page views. Depending on how you construct your poll software, every poll vote can trigger a new page view. Ditto for when a reader wants to view the results – they'll usually have to load a new page to do so. And of course, if someone likes your polls and wants to see the results of some other polls, they'll have to load more pages.
Do note that with modern CSS and AJAX techniques, it's possible to create polls that display within a single page and require no extra page loads. How you decide to implement polls is up to you, and readers might appreciate not having to reload everything each time they vote.
What is AJAX?
Polls can be a stand-alone feature. Not all polls have to be associated with a story. Sometimes just the questions, and the responses, can be interesting.
Polls can be fun. The truth is, polls can be a place to inject a bit of levity or serendipity into your site. You don't want to go overboard with this but there's still a chance to amuse your readers and yourselves with a poll. Maybe you can get a local historian to contribute a fun-facts quiz about your community and solicit a local business to donate small prizes for the winning answers.
Among the biggest risks of Internet polls are that users will flood the poll with multiple responses and that users will think that web polls are statistically valid.
Self-selected polls, even if they are completely tamper-proof, are simply not valid indicators of the community at large. Professional pollsters spend much time and money trying to get a representative sample – old, young, rich, poor, ethnically diverse — to answer their questions.
A web poll is limited to a rather narrow sample of people who can afford a computer and Internet service, who are interested in your site, and who are motivated to respond to your poll.
Online, you need to state clearly that poll results are not scientifically valid and make sure that your poll viewers and participants realize that the results should not be taken as anything more than the views of a few like-minded individuals.
As for poll tampering, it can, and probably will, happen. Polls that hit on particularly polarizing issues or that measure the popularity of anything with a strong fan base will invite attempts to skew the results. For example, "What's your favorite TV show, operating system or political party?" would probably see some attempts to "stuff" the ballot box. You can, however, take some steps to limit poll tampering.
How to Minimize Tampering
If you allow only registered users to vote in your poll and you tie those votes to the user name, you can cut down on most tampering. A drawback, however, is that many people won't register only to vote in a poll so it helps to have other offerings or reasons for someone to want to register.
It should be noted, however, that registration itself is difficult to make tamper proof. Requiring a valid, unique e-mail address is one step.
You can also decide to filter excess poll voting by setting a cookie each time a person votes. If the voters block or erase their cookies, they would be able to vote again – so this is not a perfect solution. It is quick and transparent to the user, though.
If you store each vote along with the voter's IP address, you can block repeat votes from one IP. This overcomes the problem of those who erase their cookie so they can vote multiple times. However, this also blocks groups of people connecting from behind a single firewall, such as people working in the same office or from a university server. They share an IP and so might not be able to have each person's vote count.
In truth, while you certainly want to put in place the best tamper resistance you can, it's just as important to monitor your polls and simply take down ones that seem to be showing evidence of foul play.
When it comes to showing the results of a poll, there are a few formats to consider.
Raw numbers in a chart is nice for those who want to analyze data, but the numbers themselves can be overwhelming for others.
Will you be buying the new Harry Potter book? Yes 217 25% No 454 53% Maybe 66 8% I'll Borrow It 117 14% Total 854 100%
Line charts often make the most sense and can be used for almost any format of poll question.
Area (pie) charts are trickier to generate but tend to break out answers most clearly for basic questions where the respondents can only choose one of a set of answers.
Really, what you need is a poll solution that works with your existing hosting, and that has some sort of protection against people unfairly biasing the poll. Beyond that, most polling solutions are fairly similar.
Most modern discussion-forum tools have built-in polling features, including the forums on this site. This is because of the natural affinity of forums with polls as well as the relative simplicity of adding polls to a system based on user registration.
Survey software can be purchased and loaded on your web site's server. A web search for "online survey software" will yield some candidates.
For stand-alone poll solutions, there are many open-source solutions although most will need a little coding work to be integrated into your site. A search for "PHP" or "ASP" (or whatever your site's programming language is) and "poll" will bring back some likely candidates.
There are also several websites that will host a survey for you. For instance, Blogpoll.com and Bravenet.com offer fairly robust tools, which you can use if you don't mind giving them a little link on your site. Other sites will charge you for this service. Two of the more popular survey vendors are Zoomerang and SurveyMonkey. You can read more about surveys here.
Finally, your web hosting service may also offer polling services as a feature of its hosting plan.
> READ NEXT ARTICLE: Searchable Databases