You Be the Judge
When Internet pollsters push too far.
By Barb Palser
Barb Palser (email@example.com), AJR's new-media columnist, is vice president, account management, with Internet Broadcasting.
THINGS LOOK PRETTY BAD for this guy, all right.
For 45 minutes a queue of friends and enemies describe the man charged with murdering his estranged wife. The TV narrators solemnly explore every possible scenario and piece of speculation, shoveling evidence into piles on either side.
Meanwhile a "guilt gauge" ticks ever upward in the lower-left corner of the TV screen. After the first few minutes, the numbers turn toward the "guilty" side; by the end of the show 57 percent have reached that verdict.
The votes are not being cast by judges, jurors or investigators. This man is being tried in the court of public opinion by mildly informed and vaguely interested upper-middle-class Internet users who will presently log off, tune out and go to sleep. Case closed.
Though the ploy is fascinating, one wonders what news purpose is served by a perpetual notice that the majority of unqualified strangers think this fellow is guilty. Moreover, what harm might be done to his reputation (should he be acquitted) and the integrity of the program itself?
Talk shows and newsmagazines like "Dateline NBC" have been toying with interactive survey tools for a few years, in various contexts and formats. In the above example, "Dateline" viewers were invited to visit MSNBC's Web site during the show to vote on the man's guilt or innocence. The results were displayed graphically on the screen in "real time" during the February 29 broadcast.
This experiment was somewhat unconventional--if there is such a thing as convention in the wild world of media convergence. Normally the results of live Internet polls aren't presented until the end of the TV show. Also, they tend to focus on policy issues or subjective opinion--from Super Bowl commercials to gas prices.
There is definitely something surreal about injecting a totally unscientific opinion poll into an ostensibly objective news report. When Web surveys are cited in the body of a story or a broadcast--even with explicit qualifications--integrity and objectivity are in peril.
Fortunately that is not yet common practice. Most networks' Web sites do freely acknowledge that their surveys are "not necessarily representative of the general population," as MSNBC.com notes, and that they "should not be confused" with the results of scientific polls, as CBSNEWS.com says. But does qualification lend license? Does the average viewer--or producer--truly understand how unscientific they are?
It's widely known that online surveys are demographically skewed, since participation is proactive and Internet users--particularly those who visit news sites--are disproportionately white, middle-class college graduates.
Also consider the duration of time the survey is available. While some begin and end during an hour-long show, others collect votes on a developing story for months. The result represents the cumulative ebb and flow of opinion instead of a snapshot at a specific point in time.
Another variable is sample size. While some sites suggest that 1,000 votes are as representative as 20,000, others fail to reveal whether five people voted or 500.
Nor should one assume those 500 votes are coming from 500 individuals. At worst, an online poll has no security checks at all. But a survey can be programmed to accept one vote per IP address--the electronic identification number assigned to each computer. That won't stop a savvy cheater from changing his or her IP to vote many times over, or a not-so-savvy cheater from voting several times from different accounts.
Finally, there is the phenomenon that people prefer to think what others think. It's why visitors should be required to register their own responses before learning how their neighbors voted.
So why use online surveys at all?
They're a free, fast and entertaining way to get the audience involved. They also affirm the premise that some people really do surf the Web and watch TV at the same time.
The harm comes with the temptation to pass off Internet polls as accurate gauges of public opinion. We should never see informal Web surveys presented without qualification.
Thankfully most news organizations honor that distinction. The week following the removal of Elián González from his Miami relatives' home, MSNBC, CBS, ABC and CNN all featured opinion surveys in their online coverage--all of them formal studies rather than Internet polls.
That doesn't mean that these Internet polls have no place on news sites. Creative Webmasters at MSNBC invited visitors to register their own views on the González saga, then compare their votes with the results of a formal NBC/Wall Street Journal poll. This idea should be applauded for showing the positive potential of media convergence in matters of opinion.###