Poll Shocker: More than half want to change DailyKos operations!

In a stunning setback for advocates of DailyKos.com's current handling of registration, hidden comments, and troll ratings, more than half of DailyKos regulars want to change the system, according to a recent online poll.  

The poll, "Should DailyKos become invite-only", was posted by longtime Kos fan and diarist "cartwrightdale", and received hundreds of votes from the community.  Opinions ranged from advocating an "invite-only" system, to closing registration entirely until 2009.  

Shocking!

Now, let's look at the actual results:

"Well, wait a minute," you might object.  "The first two paragraphs of this diary made it sound like dissent was rampant, whereas in reality, the most popular response to the poll was to keep everything as-is, and the second most popular response was, at best, a minor site tweak!"

"Ah," I would reply.  "But only 48% advocating keeping the current system, which is a minority, and therefore it's true that more than half wanted changes, and the opinions I mentioned were, in fact, chosen by some Kossacks."

"But that's ridiculously misleading!!!"

Exactly.  

The way the news media reports on polls is frequently identical to my intentionally misleading headline and opening paragraphs.  How, say, Fox News reports on a poll (or anything involving numbers, such as troop deaths) vs. how NPR reports on a poll can be wildly different, giving completely different perceptions to the reader/listener/viewer.  As seen in the recent Joseph Curl example, you can push any agenda you like by playing with the numbers to meet a preconceived conclusion.

A more accurate way of defining the results of my poll would have been "Kossacks overwhelmingly reject big changes to dailykos.com".  But both this definition and the "more than half want change" definition are technically correct despite appearing to be opposites.  

Suppose a poll shows Obama at 25%, Clinton at 23%, Edwards at 22%, and everyone else in single digits, with a +/- 1% margin of error.  The following reponses would all be, in their own ways, "technically correct", but most are purposely forcing a different perception upon the reader than an objective reading of the poll would allow:

"New poll shows an overwhelmingly majority (75%!) do not want Obama to be the Democratic nominee)!"  (True, but in a "the majority of Americans voted against Bill Clinton both times!" kind of way.)

"New poll shows Obama has 10% more supporters than Clinton!"  (After all, 10% of 23 is a rounded 2 -- the point difference between the candidates.)

"New poll shoes Obama and Clinton virtually tied for first, but Edwards second!" (Misleading, since Edwards and Clinton are even closer than Obama and Clinton.)

"New poll shoes Obama in first, but Edwards and Clinton virtually tied for second!" (A little better, but still overstates Obama's lead.)

"New poll shows a three-way-tie for first place!" (Not REALLY.  The difference between Obama and Clinton is within the margin of error, since +/- 1% means Obama could be 24, and Clinton could be 24, which is a tie.  But Obama and Edwards are outside the margin of error, so "three-way-tie" is an inaccurate approximation.)

"New poll shows a wide open Democratic race!" (Well... among the top three candidates anyway, but wording it this way makes it seem Dodd or Gravel has a shot too, which isn't supported by the poll.)

And so on.  You could probably make a million headlines based on those Obama-Clinton-Edwards numbers, all of which could lead a reader to interpret the poll said what you wanted them to think it said.  

Polls can be useful indicators of public opinion, but we tend to go crazy over minor blips and anomalies here, all while trying to push our own agendas.  Sometimes we don't even know we're doing it -- when you genuinely believe in something or someone, you read into the numbers.  Mr. Curl in today's example was likely trying to intentionally mislead his readers into thinking Bush's poll numbers were surging more than reality suggests, but it's also likely that he and other Bush defenders saw a 29% poll one month, then a 36% poll later, and thought "hey, Bush is going up!" without bothering to check the details.  We see what we want to see, and then, through our language, try to make others see it that way too.

The devil, as they say, is in the details.

(One final note from the poll from the previous diary:  Some of the commenters thought that I was advocating making the site invite-only, by nature of the poll question.  But if you read the diary, I was clearly NOT for making the site invite-only, and indeed voted to keep the site as-is.  Another sign that how you ask the question can greatly impact people's perception of the question.)