I’ve discussed several times  Public Policy Polling, a Democratic polling firm which heavily spins poll results.

PPP has a split personality; some of its polls are straight up and in actual elections it seems to do as well as many other polling organizations.   But PPP also engages in messaging polls, which help shape the debate.  And on its blog, PPP spins its polling furiously for Democrats.  PPP wants to have it both ways, being neutral professonal pollster and advocate; it’s a tough act to pull off.

It seems that every time I bother to look at the numbers behind a PPP poll, I find that they over sampled Democrats.  While I have not tested a majority of PPP polls, in the times I’ve looked at their numbers I’ve never seen a sampling which reflected party-affiliation reality or which oversampled Republicans.  Maybe I’ve just had bad luck, or maybe there is a problem at PPP.

So it is no surprise that Ed Morrissey at HotAir finds the same problem in a PPP poll purporting to show Obama doing well in Colorado:

“Is Barack Obama in trouble in Colorado?  A new survey from Democratic pollster PPP tries to allay fears that the President has lost significant ground in the interior West state that Democrats worked hard to convert in 2008, but its sample is so skewed that it ends up proving the opposite….  This poll goes beyond bad, into self-parody.”

The cumulative effect of such polling is to create a self-fulfilling prohecy, to rally the troops, to keep Democratic hope alive.

Yet, I continue to see conservative blogs link to PPP polls.  I’ve even done it in the past.

Because PPP is so good at driving news and because it comes out with new polling almost every day, it’s hard to ignore them.  It’s also hard to keep up with them because they literally swarm the news cycle with polling.

So don’t ignore them, just check their sampling … and link to Morrissey’s post every time you do.