Diary

Cooking the Books

Usually this phrase is used to describe an organization’s attempt to hide financial losses or problems.  However, we can also use this phrase to describe the lame-brain media’s attempt to mask the losses of the Democratic Party among the general population.  Following last year’s election, the major pollsters at CBS and other Democratic media outlets, decided to alter the face of our political landscape.  They continue, despite a serious slide, to do that to hide Obama’s political marginalization. 

It all started with a simple poll that didn’t make sense.  The most recent CBS poll at Real Clear Politics shows Obama with a 53% approval rating.  Viewed among the other polls, which show him at 50 percent or lower, it is an anomaly.  Anomalies are interesting because they often tell a better story than the norm.  All the polls show a radical downward spiral for the president in particular and the Democrats in general, but this poll has his political support at the same margin he won the presidency.  In this charged political climate is that possible in a fair survey? 

Republicans are almost universally opposed to the president.  Independents are also trending against Obama.  So how can his margin continue to show the same support he received last November.  Its all in the numbers.

CBS weights its numbers by political affiliation.  This is not intended, at least in theory, to mislead but to adjust polling data to fit actual populations.  Even in a randomly selected population, it is possible to survey more Democrats or Republicans than are in the general population.  Since a poll is just a snapshot of a smaller group that is intended to ‘speak’ for the entire population, you don’t want those outliers to misrepresent the whole picture.  So they weight the survey with party affiliation percentages meant to even out the totals to reflect the general population.

Where do they get these numbers?  It used to be that previous raw data numbers would become the weighted percentages after a series of consecutive polls.  In other words, if the previous three polls showed actual Democratic Party numbers at 35% consistantly, then there was a good probability the general population was at that percentage.  If the previous polls showed Independents at 40%, that would suggest the general population reflected that total.

However, the CBS pollsters, and others as well, adjusted the numbers to reflect last year’s voting percentages and voter identification following the election and they continue to weight polls to the advantage of Democrats.  Instead of accepting the changing political landscape, they are still using the older, less current metrics to adjust the data.

Their October raw data showed 27% of the randomly selected population identified themselves as Republican.  Thirty two % identified themselves as Democrats.  A full 41% of the raw numbers represented themselves as political Independents.  However, they weighted the numbers to 22% Republican, 33% Democrat, and 45% Independent.  This had the effect of negating 5 points from Republicans and raising Democrats by a point and Independents by 4 points.  That is a rather striking alignment from Republican to Independant.  But, they may have believed it reflected a more fairly constituted general population than the raw data would suggest.

Then in November, the raw data came out as 27% Republican, 35% Democrat, and 38% Independent.  That would suggest the same percentage of the population continues to identify as Republican.  It also seems that the Democrats got a little bump and the number of Independents sank.  Once again, they weighted the numbers to reflect what they believed was a true representation of the general population.  They deflated the Republican percentage again to 24%, inflated Democrat identification to 37% and Independents were adjusted to 39%. 

But, October and November raw data polls showed Republicans remaining approximately the same.  Democrat numbers increased a bit from 32 to 35, and Independents decreased from 41% to 39%.  But that only shows a range.  You’d think they are looking for a general percentage which would split the differences and not radically alter them one way or the other.  They should have weighted the numbers 27% Republican, since that number held, 33 – 34% Democrats, splitting the difference, and around 40% for Independents.  That would seem to reflect the actual data over time.

Instead, the weighted the data to 24% Republican, 37% Democrat, and 39% Independent.  That, in essence, puts a thumb on the scales for the Democrats to the tune of 6 percentage points.  Given the Republicans are almost universally opposed to Obama and Democrats almost universally support him, the poll should reflect Obama at a 48% approval rating and not a 53%.  That fair representation of the numbers would put the CBS poll back in line with the other polls.  But, the poll pushers at CBS can’t have that.  They must make it appear Obama is more popular than he really is. 

This is not even taking into consideration the other skewed ‘weighting’ mechanisms within the poll.  That would require even more consideration.  On the face of it, these CBS polls are not seriously reflective of anything other than their ideological bias.  And they are not alone.  Gallup and ABC and Harris do the same things.  They simply don’t skew the data as dramatically.  A careful look at the trends as reflected in the numbers tell the story.  Obama is upside down in every poll considered by Real Clear Politics.  CBS is just the cautionary tale.