Polling data: unsure and outliers

I often go to Real Clear Politics website to get the latest opinion polls on Obama's approval rating. I think they do an admiral job at trying to remain bi-partisan in their reporting of the many polls taken on politics and the president. However, there are two things I have come to conclude:

1) Some polling groups come out with some WILDLY out of norm responses. 
2) There are still a lot of "not sure" people out there.

Out of norm polling data
As to the out of norm responses, what I mean here is that in the world of statistics you have what if commonly referred to as "outliers".  These are statistical anomalies that should be thrown out if their are outside of the statistical norm. Let me first say, there is no mathematical algorithm for determining an outlier and it can be quite subjective (you know it when you see it).  For example, lets say we have the following numbers reported: +2, +1,0,-2,-3,+10.  The statistical variation of the first 5 is  14/6 = 2.33 so therefore +10 is outside of the variation since its closes neighbor is a difference of 7 (10-3).  It should be considered to be outlier and therefore rejected as a acceptable poll.    About a month ago, such an anomaly did occur as MOST of the polls for Obama were negative (or trending negative), yet out of the blue, Bloomberg showed a +9 rating for the president.  

Below is a screenshot from RealClearPolitics website showing a sampling of the various polls taken around the 6/15-6/18 time frame.



As you can see, most of the polls at that time showed approximately a -1 rating for Obama.  But LO AND BEHOLD,  Bloomberg comes to the rescue and gives Obama a +9 bump, sending the average from a negative (-2/5 or -0.4) to a positive level of  +7/6 (  1.16 ).  (Personally, I hope someone in the Obama administration sent the Bloomberg polling group some flowers and candy in thanks for their gift to the president) 

Not sure ??
The next issue I see that is not taken into account is the high number of "not sure".  Frankly at this point (3.5 years into his administration) we as Americans should have SOME OPINION as to how we think the President is doing.   To me, you get a WHOLE different view when you take into account the "not sure" and place them in the "disapprove" category.

Below is a snapshot from www.realclearpolitics.com on Friday 7/13/2012 at 10:00 am PST






If you took at missing "not sure" as "Disapprove" you would get a rating of  -5.5 instead of -1.3 (a much more clearer view on how the President is REALLY doing).






Note: In my opinion, all Presidential polls should only offer Approve or Disapprove as their only two choices.  We cannot be all "mamby-pamby" when it comes to the direction our country is going.  Either you are FOR HIM or you are AGAINST HIM.  So .... MAKE A CHOICE!

In conclusion, let me say I am not putting down Real Clear Politics fr their coverage of the President's approval rating.   Like I said in the beginning, overall I think they do some good work and try not be partisan in their approach. and so they just take the numbers that are given them.   But we as citizens need to understand statistics better and see that there is often more to the numbers than just averages and spreads.  We need to use also our gut instincts as well