Some time ago I noticed that after Obama was elected, Rasmussen started using a new statistic on his Daily Presidential Tracking Poll.

The poll has four answers: Strongly Approve (StA), Somewhat Approve (SmA), Somewhat Disapprove (SmD), and Strongly Disapprove (StD).

While Bush was president, Rasmussen did not regularly (if at all!) calculate Presidential Approval Index, but now he does.

While Bush was president, Rasmussen reported explicitly the numbers for all 4 categories. Now he only reports the StA, StD, and the total approval and disapproval. SmA and SmD are no longer reported explicitly, one has to calculate them on one's own.

While Bush was president, Rasmussen usually emphasized total approval percentage, TA=StA+SmA, versus total disapproval percentage, TD=StD+SmD.

However, when Obama became president, he started emphasizing only the strong approval and disapproval, and his presidential approval index is calculated as (S)PAI=StA-StD.

This, I feel, is often misleading. For example, Rasmussen's Obama (strong) approval index for yesterday (August 3, 2009) was -6 since StA=30% and StD=36%. However, if we read past first paragraph of the daily article on the tracking poll, we see that Obama's total approval is TA=51% and total disapproval is TD=48%.

Why not calculate Obama's total approval rating (T)PAI=TA-TD, which would make yesterday's (T)PAI=51-48=+3 instead of -6?

Well, let's see. In the 195 days since Obama took office, his strong approval rating was positive on 154 days, negative on 37 days (and not calculated on 4 holidays). During the same time Obama's total approval rating was positive on 183 days and negative on only 8 days (actually, consecutive days, July 24-31).

The value of the strong approval rating (S)PAI varied between +30 and -12, while the value of the total approval rating (T)PAI varied between +35 and -3 and was always (or at least nearly always) greater than (S)PAI.

So if you look at Obama's (S)PAI, you'll see that his rating has been negative for 33 consecutive days and counting, and reached into negative double digits, which the graph with strong positives and negatives and the PAI trend conveniently illustrates.

However, if you bother to look at Obama's total approval and disapproval and calculate (T)PAI on your own and graph the negative red line on your own (since only one line for total positives is provided on the corresponding graph), you'll see that his rating had been barely negative, and only for the past week out of his whole term so far, and that he has rebounded this week and is in the green again.

So why is Rasmussen doing that? He is, of course, known to be conservative, but he has shown some restraint until recently in bending statistics to his political beliefs. Why throw away that certain amount of reputation he accumulated for relative objectivity, and for what? For that little bit of extra premium self-delusion?