There have been quite a few posts referring to the Survey USA (SUSA) PA poll showing Hillary with a massive 18% lead over Obama. In light of the others polls that have come out this week which show Hillary with only a single-digit lead (and in one case tied), this SUSA seems a bit off, to say the least. However, I suggest we take the long view. These number, like everything else, only make sense in context.

CLICK FOR CHART

I have been updating this graph for a few weeks now. Hillary's points are plotted in orange, and Obama's in blue. There are two fit lines. The polynomial regression line (with 6 orders) plots the best fit (see the r-squared values for goodness of fit), while the darker line plots the mean of the last 5 data points. These are just two different ways to bring order to the chaos. Refer to the one you like best.

What we see is that over the past 6 months, Hillary has maintained her support, while Obama has picked up support as other candidates have dropped out and as PA has gotten to know him. What is most amazing, to me, is the deficit that Obama had to overcome. In 6 months, he has moved from 14% to about 42%.

Now, to point. The SUSA poll doesn't look like an outlier when seen in perspective of time. Each poll has a margin of error, and I suspect that the SUSA poll just happened to put Hillary at the top of the error range and Obama towards the bottom. I can't know that for sure, obviously, but the values don't particularly stand out (FYI, the point mostly in question is the one that appears right under Clinton's name on the chart).

So, I don't think the SUSA poll is an outlier. I also don't think that it is any more predictive than the rest. It is just part of the distribution of polls. When all are brought together, the picture is much clearer. Context is key.