As you're probably already aware (being a TPM reader and all), the recent PPP survey out of Pennsylvania shows Senator Obama with a 3 percent lead over Senator Clinton.

While encouraging, the difference is statistically insignificant. The  real usefulness of the poll lies in the insights it provides into the demographics of the crucial 13% of PA voters who remain undecided.

Shuffling through the crosstab numbers, and in absence of additional direct insight into PPP's weights, we can draw some conclusions how the undecided might break down. Sparing the gory details we find:

By Sex:
Women (net differential): + 0.77% Clinton
Men (net differential):  + 1.22% Obama

Or alternatively by racial affiliation:
White: +1.38% Clinton
AA: +1.36% Obama
Other: +0.21% Obama

Did not immediately find enough data to run a similar comparison on the age break down.

Now if you accept PPP's weighting schema, completely ignore potential sampling error, and accept the axiom that undecided voters from the groups above break down identically to those who claim to have already decided, I'd say things look pretty sunny for Senator Obama's PA outlook.

These are all <I>really</I> big "ifs", but still the ingredients are there for a potential primary watershed.

Disclaimer: I often use an analytical model to predict results completely independent of polling data, and it computes a 8.9% victory for Senator Clinton. However, the model has some inherent faults - notably in capturing dimensions representing just the types of factors that have recently come into play in PA.