A significant fraction of the people voting for Clinton in the Pennsylvania Democratic Primary have no plans to vote for her in the fall.
According to the CNN exit poll, in an Obama:McCain match up 15% of those voting in the Democratic primary would vote for McCain in November. Of these voters, 90% reported voting for Hillary Clinton. In a Clinton:McCain match up, 10% of the Pennsylvania Democratic voters indicate that they would vote for McCain. Of these, 69% voted for Obama in the primary and 31% voted for Clinton.
Clearly, Obama voters were reflecting greater Democratic Party loyalty than the Clinton voters. Were these disaffected Regan Democrats? Were racism and misogyny or did these voters never intend to vote for either candidate and were simply creating mischief.
If these votes are discounted from the total, the margin of Clinton's victory was 6.2 rather than 9.2%.
On Page 1 at the bottom of the poll, you will see the following:
Vote for President in November {Clinton vs McCain}
Response Voted for
(% response) Clinton Obama
Obama (82%) 62% 38%
McCain (10%) 31% 69%
Would not vote (6%) 5% 95%
On page 4 in the second question, you will see the following:
Vote for President in November {Obama vs McCain}
Response Voted for
(% response) Clinton Obama
Obama (73%) 41% 59%
McCain (15%) 90% 10%
Would not vote (9%) 98% 2%
The pollster has already determined whether the voter voted in the Democratic or Republican primary. The question is, in a match up between Clinton and McCain (or Obama and McCain) "Who do you plan (or expect) to vote for in November?"
Of the respondents, in the Clinton:McCain match up, a total of 16% either would vote for McCain or not vote; In the case of the Obama:McCain match up, a total of 24% either would vote for McCain or not vote.
One reason Republican voters might chose to participate in the the Democratic party voting process is that they are highly dissatisfied with the Republican candidate. On the Republican side, running basically unopposed, McCain was able to garner only 73% of the vote with 16% going to Ron Paul and 11% sticking with Mike Huckabee. Not a very solid endorsement.
In looking at the first table, 6% of the respondents report that they "would not vote" in November, yet 5% of those reported voting for Clinton. Now this is a very small number of voters (5% of Clinton voters or 62,295 votes), but why would anyone turn out to vote for Clinton in the primary if they did not plan to vote for her in the fall, unless they were simply trying to distort the outcome?
In looking at the second table, in the Obama:McCain match up, even though larger percentages of voters indicated that they "would not vote" (9%), only 2% of the latter (20,759 votes) reported that they voted for Obama in the primary. Again, these could be considered "mischief voters" - in this case distorting the vote toward Obama.
Numbers that are a bit more difficult to interpret are those reporting that they would "vote for McCain" in the fall. In the Clinton vs McCain match up, 10% of the voters responded that they would vote for McCain in November even though 31% voted for Clinton in the primary. This represents about 3% of the total vote in the Democratic primary, or about 70,799 votes.
In the Obama vs. McCain match up, 15% of the voters responded that they would vote for McCain in the fall. Of these, 10% reported voting for Obama. These represent about 1.5% of the total or 34,257 votes in the Democratic primary.
If the relevant totals are subtracted from each of the tallies, one sees the following:
Votes for Clinton 1,245,911 - (62,295 +70,799) = 1,112,817 net (53.1%)
Votes for Obama 1,037,953 - (20,759 + 34,257) = 982,937 net (46.9%)
The difference between these two percentages is 6.2%
If it had been reported that Clinton beat Obama by 6.2% rather than by 9.2%, the perception would have been considerably different.