While I'm sure this district probably ranks toward the bottom of the country for 18-29 population, there have been a number of Senate polls this year with extremely low participation rates for this age group out of line with reality. 4% is not realistic and says more about polling than demographics. That said, I doubt it makes much of a difference.
The Census claims 12.7% for 20-29 with an additional implied 3.5% for 18-19, for a total of 16.2%. Voting participation in a special election will be lower, even disproportionately lower than it is in a general election, but not that much.
The 2009 New Jersey and Virginia exit polls showed 18-29s comprised 9% and 10% of the electorate, respectively. Based on my imputation of the 2008 census estimates, about 18.5% of New Jersey voting-age residents and 21% of Virginia voting-age residents were 18-29s. That means about half of all 18-29s showed up to vote in the big 2009 elections, relative to their percentage of the voting-age population.
If your numbers are correct and about 50% of PA-12 18-29s show up to vote, they should be about 8% of the electorate.
But 18-29s might not turn out in the same proportions in an irregularly scheduled special election as they would for a regularly scheduled general election. FWIW,
this website claims only 15% of Massachusetts 18-29s showed up to vote in the Massachusetts Special Election based on some bizarre imputation of Rasmussen's last poll (remember - there was no exit poll). According to that website, 18-29 MA turnout was lower than NJ (19%) or VA (17%).
So the 18-29s in PA-12 could very well end up being lower than 8% of the total turnout in the special election. Probably not 4%, but 4% is not so far off as to skew the results much.
