That average seems a little high. The problem with straight averages is that just a couple of elections with really large gains can really bump up the number even if most elections are relatively sedate in terms of turnover.

Second, for the past 70 or so years the 6th year midterm has been beset with a lot of other factors. But key here is that in 1998 and 1986 (the last two 6th year midterms), the net change for the President's party was 5 seats, and in 1998 the Democrats gained 5.
The others:
1938 - The GOP only held 88 seats in the House going in, so they really had nowhere to go but up big. Plus FDR had just miscalculated with Court packing.
1950(?)- Some might not count this one - but it was a 28 seat shift.
1958 - 48 seat loss by GOP - economy was in a recession.
1966 - Vietnam
1974 - Watergate

While the lefties might be running around bemoaning the state of affairs, I don't think the average American sees the situation today as being like any of these prior elections. Iraq may be bad, but it is not another Vietnam, there is no recession (quite the opposite), most people support things like wiretapping and such (as opposed to Watergate and Court Packing). And finally, there just aren't as many competitive districts as there once were. The "science" of gerrymandering has improved - modern computer databases have made it possible to carbe out districts down to the street level knowing exactly how many registered Republians, Democrats and Independents live on any block.

On the Senate side the number appears to be high due to the single anomaly election of 1958, where the Republicans lost about 15 seats. Second was 1986 with an 8 seat loss (most of the losses being seats held by freshman elected with Reagan in 1980). Others back to FDR seem to be much more sedate and appear to max out at 5 seats.

You are correct - the "sea-change" elections do skew the averages.

I suppose the point I'm dancing around is "How do we measure the results?"

Even adjusting averages downward to discount those landslide years, we still come up with Democrats who ought to win back the House and forge a tie in the Senate. They almost have to do that much to claim a "victory."

The Senate's expectations are different, and have been all year, and it is almost impossible for Democrats to win control there anyway, so suppose we just rule that one out. They still MUST win the House, or they have LOST the midterms.

It's been almost impossible to pin any Democrat down on a concrete prediction, though. They won't even low-ball their expectations to put a number on it.

Jim's point is made. I spent some time taunting the left whenever I would read comments like "the WHOLE COUNTRY is fed-up with Bush and the GOP" - then it should be an absolute cakewalk to win the measly 15 House seats necessary to take back the House.

This is their talking point. THE WHOLE COUNTRY IS UPSET over (Iraq, economy, theocracy etc).

If they do not win the House back - even if they win 8 or 9 seats - they have a MAJOR loss on their hands.

Speaking of 1998, the GOP kept the same balance in the Senate that year, and it was seen as a crushing loss, because they were expected to make strong gains (and even get a filibuster proof majority).