Compared to this point in the 1994 cycle, the Democratic Party is in relatively decent shape for a party that is on the verge of losing their majority. There are between 70 and 100 plausible House seats in play, and depending upon one's inclination, Democrats remain competitive in most of them. The sky has not fallen. Early voting is proceeding apace. Democratic targeting efforts, which are persistent and based on science, are kicking in.
Republicans have more money, more outside help, a more resonant message and have yet to make the type of mistake -- this is a nationalized election, after all -- that would pull the rug out from underneath the party's feet at the last moment.
But if you wonder why prognosticators have begun to forecast a bigger Republican victory, it's because they're reading below the top lines of polls, below, even, general questions about voter enthusiasm. (It's also because Democrats are now spending money to shore up seats like Rep. Costa's in the San Joaquin Valley in CA -- the 20th.)
Voters who are the most enthusiastic and likeliest to vote are going for the Republicans. And -- infrequent voters are more susceptible to Republican messaging, pollsters and strategists in both parties tell me.
At this stage in the election, parties are measuring likely voters on a propensity scale. It's usually broken up into thirds, with voters being designated as low propensity, medium propensity, and high propensity. Microtargeters use a variety of metrics (past voting history, age, location, response to messaging, commitment to call-takers, race, etc.) to assign each voter a propensity score, much like Democrats break their Iowa caucus supporters into "ones," "twos," and "threes." Political scientists are coming to see these propensity metrics as fairly predictive
, although a person's propensity changes over time. (The older you get, the more likely you are to vote, for example.)
SpatialLogic, a microtargeting firm in Virginia, has this definition of the measurement:
A voter propensity index measures multiple observances that have voter commonality (i.e., election results, survey responses, demographic characteristics) and calculates a value to indicate the strength or propensity of a voting group. All of the variables collectively measured in the index must share the same base geography (ex. voter precinct, ZIP code, block group). Some indices are calculated by taking the average of the variables examined, others are simply a total of the variables measured. More sophisticated voter indices may include variables that have been multiplied by a weighted measure to give importance of one variable over another. How one derives a propensity index should be based upon the target group and the data available.
The secret of voter targeting is that people who get high voter propensity scores are USUALLY quite likely to vote already, and therefore targeters don't have to spend too much time on them. These are party regulars. For Republicans, there are no problems with this group of people this cycle. Democrats have had trouble, until recently, in ensuring that their "mid-term base" turns out. That is, Democratic voters who SHOULD turn out -- who have high propensity scores -- are being more ornery than usual. They might squeeze through a likely voter sample in a poll, or they might slip outside of one.
But where Republicans have a massive advantage this year is in turning out what political scientist David Niven termed "infrequent" or occasional voters. These voters tend to be most susceptible to persuasion, in that microtargeted appeals can move them to do something they otherwise wouldn't. Democrats aren't moving their universe of infrequent midterm voters. Republicans are.