In most developed democracies the percentage of people who vote in a general election is between 50 and 90 percent. Has anyone tried to measure how different election results would be if only the people who don't normally vote voted? Would we see a uniform swing to the left for example?
This website gives differences based on different levels of electoral participation in a variety of demographic measures. As one might expect, non-voters are more likely to be ethnic minorities, poor, lower education, younger, less politically engaged, more cynical, less socially engaged, and more liberal. Most pertinent to your question:
Roughly the same proportions of self-identified Republicans and Democrats are regular voters (41% vs. 39%). But Democrats are more likely to be non-voters: 20% of Democrats say they are not registered to vote, compared with 14% of Republicans; among political independents, 27% say they are not registered to vote.
A registration gap also exists between liberals and conservatives, with 29% of self-described liberals saying they are not registered to vote compared with 20% of moderates and 17% of conservatives. However, there are only modest differences in the percentages of conservatives (38%), moderates (35%) and liberals (34%) who are regular voters.
Doing some math, that means that people aren't registered to vote but identify as Republican make up 5.74% of the population, Democratic, 7.8%, independents, 5.4%. Thus, of the people not registered to vote, 30.3% identify as Republican, 41.2% as Democratic, and 28.5% as independent.