Two things:

All major pollsters have adjusted their sample weighting to avoid the under-representation of Trump voters. Some commentators believe they may have over corrected.

Other observers believe that if there is a 'shy voter' it may be women in male dominated households were voting a different way than their Trump supporting husbands.

At the moment no one knows the extent of polster accuracy and the result will depend on turnout. Do some of the 'low information' male voters who lean towards Trump turnout? On the other side of the ledger, how many of the female voters who dislike Trump do similalry? This is the so-called potential 'enthusiasm gap' which could make a difference. At the moment, we just don't know, but I doubt if 'shy Trumpers' are the major issue.