As far as I understand, the traditional sense of "shy voter" is to describe your poll numbers (i.e. the people who did take the survey) which answered "undecided" when they were shy about their choice. This also goes hand in hand with things like observed differences in answers of in-person surveys when the person doing the survey is/isn't from the same demographic as the person being surveyed.
A disproportional amount of people (after controlling for race/gender/income/education/etc) simply refusing to answer the survey has similar but conceptually different effect; one that (as the OP claims - perhaps they're mistaken) is new and had not been a significant factor before.
A disproportional amount of people (after controlling for race/gender/income/education/etc) simply refusing to answer the survey has similar but conceptually different effect; one that (as the OP claims - perhaps they're mistaken) is new and had not been a significant factor before.