Text this: Could differential underreporting of loneliness between men and women bias the gender-specific association between loneliness duration and rate of memory decline? A probabilistic bias analysis of effect modification.