Problems arise, however, where initial sampling paths can be seen to affect subsequent referral paths, thus skewing the “recapture” process to those in the original sample (and resulting in an inaccurate recapture number, see Berchenko & Frost, 2011).