Laboratory faking research often instructs participants to respond honestly (generic instructions [GI], control condition) or to fake (a personnel-selection scenario [PSS], faking condition). Research has shown that instructions can impact the frame of reference (FOR), which subsequently impacts responses. Thus, not only might using a PSS to investigate faking motivate respondents to fake, but it might also influence their FOR. Consequently, differences between faking and control conditions may partly result from different FORs and thus confound faking effects. We combined instruction manipulation with item-level contextualization to disentangle FOR and faking effects, conducted a 4-wave longitudinal study with a 2 (GI vs. PSS) x 2 (full item-level contextualization absent vs. present) repeated-measures design (N = 309), and compared the impact of these conditions on three HEXACO-PI-R scales (Conscientiousness, Emotionality, Honesty-Humility). Irrespective of the personality trait under investigation, the ANOVAs revealed significant main effects. As expected, the PSS induced a work FOR, although the effects were smaller than those caused by item-level contextualization. Also, as expected, the PSS and the item-level contextualization changed respondents’ scale mean scores. However, there were no interactions. The study provides evidence that the internal validity of faking research is not threatened by the adoption of different FORs.