In laboratory faking research, participants are often instructed to respond honestly (generic instructions [GIs], control condition) or to fake (personnel-selection scenario [PSS], faking condition). Considering the research on instruction-level contextualization, a PSS might not only motivate participants to fake but might also promote the adoption of a work frame of reference (FOR). Thus, differences in responses between faking and control conditions could partly result from FOR effects. (Full) item-level contextualization can also be used to promote the adoption of a work FOR, and the adoption through this route is stronger than through instruction manipulation. We combined the two approaches to disentangle FOR and faking, conducted a 4-wave longitudinal study with a 2 (instructions: GIs vs. PSS) × 2 (full item-level work contextualization absent vs. present) repeated-measures design (N = 309), and compared the effects of these conditions on three HEXACO-PI-R scales (Conscientiousness, Emotionality, Honesty-Humility). Irrespective of the investigated personality trait, the ANOVAs revealed significant main effects. As expected, compared with GIs, the PSS increased the adoption of a work FOR, and the effects were smaller than the effects of full item-level work contextualization present (vs. absent). Also, as expected, the PSS (vs. GIs) and full item-level work contextualization present (vs. absent) changed participants' scale mean scores. However, importantly, there were no interaction effects. Exploratory mediation analyses indicated direct rather than indirect (mediator: adoption of a work FOR) effects of instructions on participants' scale mean scores. In conclusion, the internal validity of faking research is not threatened by confounding FOR effects.