A theory and model of the ‘socialization impact’ of selection methods is proposed. Developing and extending an earlier model proposed by Anderson and Ostroff (1997), this article proposes an empirically testable theory comprising four fundamental postulates and six corollary hypotheses. The fundamental postulates of socialization impact, varying degree of impact, individual differences between candidates, and sub‐group differences between gender and racial/ethnic origin are put forward as essential cornerstones of this theoretical approach. Socialization impact is articulated across five constituent domains: information provision (IP), preference impact (PI), expectational impact (EI), attitudinal impact (AI), and behavioural impact (BI). Contrasting this theory against the predictivist paradigm which has dominated selection theory, research and practice over the last fifty years, this article sets out an alternative but complementary perspective capable of being tested through subsequent empirical research. The implications of conceiving of selection methods not as ‘neutral predictors’ but as ‘interventive affectors’ are discussed in conclusion.