1 Psychometry and GPA
Intelligence is a good predictor of job performance, and accounts for 60% of variance, even when controlling for various measurement errors. The argument is that intelligence greatly contributes to the acquisition of job knowledge, which in turn results in job performance. Granted, you can find examples of pathologically shy Mensa members, and that is partly why we also use additional tests, like e.g. personality tests. Personality tests offer incremental validity over intelligence, and tap into very different psychometric properties that is of great relevance in selection. Conscientiousness is of particular importance, and is a central part of integrity tests, yet there is a curvilinear relationship such that an ideal value is better than the highest.
It is very common for employers to put allot of weight on GPA, which counts for 30% of variance, but the theoretical rationale for why GPA would predict job performance is not clear. One possible explanation for why GPA is not a better predictor is that the amount of relevant job knowledge students acquire varies between colleges, degrees, and courses; e.g., a business degree arguably result in more generally relevant knowledge than one in medieval history. Another explanation for why intelligence is a better predictor of performance is that, while GPA and intelligence tap into similar constructs, GPA achievements might be moderated by conscientiousness compensating for intelligence.
2 Work Sample and Reference Checks
A big benefit of these is that they exhibit a high degree of realism that is not found in the other selection methods. Work samples tests tap into amount of job knowledge, with higher amount clearly leading to better job performance, and it adds incremental value above intelligence.
3 The Selection Interview
Interviewing is much more subject to extraneous variables than the other selection methods. Structured interviews have higher validity than their unstructured counterparts, accounting for 51% variance, though structured interviews tap poorly into personality factors. Still, when it comes to structured interviews, same level of criterion-related validity where found with nontraditional employment interviews. Unstructured interviews are very difficult to execute successfully.
Interviews gather more thick qualitative data than other instruments, as seen in a recent taxonomy of constructs the interview measure; mental capability, knowledge and skills, personality, social skills, interests and preferences, organizational fit, and physical attributes. Interviews can be categorized as conventional or behavioral, with the former tapping more into intelligence, personality, work- and social skills, while the latter is better at assessing job-specific and social skills. It is common to have multiple interview sessions before a hire, so it is fully possible to triangulate various kinds of interview types in order to compensate for the deficiencies in each.
Video background: Fallout 4, by Bethesda Game Studios