In particular, the prevalence of the traditional employment interview as the primary selection method will be compared to the alternative methods of cognitive ability tests and biographical data.
Whilst the assessment and selection process provides information for decisions by both the employer and the potential employee, this is not the traditional view as employment decisions have long been regarded as a management prerogative (Torrington & Hall, 1991). However, given the predicted skill shortages and the fact that selection is also concerned with the future life plans of individuals, the predictive validity of selection methods is an important issue (Meijer, 1998) both for organizations and for individuals. Predictive validity refers to the extent to which an assessment measure can predict subsequent job performance (Smith et al, 1993) such as error rate, production rate, appraisal scores, absence rate, or other criteria that may be important to the organisation. Relationships between assessment outcomes and future performance are expressed as correlation coefficients (r), where r = 1 represents a perfect relationship, and r = 0 signifies that no relationship whatsoever exists. A correlation of, say, r = 0.4 is regarded as comparatively good in assessment and selection (Torrington et al, 1991), but this does illustrate that there are no methods of selection that represent outstanding predictors of future performance.
In their survey of management selection methods used in French and British organizations, Shackleton & Newell (1991) illustrated that, although there was an increasing use of personality assessments, cognitive ability tests, assessment centers and biographical data between the years of 1984 and 1989, the traditional method of face to face interviews continued to represent the dominant method of assessment and selection in the UK. The format of interviews may range from totally unstructured where no objectives are set, to highly structured, pre-planned processes in which topic areas are closely related to job behavior, with varying degrees of semi-structure in-between (Anderson & Shackleton, 1993). Huffcutt & Arthur (1994) used meta-analysis to demonstrate that predictive validity increased as interview structure increased; the degree of standardization of questions and response scoring mechanisms resulted in validities ranging from r = 0.20 for less structured interviews to r = 0.56 for more highly structured interviews. Nevertheless, even highly structured interviews may vary in their predictive validity. In a comparison of situational interviews (future-orientated) and behavior description interviews (past-orientated) Campion, Campion & Hudson (1994) found higher validities for past-orientated (r = 0.51) than for