1. Reliability is related to the precision, trustworthiness or constancy of results obtained with a measuring instrument. Basically, a reliable result is not distorted by random events (e.g., a candidate's momentary indisposition, an ambiguous question variously interpreted by different candidates, or varying behaviours of different interviewers during the course of the interview ¹). Reliability can take several forms.
   
   In the case of a selection interview and numerous psychometric tests, reliability takes the form of stability, which can be verified in two ways:
   a) Stability of evaluations from one interviewer to another—if two or more interviewers evaluate the same candidate, their evaluations should be similar.
   b) Stability of evaluations between interviews: if the same interviewer evaluates two equivalent candidates, the evaluations should be equivalent.
   
   Reliability is often the focus of litigation. Candidates can contest their evaluation, attesting that they were not given the same questions or treated the same way by interviewers, for example, or that identical responses by different candidates led to different evaluations.
2. Validity is the most important quality of a measurement instrument—its ability to measure what it is supposed to measure, or predict what it is supposed to predict. In personnel selection, a valid assessment tool makes it possible to find the best candidates for a given position. In other words, the validity of an instrument is verified when the assessment tools predict candidates' future performance adequately and without bias, i.e., without prejudice based on sex, race or favouritism.
   
   Validity is therefore defined by the relationship that exists between the results obtained by the candidates on the assessment tool and their job performance once they have been hired. This relationship between the measurement instrument and performance is usually quantified using a statistical index, the correlation coefficient (in these circumstances, called the criterion-related validity coefficient), whose absolute value varies between 0.00 and 1.00. The higher the validity, the higher the coefficient.
3. Respect for the law and the organization's policies, as well as legal defensibility. The selection interview used for hiring or promotion purposes must proceed within the limits of a formal framework. If the interview becomes the subject of a dispute, the appellants' allegations must be responded to and evidence that the law or regulation in question was respected must be produced. Although each case is different, the validity and reliability of the interview will become decisive issues. These legal considerations are another important criterion of the effectiveness of the selection interview, because in many Western countries, selection practices are subject to various laws, policy statements or contractual agreements.
4. The candidates' reactions. It is desirable that the candidates have a positive perception of the interview to avoid harmful consequences to the work atmosphere, the reputation of the organization or its recruiting power. In addition, the candidate's perception of the interview will be instrumental in his or her decision to mount a legal challenge or take a case before the appropriate tribunal. The candidate's attitude can be influenced by any number of factors including:
   a) the relevance of the questions,
   b) the professionalism and respect shown by the interviewers, and
   c) the candidate's own perception of the fairness of the process.

Several hundred empirical studies must have been done on the selection interview over the years ². The invaluable data they have produced have made it possible to establish a set of best practices. Among these best practices, two are obvious: structuring the components of the interview and focusing them on the job to be filled.

¹ Without reliability, validity is not possible. Reliability is a necessary but not sufficient condition of validity.

² For an overview of these results, consult the following meta-analyses: Conway, Jako and Goodman (1995); Huffcut and Arthur (1994); Marchese and Muchinsky (1993); McDaniel, Whetzel, Schmidt and Maurer (1994); Weisner and Cronshaw (1988); Wright, Lichtenfels and Pursell (1989); or the most recent narrative reviews: Campion, Palmer and Campion (1992); Harris (1989); Posthuma, Morgeson and Campion (2002).