Psychometric Testing vs. Interviews: What the Science Says

Psychometric Testing vs. Interviews: What the Science Says

Interviews are, on a per-hire basis, the most commonly used employee selection tool. Almost every role will involve at least one interview as part of the process, and most hiring managers wouldn’t dream of hiring a candidate without interviewing them first. Psychometrics, on the other hand, are used less frequently on a per-hire basis but are used in substantially higher volumes than interviews, especially for early careers. For every candidate interviewed, dozens or even hundreds of candidates will need to complete psychometric assessments for short-listing, making psychometric tests far more frequently used in absolute terms.

However, what does the evidence say about these two assessment methodologies, and which is the more effective tool for hiring?

Validity of Interviews

The research in industrial and organizational psychology suggests that interviews are a very mixed bag when it comes to predicting future performance. When done well, they rank among the most powerful predictors of performance known, but when done poorly, they exhibit a fraction of their potential predictive validity. Moreover, given the subjective nature of interviewing, they can result in significant adverse impact against certain protected groups, depending on the interview structure and format.

Of all the variables that influence validity, interview structure has the greatest impact. Structured interviews, which have formal scoring criteria and a fixed set of questions, show the greatest validity. Unstructured or conversational interviews show substantially less validity, likely as a result of lower reliability. Naturally, if candidates are each given different questions, some will inevitably receive a harder time than others, making it harder to compare. Although conversational interviews may seem more natural, this loss of consistency hits their effectiveness hard, making them inferior selection tools.

Lastly, the format of the interview also seems to make a difference, albeit less so than interview structure. For example, panel interviews surprisingly underperform relative to one-on-one interviews. The reason for this isn’t well understood, but it is suspected that panel interviews make candidates more nervous, interfering with their ability to perform. Additionally, asking candidates future-oriented or hypothetical scenario questions seems to outperform past-oriented questions, i.e., “Tell me about a time when…”. This is because candidates generally lie about their past, confounding the result. Hypothetical situations, however, are a truer test of knowledge and skill, making them more effective.

Validity of Psychometric Tests

The validity of psychometric tests tends to vary even more significantly than with interviews, and for many reasons. Some psychometric tests are designed to measure constructs that are highly relevant to job performance, and others measure traits that are entirely irrelevant, making them useless for hiring. Additionally, individual tests have varying levels of reliability, making some assessments more useful than others for hiring purposes. Lastly, some psychometric constructs are ubiquitously relevant to the workplace, whereas others are highly role-specific and offer little value elsewhere. Consequently, some psychometric tests are extremely effective hiring tools, whereas others could offer almost no value or even potentially negative value to the recruitment process.

Based on the research, cognitive ability tests show the highest levels of predictive validity among psychometric tests. Cognitive ability underpins one's ability to learn, solve problems, and make decisions, and is a powerful predictor of performance in cognitively complex professional, technical, and managerial work. In these contexts, well-designed cognitive assessments outperform structured interviews, making them more powerful hiring tools. However, in more simple and routine work, interviews tend to outperform cognitive assessments, and thus the context matters significantly.

Personality questionnaires are also highly effective tools, but also particularly situational. Some traits tend to be ubiquitously predictive of performance, such as resilience and conscientiousness, but others are highly role-dependent. For example, extraversion—how social, gregarious, and assertive a person is—tends to be useful for hiring salespeople but irrelevant for hiring web developers. Overall, personality questionnaires tend to be somewhat weaker predictors of performance than interviews, but they have the added benefit of predicting employee retention as well as performance, making it debatable which is better for hiring.

Conclusion and Recommendations

When looking at psychometric tests in combination, they are substantially more powerful than interviews. This is because different psychometric tests complement each other and have the capacity to cover a wide variety of psychological constructs that are relevant to performance. For example, using cognitive ability tests, numerical reasoning test, a personality questionnaire, a situational judgment test, and a range of knowledge tests in combination will almost certainly outperform any interview by a significant margin. However, when comparing a single test to a single interview, the context matters considerably, with either the psychometric assessment or the interview coming out on top depending on the situation.