The Case Against Assessment Centres for Recruitment

Almost every major early careers employer in the UK uses assessment centres within their selection processes; they are surprisingly ubiquitous. The logic behind assessment centres makes intuitive sense: finding the best candidates at almost any cost. A whole day of exercises, interviews, tests, and assessments—sometimes multiple days. Surely, that is the best possible way to select candidates?

A quick thought exercise for the early careers professionals reading this:

1. How many peer-reviewed research articles have you read regarding the effectiveness of assessment centres?
2. How many assessment centres have you run?

In this article, I will outline the surprising downsides of assessment centres for recruitment and selection, along with some recommendations to get the most out of assessment centres moving forward.

Downside 1: Surprisingly Low Predictive Validity

By all accounts, assessment centres should be the strongest predictors of job performance known. After all, they actually include a number of assessments that are themselves powerful predictors of performance, including interviews and psychometrics. However, for whatever reason, assessment centres seem to be less than the sum of their parts and are only modestly associated with job performance. For example, based on the oft-cited Schmidt and Hunter review of employee selection methods (1998), structured interviews correlate at around .51 with job performance, whereas assessment centres correlate at only .37.

To put this in context, the personality trait of conscientiousness showed an association of .31 with job performance and can be reliably measured in a couple of minutes remotely online. Assessment centres, however, are a whole-day affair and require a huge amount of time, energy, effort, and expense. Consequently, if we evaluate the utility of selection tools based on their effectiveness at predicting performance relative to their costs, assessment centres rank among the least effective selection tools known.

The reasons for their modest validity are open to debate, but I suspect that the effective exercises (interviews, psychometrics, etc.) have their impact diluted by largely ineffective exercises (group exercises, case study exercises, etc.). I have yet to see any compelling evidence that these assessment centre exercises are useful predictors of performance and instead amount to little more than a popularity contest. Consequently, the overall assessment centre validity ends up being fairly middling, dragging down their predictive validity.

Downside 2: Too Little, Too Late

The other problem is that assessment centres are used at the end of the recruitment process, resulting in a "restriction of range" (Hermelin et al., 2007). If the organisation has thoroughly screened applicants beforehand, the quality of applicants should be fairly similar. As a result, there isn’t much to distinguish between them, making assessment centres fairly redundant. Indeed, best practice often dictates that you should be in a position to make an offer to 60%–80% of the applicants who attend, suggesting that most of the hard work should already be done by the assessment centre stage.

Even more troubling is when organisations have done a poor job at screening beforehand. Imagine you have 1,000 applicants and plan on inviting 20 to the assessment centre, aiming to hire the top 10. How you shortlist from 1,000 to 20 is MASSIVELY more important than how you shortlist from 20 to 10. That initial pool of 1,000 applicants, statistically speaking, will have well over 100 high-potential applicants, and your number one priority should be identifying them. However, if your entire recruitment budget is spent on assessment centres, you will just have 20 average candidates show up, with maybe one or two diamonds in the rough. Instead, organisations must recognise that the important work is done at the early stage, and the late stages are purely for due diligence.

Realistically, recruitment budgets only go so far, and the costs associated with assessment centres are particularly steep. Organisations should give serious thought to where best to spend their money. The costs of running an assessment centre could easily be put towards psychometric assessments during the shortlisting process, dramatically enhancing validity and scalability compared to old-school CV sifting. But instead, many organisations overinvest in the final stage of the recruitment process, by which point the strongest candidates have long been screened out.

Downside 3: Difficulties with Wash-up Sessions

One contributing factor to the disappointing effectiveness of assessment centres is the use of "wash-up sessions" when making decisions. Organisations will systematically evaluate candidates on multiple criteria and exercises, only to have an informal conversation at the end and decide who gets selected subjectively. This naturally undermines the effectiveness of the process and the assessments themselves, reducing the predictive validity of the assessment centre (Feltham, 1988).

Wash-up sessions also allow for the introduction of personal biases and irrelevant opinions (Dewberry, 2006). Implicit bias can be particularly insidious, working its way into selection processes without assessors noticing. For example, candidates from higher socio-economic backgrounds could come off as more “professional” in the eyes of assessors, giving them an advantage regardless of how they performed in the exercises themselves. This implicit bias could certainly impact candidates from other demographics, hampering diversity and inclusion objectives.

Instead, an "actuarial" approach is preferred and tends to show higher levels of predictive validity. This simply involves averaging the scores on the various assessments and exercises, then ranking the candidates by that score. The highest-scoring candidates then receive offers, regardless of the personal preferences of assessors. This reduces the scope for bias at the decision-making stage, ensuring decisions are made solely on the data collected.

How I Would Recommend Designing Assessment Centres

I am under no illusions that organisations will suddenly stop running assessment centres. The evidence that I’ve referenced isn’t exactly breaking news; the research is even older than I am. Moreover, I suspect that organisations are at least vaguely aware of the weaknesses of assessment centres but do them anyway for two reasons:

1. Their competitors use assessment centres, and they don’t want to be seen as less invested in their candidates than other employers.

1. Assessment centres serve as open days for the organisation, giving them the chance to impress candidates and showcase their corporate culture.

Both of these are good reasons to use assessment centres, and I certainly don’t fault organisations for wanting to give this impression. However, I believe we can improve upon the traditional assessment centre by acknowledging that half the benefit is about candidate experience and then acting accordingly. Consequently, here are some changes to assessment centres that I would recommend:

• Only use structured interviews. Drop group exercises, case study exercises, analysis exercises, psychometrics, etc. These are either just fillers that dilute the effectiveness of assessment centres or tools that should be used earlier in the recruitment process.

• Turn a sizable portion of the assessment centre into an open day. Invite speakers from the organisation, introduce candidates to different department heads, give out promotional materials, etc. Really focus on the employer branding piece.

• Avoid using wash-up sessions. Instead, use numerical scoring only and select the highest-scoring applicants for offers. This eliminates the scope for biases at the decision stage while also boosting the predictive validity of the assessment centre itself.

By following these steps, you retain the benefits of assessment centres without the reduction in selection process validity. This does, however, require organisations to acknowledge the fact that their assessment centres have historically been an inefficient use of resources, which can be a bitter pill to swallow. Nevertheless, I strongly recommend that organisations take time to honestly reflect on their selection processes and make much-needed adjustments when required.

References

Dewberry, C., & Jordan, D. (2006). Do consensus meetings undermine the validity of assessment centres?.

Feltham, R. (1988). Assessment centre decision making: Judgemental vs. mechanical. Journal of Occupational Psychology, 61, 237?241.

Hermelin, E., Lievens, F., & Robertson, I. T. (2007). The validity of assessment centres for the prediction of supervisory performance ratings: A meta‐analysis. International Journal of Selection and Assessment, 15(4), 405-411.

Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology: Practical and theoretical implications of 85 years of research findings. Psychological bulletin, 124(2), 262.