Should small businesses rely on personality tests for recruiting staff?
Research suggests that 46% of all new hires fail within 18 months, but 89% of them fail for attitudinal reasons (personality traits, human interactions, etc) rather than cognitive ability (brain-based skills like logic and reasoning, problem-solving, language, etc). So as a small business owner and hiring manager, what can you do to ensure you hire the right people?
Can do - will do - will fit
As business owners, we are responsible for hiring the right people and building teams which perform well together. And the smaller the business, the more costly it is when you get this wrong.
A senior executive I worked with many years ago once shared with me a simple yet useful rule for effective recruitment, which describes a stepped approach to candidate selection, focusing first on technical ability, next on attitudinal or motivational disposition, and finally on cultural fit: can do - will do - will fit.
The following assumes you have identified a candidate who can do the job, and focusses instead on the latter two stages of the assessment model.
Personality testing
One tool in the recruitment process that larger companies tend to rely heavily on, but is often overlooked by smaller businesses, is the personality assessment. Personality testing, also known as psychometric testing, is designed to predict how people will behave in the workplace. In other words, it attempts to predict how the candidate will work, rather than worry about whether they have the technical skills to do the job. It might cast light, for example, on how the candidate will work under pressure, how they will interact with co-workers, or whether they will fit into a given team, given the existing team members' personalities. In other words, companies rely on these tests to screen candidates for 'good fit' - the end goal being to reduce turnover and improve productivity.
The argument against using such tests - other than the fact they add time and expense to the recruitment process - is that they don't successfully predict behaviours, or that they are easy to fake.
There are a number of different tests available. Generally, these tests have been developed following a rigorous process relying on academic research and statistical analysis. In other words, academics will test groups of people and identify correlations between certain personality traits and certain workplace behaviours, and then try to capture the essence of those traits through a range of multiple-choice questions. The resulting questionnaires can then be automatically processed, so that (hypothetically at least) no human interaction is required to analyse the data.
One of the most common tests used is the Myers-Briggs Type Indicator (MBTI). Co-author of Nudge and Harvard professor Cass R Sunstein suggests that while 90% of major US companies rely on it, the MBTI is not a good behavioural predictor. In fact, he points to further research suggesting that all personality tests fail in effectively predicting behaviour over time.
In my professional life I've had quite a bit of exposure to personality tests (and in fact I spent a bit of time researching their effectiveness as part of my Masters). I want to share some of my insights here, if only as a cautionary tale.
Low test-retest reliability
Over the past few years, I've taken a number of personality tests, include the MBTI, the Competing Values Framework (CVF), the DISC profile and the Life-Styles Inventory (LSI) assessment.
The DISC profile suggested I was “passionate and expressive and my enthusiasm is contagious” and I “show an ability to persuade others to adopt my vision”, whereas the CVF assessment suggested that this was my lowest rated characteristic out of 100 items.
The LSI assessment suggested I have “an excessive concern with avoiding mistakes” and “a need to look for flaws in everything”, whereas the DISC profile found I can be “overly optimistic at times, dismissing potential obstacles too quickly”.
And finally, the DISC profile described me as “adventurous”, ‘taking risks” and going on “gut instinct”, whereas the LSI suggested I am “very conventional”, with the CVF rating me relatively low on “initiating bold projects” and “starting ambitious programs”.
The reason I share this personal insight is that I believe there are significant flaws with relying on these assessments to predict individual behaviours and performance. While it is possible that there is something unusual and quirky about me and how I take the tests which results in such contrasting results, there is quite a bit of literature out there on the issues with personality tests, and in particular what's known as their low test-retest reliability.
Sunstein suggests that in 50% of cases, retaking the MBTI after a one month gap results in the person being assessed landing in a different personality category. A bit problematic if the person was hired a few weeks ago on the basis of their original category being a good fit to the team they were joining.
One of the concerns I have with these tests is that, in order to offer an automated and one-size-fits-all solution (which is necessary to ensure widespread adoption by unqualified assessors), they often fail to capture the conviction behind the responses given by the candidate. Completion of an assessment will require all questions to be answered, even those where the respondent doesn't really 'get' the question, or is not particularly drawn one way or the other by the available answers. And yet the test does not adequately differentiate between a response which is "yes, that's 100% what I would do in that situation" and "well I don't feel strongly about this question but since I have to pick an answer, here goes." Expecting a piece of software to process multiple data points of varying factual quality, and then spit out a true and consistent assessment of the respondent's personality, is probably asking too much of it!
Developers of these assessments tend to counter that the tests do have controls that detect inaccurate answers, typically by asking a question several times in different ways to test consistency of responses. While this may help correct one misunderstood question, or a slip of the mouse, I'm still not convinced it fully addresses the test's failure to capture the varying degrees of conviction behind a respondent's answers - and so the extent to which certain personality traits dominate, while other traits are only occasionally present and may even be prone to vary.
While personality testing as part of the recruitment process is certainly supported with some academic credibility, businesses should nonetheless apply caution in deploying these tests, in particular if they are being used exclusively to predict a candidate's fit. A lot of the time, the candidate interview and the psychometric test are seen as two separate processes, or sequential 'hurdles' that the candidate must jump over. Instead, my recommendation would be to take a more integrated approach, and address in a second interview (after the test is taken) any concerns which arise out of the test results, probing any potential 'unwanted' personality traits.
Ironically, sometimes the 'will fit' (personality fit) assessment may conflict with the 'can do' (skills) assessment. Research conducted on characteristics of successful salespeople suggests that employees who are assertive and display a strong will to achieve are more likely to sell more than those who gregarious, dependable and persistent. And yet psychometric testing may well filter out the former types on the grounds that they are a 'poor fit' to the existing team. It's therefore necessary to always relate your recruitment assessments to a conscious consideration of the specific skills and attributes you need in a particular role, rather than blindly trust in an automated assessment process.