By Paul Barrett on August 22, 2016
If we were to believe the calculations of assessment utility from the major test publishers such as CEB, based upon the conventional utility formulae we see in the textbooks, then using any kind of psychometric assessment will result in dramatic financial benefit.
But, these simple calculations are unrealistic, as the economist Michael Sturman demonstrated empirically in an article published in 2000. As he factored in several real-world conditions which operate in any job market, utility estimates were incrementally reduced by a total of 96% from their initial simple calculation. That’s the reason why the consequences from the use of candidate psychological/ psychometric assessment is not always noticeable, in contrast to what many test-publisher salespeople claim. As Sturman notes in the abstract to his article:
“These results imply that human resources programs do not invariably yield positive returns; rather, intervention success is contingent on program-specific, organizational, and environmental factors.”
However, to ask for a Return on Investment (ROI) indication is a sensible request associated with any intervention deployed into the working environment, whether software, robotics, plant machinery, office equipment, through to psychological assessment of candidate employees. And that return may be measured in various ways, which may or may not be expressed in terms of straightforward ‘outcome financials’.
Take for example the use of a personality assessment as an employment pre-screen for administrative assistants in a government social services department. What is the “Return” on any financial outlay for psychological assessment? If performance assessment evaluates ‘meeting KPI and role-expectations from a manager’, then the outcomes may or may not have any direct financial consequences. But if the Return is seen as reducing employee turnover, then that outcome conveys a more direct financial consequence; a reduction in costs of having to replace employees. In this particular instance, you need to evaluate qualitative, indirect, as well as direct quantitative returns.
A Typical ROI work-up
Consider the sales turnover of a commercial organisation (say $40m pa), the percentage (20%) of pre-tax profit ($8m), the average salary of employees at operational (n=160 @ $50k pa) and executive levels (n=10 @ $200k pa) in the organisation is $8m and $2m respectively, and the percentage of staff annual turnover in both these categories is 15% for operational and 10% for Executives. With the costs incurred in the exiting and subsequent hiring of operatives put at ~$20k each (about $480k), and about $300k for each executive, these turnover costs can amount to almost $800k a year.
If by using some form of formal psychological assessment, you could reduce executive and operational staff turnover by 50%, the potential savings each year might amount to nearly $400k. The cost of that psychological assessment is likely to be around $100-200 pp, which even if allowing for the testing of 500 candidates as a pre-screen, would amount to somewhere between $50-100k. So, an overall reduction in costs of about $300k might be expected to be achieved. However, we have not factored in the potential increase in sales turnover caused by a more stable and more ‘selected-for-suitability’ group of employees. So the $300k cost reduction might also be augmented by a 1% increase in sales turnover ($400k), with all other costs remaining static.
We can calculate a potential ROI as:
((cost-reduction + increased sales) – cost of assessment) / cost of assessment
which even with cost of assessment at $100k, shows a positive ROI of 6x the investment, or 600%.
This is all “if”, “might”, “should”, and “maybe”. Furthermore, using validity coefficients from one-shot or even meta-analytic studies in the literature is rarely helpful, especially when those validities are small (as with the impact of employee engagement on organizational profitability). What many forget is that small effect sizes rarely replicate from sample to sample, as Gregory Francis points out, and as Paul Barrett demonstrated with the ROI calculations around Gallup’s Q12 engagement assessment and organizational roll-out in a whitepaper, making use of the Gallup Q12 evidence-base.
From our perspective, best-practice ROI calculations are organization-specific. For clients who require such estimates in order to perhaps justify the proposed assessment expenditure to the CFO or CEO, we interactively model all the factors relevant to intervention deployment for justifying high-stakes/high cost assessments, both qualitative and quantitative. It’s complex and sometimes not worth the effort given the often small cost of intervention/assessment relative to the annual financials. But, when deploying assessments across a large corporate, those costs can be substantial relative to any potential return.
Applying simple utility or turnover formulas as so many do, or even using the kind of workup we demonstrated above, is clearly sub-optimal when you need to develop a more accurate evaluation of the likely real-world ranges of potential ROI for your workplace intervention deployment. We will be presenting an example of a realistic ROI workup for deploying Cognadev Assessments in a forthcoming Cognadev Technical Report.
 Sturman, M.C. (2000). Implications of Utility Analysis adjustments for estimates of human resource intervention value. Journal of Management, 26, 2, 281-299.
 Francis, G. (2013). Replication, statistical consistency, and publication bias. Journal of Mathematical Psychology, 57, 5, 153-169.