Cognitive dissonance and HR Analytics is a bad cocktail

HR Analytics (or workforce analytics) promise to give HR executives better data to make better decisions. The potential of analytics is however easy to evangelize and difficult to achieve. I find two psychically challenges with workforce analytics especially interesting; the problem with too much information and the problem with making wrong decisions even with the right data.

The first problem is eloquently described by Jeremy Shapiro, who in his recent post writes about “cognitive load“. Here he points out that too much information will make us make decisions which are either emotional driven or irrational. He use research from neuroscience to back this up. I agree with his point completely when he writes that “know what decision you are asking someone to make. What information is needed to make that decision? Keep that data, and strip out the rest”. In other words; although analytics can provide you will ton of data don’t use it all, but keep it down to the essentials. (Check out this TED-video to see how this can be done with medical data).

The second problem is however a bit more problematic. I would propose that data in itself – even in the right measure and presented the right way – will not necessarily lead to better decisions. Why? This is explained by the social psychological concept called “cognitive dissonance”.

Cognitive dissonance is essentially the discomfort we feel when we have two conflicting cognitions (beliefs, emotional reaction, values and ideas). Take for example the doctor who is smoking. He knows that smoking is bad for his health.  He may even know the exact science behind all the health problems smoking can cause but he continues to smoke anyway. If the doctor does nothing, he will continuously feel bad about his smoking habit – he will be plagued by guilt. However, because we humans don’t like this feeling  we will add a cognition to relieve us of this pain. In this example, the doctor can do several things: He may ‘accidently’ find research which questions how unhealthy smoking really is. He may conclude that smoking relieves him of his stress at work and therefore is worth the potential problems he may suffer later. He may conclude that because his father smoked all his life and didn’t suffer of any medical problems that he is genetically immune to smoking-related health problems. Whatever he choose to do, the doctor will create a belief that can make him live with his smoking habit.

In short, cognitive dissonance will create a bias for a certain decision despite other factors such as facts and evidence may favor the alternative. So even if HR executives are faced with facts supplied by analytics, it does not mean that he/she will use that data to make better decisions. Not if that data will create cognitive dissonance.

Let’s look at a simple example: The head of Talent Management has been presented with evidence which suggests that the current talent program has no tangible impact on productivity and talent turnover. Analytics is also able to show that the ROI on the program has been negative the last three years. However, the head of TM has not only designed and approved the project but she has also told the board of its successes and won praise for them. This presents her with a problem as the data suggests that she has not done her job well. The analytics data should now be used to change the program or to scrap it altogether, but it may not happen. The head of TM will have to find a way to live with this cognitive dissonance and not make that best decision.

So what to do? Social psychology theories suggests ways to overcome this bias. You can find a good overview on Wikipedia. Essentially, I believe a solution is to have a CAO (Chief Analytics Officer) who is powerful enough to challenge HR on the results. If analytics is a sub-department of HR or a non-powerful support function, the decision maker can get away with some of the typical cognitive dissonance strategies (avoidance, distortion, reassurance, confirmation, re-valuation).

Analytics is not a end it itself – it is a mean to create better HR decisions. Cognitive load and cognitive dissonance may stand in the way unless it is proactively dealt with.


  1. I think there’s a few other hurdles:

    – most executives don’t believe data that comes from HR, so time will be required to improve credibility.
    – many HR professionals are not highly skilled in interpreting data and analytics
    – much of the data produced won’t be clearly actionable in the short term.

    1. Hi Jason,

      You’re right. Cognitive dissonance (and other psychological theories) are certainly not the only (or even largest) reason. The three hurdles you mention are certainly big and for some gaining the fundamental trust that you as HR executive can deliver credible data is in my experience probably the biggest.

      Do you have any experience in how this may be achieved?

    1. Hi Charles,

      An excellent post you have written – thanks for highlighting it to me. I totally agree with your points, especially the one with the danger of pushing predictive analytics for HR.

      I have read the book by Kahneman – a great book. He is a long term favorite of mine and glad that you appreciate him and his work also.

      Question: You give an example with the Stepford-effect in recruiting. Are there any analytics-software out there which can take in and process data which overcome this issue?

      1. Morten,

        thanks- I think you and Kahneman covered it much more articulately 🙂

        As for recruiting specifically, this could be a completely new post, but I’ll keep my 2 cents as short as I can:

        1) That “Stepford-effect” I have witnessed in organizations without any recruiting or analytics software in place so, to be fair, it certainly can be a cultural issue as well as a technical one.

        2) Yes, plenty of new HR and Workforce Analytics vendors (nearly all of the most well known) are claiming predictive analytics in recruiting and many other areas- it’s a holy grail of sorts for software vendors trying to sell in this highly competitive space. The concerning part is they are all in their infancy. The sales pitch is chronically far ahead of the actual functionality in many cases.

        The fundamental issues I see:

        a) the volume of data to perform this kind of statistical analysis isn’t there or isn’t clean enough.

        b) the software is still very immature.

        c) as Kahneman said, “WYSIATI” What You See Is All There Is” …and that leads to very poor decision making.

        keep up the insightful work here, glad to have found your blog!

  2. ironic (or maybe not) timing for this article:


  3. The issue that your Head of Talent Management faces can, in my view, be better dealt with by using a Bayesian approach, where beliefs are updated as new data becomes available. With this approach the new data can be dealt with in a rigorous manner, modifying the original predicted ROI, not necessarily rejecting it.

    I fully support your view that measurement and determination of effectiveness of any activity should be independent of the person conducting the activity being measured. It is this reasoning behind why an internal audit function reports to the chairman, not the CEO. A HR analytics centre of excellence is the right approach (disclaimer: we provide these capabilities as a service).

    The other recommendation would be that measurement should be regular and frequent. We use prediction in HR in this manner (see: Our view is that the prediction should be used to increase confidence in decision making. A key issue with software making decisions based on predictions is to do with the absence of a loss function, especially as the costs of a wrong decision are likely to be asymmetric.

    Charles is absolutely right with the issue of data. In most instances in HR the issue is not big data (needed for accurate predictions) but small data. Good post on this here:

  4. Leaders ignoring HR data is no different from leaders ignoring any other type of data. This is commonplace, and it requires analytics professionals with the ability not only to analyze data, but to communicate it in a way that is impossible to ignore and to exert the influence necessary to compel leaders to make good decisions.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s