Hack:
Appraisals and the need to bell the ‘Bell Curve’

The performance of a company is intrinsic with the employees associated with the organisation. Every employee and his contribution is part of the core ethos of a successful organization. This has lead to various appraisal techniques used to analyse the individuals share to the company’s objective; one such popular model used is the Bell Curve.

The Bell Curve or the Gaussian curve is one such model. This model is based on law of averages or standard deviation. Picture a Bell shape on a graph paper; the peak of the curve is the performance of majority of the employee within an organization who will gravitate towards the ‘centre’ where as the tail accounts for the underachievers and top performers. However ‘Bell Curve’ as a fitment model is anything but ideal. For a start appraisals involve people.

The most important drawback of using the Bell curve is that it cannot account for individual behavior and traits. This is true for any model used where people are part of the system. In all such cases the conventional models fail to stand the test of time. Take the example of the use of Gaussian Curve in financial market by risk analyst or other statistical applications like predictions of earthquakes (I would do better if I was to toss a coin to make my next prediction; the odds are I will be on par with the analysts using such models). Relevant to this discussion how well does this BELL CURVE fitment work for Performance management system?

For the sake of argument considers that you have worked exactly on the same problem for a year each with equal diligence. But you have a different manager for the each of the 6 months. Odds are that your ratings will wary because the perception of 2 managers will almost never be the same. There cannot be a consistency in measuring performance and fit individual’s to the model. Some people are bound to get a better deal while others may feel that they have been undone.

So what are the points that need consideration in addition?

• Variability matters – Every project, every team and every member is different. To measure someone using the same yard stick may not always be correct. As is the case with anything that involves people the strengths of individual may vary. Accounting for this variability is important. It is also important to consider the kind of work that an individual is doing. An organization has many roles that need addressing. Some people are required to dabble with multiple roles due to the nature of their work while others are required to focus on a particular role because of its criticality. All these factors cannot be accounted for in appraisals.

• Predictions are susceptible to failure - Account for forecast degradation against the period (difference between short period planning 3 months and long period planning say 1 year) - predictions for a year almost never materialize the way we plan them. The fundamental problem is actually with us i.e. people. It has to do with the way we have evolved. Plans fail because we fail to comprehend the source of uncertainty outside the plan associated to the activity. Let me explain with an example. Researchers in an experiment asked college students to plan their final year project. They divided the students into two groups; optimists and pessimists. The optimists promised to do it in 26 days while the other lot had 47. The result- the average completion took 56 days. There will always be differences in perception. Of course there are incentives that push for a shorter time frame. But planning failure exists even when there is no client or external pressures. We are so focused on matters internal to the project that we fail to consider the unknown. Focus is always on what is known and tangible, the abstract unknowns like loss of days in office due to heavy rains or sudden ill health of team players or economic down turn that results in adverse impacts on the project etc are too abstract for us to talk about or consider.  We do not plan because we do not understand the future. That does not mean we cannot plan at all. We can plan if we account for our limitations therefore planning and accounting any individual or team against that needs to be shorter not more than 3 months, in a feedback loop mechanism rather than an elaborate system. A paper - ‘Financial Applications of Random Matrix Theory’ (you can find it http://arxiv.org ) by Jean Philippe Bouchaud drives home this point. It summarised that we are very bad at predictions be it in any field. The research concludes that the top brokerage houses and the security analysts predicted practically nothing. Someone who perhaps looks at the previous quarter of the company’s record would probably do equally bad as compared to these white collared gurus. This in spite of all the information that they have on hand future contracts, planned expenditure current fiscal health. In fact these crystal ball gazers had significantly larger errors then the average errors made by individual forecaster’s ordinary people who deal in shares and study them. Again importantly this is also applicable to various other scenarios like performance where people are an integral part of the system.

• Promoting Mediocrity - Bell Curve essentially works on averages and promotes mediocrity. The notion of a Man (or women) deemed average is different from that of a Man who is average in everything that s/he does. When we say ‘Average Man’ then what it really means is to be half male and half female! This distinction of being average in the tasks that the individual needs consideration is completed missed and that he can perhaps excel in some other task cannot be identified in the Bell curve model because the Bell curve cannot quantify against individual skills. By labeling people into bands of average, below average and above average we end up doing the first labeling i.e. calling them as ‘Average Man’ which is the fallacy of the Bell Curve.

What is the alternative? I believe it’s more important to suggest alternatives rather than just highlighting the short coming of anything we analyse. And I put forth 2 here

• Collective Marking: this approach involves scaling people collectively not as individuals.

·    This will ensure that there is a cohesive, coming together of people because all of the team players will tend to align themselves with each other to ensure that collectively the team does better each day.

·    Group assessment will also ensure that the manager is not rating people but a team and many of the discrepancies highlighted above will be negated.

·    More often than not a team involves people from different streams particularly in the delivery styled projects. This will ensure that the people who are the closest to the day to day activity of the project in the entirety are the ones who are evaluating the team as a collective unit and rating them to that effect.

·    It encourages bonding individuals within the team and encourages them to push each other for excellence since their individual performance is also dependent on the performance of ‘the other team mate’. And in a multi vendor environment where you have multiple players working together if all adapt to this kind of evaluation system as a standard certified system this would work wonders.

Certainly there will be a flip side to this. The above is not a mathematical model and therefore though the idea sounds good it may not be accepted as a standard across all organisations.  The team branding will discourage the mavericks to push for more and try and supersede their already excellent acumen.

The other option however is a mathematical model that does caters to the individuality, the Mandelbrot’s fractal set and I propose this model can be used for a more effective way of performance Management System.

• Mandelbrot’s Fractal Model:  A fractal is a pattern or shape whose parts echo the whole.  Fractal geometry is about spotting repeating patterns analyzing them and quantifying them. It is a tool of synthesis and analysis. The pattern that arises can take many forms and shapes. When we talk about performance appraisal we know that the factors that govern the appraisal are common across the organization and this is the key. A matrix of factors that are deemed as critical factors for each individual in the organisation.

·    We plot every individual employee’s performance in the organization (y axis) against the combined set goals project wise and organization wise (x axis) you will see a pattern emerging. The job of plotting this on graph can be easily achieved by using the help of computer programmes. 

·    Once this is done the job of aligning the resultant pattern into bands based on the fortunes of the company for that year is step two. This will be a mechanism where rather than fitting people into bands we do the reverse; evaluate people purely on their performance and then quantify the growth of each individual based on the company’s performance for salary and the organizations policies for promotions.

The idea of using Mandelbrot’s fractals for appraisal is still very much just a concept (work in progress ) something I am currently perusing. However I table this idea to encourage the reader to contemplate and facilitate a better mechanism of apprising for the future.