College days

Those  are the days, I see through the haze,
Of the time gone by, With my vintage eye,
Lovely friends and friendship too,
Feelings of a myriad hue,
Days of arguments and fights,
Of all the tiny wrongs and rights,
Days of vast leisure and play,
Feeling full of hapiness and gay,
Of time spent in wonderous talk,
Around the campus on a serene walk,
Of the many tests and night-outs,
Followed by disappointed shouts,
The scratchy movies in the auditorium,
The audio punishing the ear drum,
Waiting for the posty to come,
What tidings does he brings from home ?
Of sickness and home-sickness too,
Only friends to take care of you,
The warden in his haughty march,
Causing a naughty throat to parch,
The cold winter and the hot days,
Withering dryness of desert ways,
Does you eye fill up with tears,
Longing for those wonderful years ?

The obnoxious practice of bell-curve fitting

Come appraisal time, there is one term that is on everyone's lips. A hated term that raises hackles. A term that conjures up the worst nightmares. It is the "bell-curve fitting". For no fault of its, the bell curve has come to be associated with a practice that employees of most corporates have come to despise.

From a statistical perspective, the bell curve is a frequency distribution curve. It shows the frequency of the measured variable for different values of the variable. It also known as the Gaussian or Normal distribution, the latter nomenclature being used in Statistics more frequently. The distribution is shaped like a bell and hence the name. A bell curve is fully defined by two parameters - the mean and standard deviation (SD). It is used widely in quality control. For any set of data, it is always possible to calculate the mean and SD. But that does not necessarily mean the data itself is normally distributed. The data must be tested for normality. There are various tests that can be used. The variable must be a continuous variable i.e., it must be able to take any value in a given interval. Typical variables are height, weight, age etc. This is about the bell curve.

Where would the bell curve apply ? If in a company there are 1000 Sales representatives and each representative has a sales target of Rs. 100,000, then the actual sales figures could reasonably be expected to lie in a bell curve. However, this needs to be confirmed with a test for normality. If confirmed, the data could be plotted as a bell curve. The performance of the Sales Reps could be gauged from the plot. The Average performers would be bunched around the mean. To the right would lie the Stars and to the left, the Laggards. The Average performers would be the largest in number. The other two categories would be less and, ideally, equal. One could have 5 categories instead of 3. The company could decide how to split the categories and the percentage of people that should be in each category could be arrived at from the graph.

The trouble starts when this innocuous curve is applied to the performance of employees. An employee typically has a number of key performance parameters or KPP. These could be revenue, profit/project margin, attrition, value addition etc. Each has its own unit of measurement. Revenue and value addition are measured in currency - Rupees, Dollars, Euro etc. Margin and attrition are measured in percentage. Each KPP target is given a weightage in percentage such that the sum of all weightages is 100%. An employee's performance is measured by scoring the employee for each KPP target on a scale of, say, 1 to 10, with 1 being the lowest and 10 the highest. A weighted average is determined and this is the overall score of the employee. The scores for all employees are similarly calculated and plotted. A bell curve is applied or rather, forced on the data. Stars, Average performers and Laggards are determined. All this seems very logical.

But let us look at the hidden facts.

1. What is plotted is the score of the employee. This is just a number and there is nothing normal about it.
2. The score is calculated from a number of numbers and their weightages. Nothing normal here.
3. No test is carried out to check for normality.
4. The score includes the bias of the appraiser. One appraiser may score an average of 9 for a group and another may score 7 for the same group. One is lenient and the other is more stringent.
5. The bands are split arbitrarily. There is no logic in saying a team of 100 has 20 Laggards. How is this determined ? This  percentage is applied across the company with no rhyme or reason.
6. There is a vague and opaque process called normalisation that is supposed to take care of the appraiser bias. How this works is a mystery.
7. Corporates have a delightful way of selling this and other such practices to employees. They employ a consulting firm that comes up with such wonderful ideas. These are "best practices across the industry","industry standard practices", "modern methods of evaluation" etc. If it does not work, they quietly dump it for another system. No apologies, no regrets. And if at all anyone is to be blamed, it is the consulting firm. Sweet, isn't it ? And don't be shocked if the company goes back to the same consulting firm for the next set of "industry bench-marked practices". 😠
8. And the best piece of cake ? Ask the HR what the bell curve is and chances are they will not know !😁😁😁

So what is done, in effect, is that the bell curve is forced on data that does not fit the bell curve. Its like wearing a shoe on the head - incongruous and painful. But there is something to cover the head, so why bother if it doesn't fit !👻👻👻
But the consequences could be bad for employees. Laggards would get less or no bonus payout. Average people would get paid but less than the Stars. Career growth would also be adversely affected.
But then who cares ?

Note :

1. The term Laggard is used only for purposes of illustration and nothing derogatory is implied.
2. The word "normal" used here is in the context of the Normal distribution.
3. The terms used could vary from company to company.
4. The method of bell curve fitting and scoring could also vary but the practice would be similar.