Can We Adjust the Bell Curve in Performance Reviews to Reflect Real Employee Contributions?

Hi everyone,

I understand the purpose of a bell curve in performance management and reward systems. My question is: can the bell curve be skewed? In other words, instead of using a normal distribution where 75% are average performers, 10% are above average, and 15% are below average every year, can the bell curve change shape based on yearly performance? For example, could 80% be considered average, 5% below average, and 15% excellent?

In my company, we currently utilize the standard distribution, where individuals are "fit" under the curve. However, this approach seems unusual to me. Shouldn't the bell curve be determined by the role of performers rather than the bell curve determining the role of performers?

Thanks

Hi everyone,

I understand the purpose of a bell curve in performance management and reward systems. My question is: can the bell curve be skewed? In other words, instead of following a traditional bell curve distribution where 75% are average performers, 10% above average, and 15% below average, is it possible to have a different distribution based on yearly performance, such as having 80% as average, 5% below average, and 15% excellent performers?

Currently, my company uses the standard distribution, and individuals are placed under the curve accordingly. However, I find this approach somewhat unusual. Shouldn't the bell curve be determined by the role of performers rather than the performers being determined by the bell curve?

Thanks.

Namaskar,

From the situation described above, I do not see the typical bell-shaped curve or normal probability curve. The employees are already screened, so in performance measurement, a negatively skewed curve is likely to be obtained, with poor performers tapering towards the left side. The specific percentages are cutoff points that depend on management decisions. Performance measures should determine the curve, not the other way around.

Regards,

Statistically speaking, a bell curve can be skewed, but here the problem is not about the shape of the curve or the level of skewness. The issue lies with the company's policy of performance evaluation. Your company has decided on a fixed bandwidth, which has forced the distribution to rank employees in either of the three categories. Only x% of people can belong to each category.

Thanks, Dr. Jogeshar,

That's what I needed to confirm. We seem to be doing the opposite by putting people under the standard distribution.

How can you put people under a standard distribution? They will occupy positions based on their performance scores. The performance scores will determine the nature of the curve. You cannot manipulate it.

Regards

Yes, why not... Depending upon:

1. Business Performance
2. Age of your corporation

For example, if you are a startup company, depending upon the gestation period, you may or may not be profitable. However, most of your team members could be high performers, which will determine your positively skewed curve and vice versa.

As a company, you need to decide whether you need to have a normal curve or a skewed curve, and accordingly, fit people. Trust this helps.

Yes, why not... Depending upon:

1. Business Performance
2. Age of your corporate

For example, if you are a startup company, depending upon the gestation period, you may or may not be profitable. However, most of your team members could be high performers, which will determine your positively skewed curve, and vice versa. As a company, you need to decide whether you need to have a normal curve or a skewed curve and accordingly fit people. Trust this helps.

1. High performers will determine a positively skewed curve; this is news to me.
2. I believe a company always needs high performers. It is also news to me that a company may choose low performers too.

Regards

Hi,

Can anyone please explain to me what exactly a bell curve is and its functionality? As a student, I have never come across this concept. A little insight into its benefits and drawbacks would also be helpful.

Thank you.

Regards,
Vinisha.

Vinisha ji,

Here is the concept:

1. Take the heights of all the girls of your age, let's say about 100 girls.
2. Take out graph paper. On the graph paper, the X-axis represents inches, and the Y-axis represents the frequencies of girls on each inch. Plot the frequencies over every inch and join the points. You will see a bell-shaped curve. If you fold the graph at the point of the mean, the two halves will align exactly. Almost all human characteristics, like job performance scores, are distributed on a graph like this.

To understand the concept better, please refer to any elementary book on statistics for behavioral science and go to the chapter on central tendencies, skewness, and kurtosis.

Regards.

Hi Siddiqif,

When you analyze your data, if your trend looks like a normal (not skewed) curve, then that means that the performance variations are random and that there is no single factor that affects the performance issue significantly. The curve will be a normally shaped curve and will not be flat (it will have some height). Thus, a normal distribution of the variations causes a bell-shaped curve.

If the analysis reveals a skewed curve, then you are presented with an opportunity to identify high and low performers on either side of the average. The curve will appear as if it has been squashed to the right or left. Either way, you can see who has done better than others and zero in on their practices and experiences.

With these kinds of tests, all comparisons are relative - for example, high performance could be one standard deviation above the mean, etc. The exact value of the mean will fluctuate and be determined by the data that goes in.

I think your organization has set limits on metrics, such as 80% and above always being excellent, and 60% and below always being poor, and so on. The problem with this is that if your organization's performance improves with time (which it will in any case), that 80% and above becomes a very poor cutoff and should be upgraded to 90% or something like that.

The true use of this curve is to compare performance over time. I would suggest that you get your people together, show them the results, and do the same after a month so they can compare and see what you are trying to do. This improvement is what the management will want, rather than arbitrary metrics and numbers.

My Rs. 2/two cents.

Thank you, Vadivelu,

That gives me something to think about and clarifies the issue. I think the problem is the use of the bell curve. We are using a performance management system, but as far as I can identify, the bell curve is used more for rewards and increments than to identify the weak and strong performers and their development plans. What other ways are there to identify strong performers versus weak?

Hi Siddiqif,

Glad to hear that you found my response somewhat useful.

Some other methods that come to mind with regard to identifying strong performers:

* Extant data: Records like previous performance appraisals, customer surveys, sales reports, etc., are a good source for obtaining information on performers who consistently perform above the average.

* Surveys: It is not too difficult to implement a survey within your organization and identify key stakeholders. One of the questions could be "Who do you go to when you have an issue with..." If a person's name consistently surfaces, then he/she is likely to be a good performer. You can then interview them directly to get their opinions.

* Competency analysis: Analyze the behavior of some good performers and break it down in terms of competencies that characterize optimal performers. This will enable you to see which skills are lacking in some people, and you can then undertake specific training/non-training interventions.

* 360-degree evaluations are a hot thing these days. Get someone evaluated by seniors, peers, and juniors to identify what they are doing right.

As an HR representative, you should be able to approach people with a high degree of confidentiality and extract this kind of information.

A word of caution, though: identifying strong performers is easier and preferable to finding out who weak performers are. For example, if you ask people to rate others on a scale of 1-5, people rarely will speak ill of their colleagues and would try to rate everyone as 3, 4, or 5. This doesn't give you any useful information at all.

Instead, try a positive approach by asking them to rate one person who they would approach if they had a work-related problem. This will get you useful information like who is knowledgeable about something, etc.

Take care,

Ram

Hello,

This theory fits only for the armed forces and other government departments where "chamchagiri" outwits performance.

Harikeyel

So, is forced ranking about fitting people under the curve? What are the advantages or reasons for using forced ranking? I suppose that, because my company is a large public sector organization, forced ranking makes reward management easier.

Clearly, the performance management process is not working at your organization. Scrap your current system of forced ranking, but be sure to keep the concept of performance management intact.

Forced ranking may work for some organizations, but it can be very risky. You must delicately balance your company's vision and organizational strategies while performing activities that support them. Forced ranking systems are difficult to communicate, tough to administer, and rarely simple to execute. Ultimately, strive for simplicity in managing performance. Few corporate cultures can truly weather an approach that devastates morale, with employees essentially competing among themselves for coveted higher rankings.

The most important thing to remember is that all processes aimed at managing performance should adhere to three simple and basic principles. They must:

- Link to your company's vision.
- Support your business strategies: the tactical "what you do" portion of your vision. Strategies normally disseminate from the top down into each department, business unit, division, etc., and define the individual performance criteria against which employees are managed.
- Be simple to use and be practiced regularly. Your managers and employees should discuss performance often. Make sure there are no surprises. Performance management works best when employees know exactly what they have done well (or poorly) and how their activities clearly relate to whatever financial incentives are offered.

This ideal process is simpler to execute than forced ranking, but it takes time to define and build, not to mention a commitment from your top management as well as your employees. You also must make a commitment to training and communicate the process to ensure consistency and efficiency. Ideally, performance management should tie into other aspects of managing talent, including job descriptions, performance review forms, and your overall compensation philosophy and programs.

Regardless of the process you adopt, pay special attention to any concerns, likes, and dislikes voiced by your employees. This could lead you to discover new opportunities for improving the process as you get feedback. Listening to employees is crucial in creating an effective performance management system.

An interesting thread - thank you to all who contributed.

Some thoughts and observations spring to mind.

If you have a performance scoring system that looks for below average, average, and above average, how can you be sure that even the 'above average' are actually good enough? Such a scoring system is relative - it's all about scores relative to each other - in effect ranking everybody. The No 1 person can still be below the required performance.

I believe the normal distribution curve can be a distraction here - what we in the UK call a 'red herring,' meaning that it takes your attention away from the real issue.

If you have a selection and hiring process that virtually always brings you good quality hires and by definition, these people are above the minimum required standard, depending upon your scoring system, you may have no kind of curve at all! Trying to force the distribution into a normal one is possible - it can be manipulated - but only by changing something fundamental in your processes.

If you manage to develop and use a scoring system that both ensures a minimum acceptable standard of employee and is able to produce sufficient definition that it is possible to plot a distribution, whether it is skewed or normal can tell you something about the nature of your processes used to measure and/or deliver performance.

Those of you familiar in detail with the tools and concepts within Six Sigma and Statistical Process Control will know that a normal distribution represents a process that is stable and in control - exhibiting common cause variation - i.e. the variation is random, natural, and entirely the product of the process concerned. To improve the distribution in the sense of reducing variation - i.e. making the curve more 'peaked' means changing the process concerned. A skewed distribution will tell you that something, in particular, is dominating - what is known as special cause variation. It could be variation in the measurement system itself or any of the inputs to the process and/or constraints upon it. Usually to reduce variation, you will be focusing on something outside of the process.

Be wary of getting overly concerned with distributions of any kind and more concerned with ensuring that as many employees as possible exceed minimum standards of performance. Once you have achieved this, then it is time to start getting into the detail!

Kind regards,

Martin

Hi!

The initial question raised was quite simple, but the responses revealed the complexity of the subject matter. The curve referred to as "bell" refers to the "normal distribution curve." It got its popular "bell curve" name from the shape of the curve that it created.

The use or application of the curve in performance management has nothing to do with its real statistical definition. The curve simply became a good tool to rationalize organizational rankings (forced ranking) and benefits distribution/entitlements.

As such, to answer the original question: the answer is YES. You can skew, depending on the intent or objective of your organization. But be sure that you can explain and defend your action. Never go back to the original definition of the normal distribution curve because it will not support your action.

Best wishes.

Ed Llarena, Jr. Managing Partner Emilla Consulting