Skip to end of metadata
Go to start of metadata


Purpose

The purpose of this page is to clarify the process of rating normalization for the Matrix Grid Reports, how to use Custom Weights to customize this process, and how it is calculated the ratings when using the average, instead of latest.

Overview

Normalization of ratings is the process of converting a rating from one rating scale (1-5 rating scale) to a target rating scale (1-3 rating scale, for example) so that it can be displayed in some context. For example, a 1-5 rating coming from a Performance Form needs to be displayed in the Matrix Grid Report, which uses a 1-3 rating scale. Custom Weights can be configured to customize how this process happens. 

Normalized Ratings

Many instances are configured to use different Rating Scales for each module. As an example, you might use a 5-star rating scale for Performance forms but you have a 3-star rating scale for the Matrix Grid Report (3x3 box). Due to these differences, the system identifies when a rating is on a different scale and normalizes it to fit the scale being used by a different module.

Using the same example as above, let us assume the following Scenario:

  • Performance forms use a 5-star rating scale, from 1-5.
  • Matrix Grid Report is configured as a 3x3 matrix, so the Performance axis will use a scale from 1-3

Since the Performance Rating Scale (Admin Center > Succession > Matrix Grid Rating Scales) is set to a 3-star scale, any ratings added directly through the Employee Profile will use this 3-star scale. In the meanwhile, completed PM forms will show as a 5-star scale in Employee Profile.

When those ratings are used in the Matrix Grid Report, and one rating is on a different scale than the one being used by the Matrix Grid, the system will automatically normalize it. Below is how the system will do this for this specific example:

If a rating of 2 comes from PM form, it will show up at the Matrix Grid as a rating of 1, and so on. The system will do this for any ratings that come from a different scale; this behavior cannot be changed.

Below, we present two ways to calculate the normalization of ratings. 

Explanation 1

Calculating the ranges of the original rating that map to each of the target ratings is a simple process.You simply need to take the total range from the original scale and divide it by the number of ratings in the target scale. For example: 

Range = (maxO - minO) / nTarget

  • maxO = maximum value of original scale
  • minO = minimum value of original scale
  • nTarget = number of ratings in target scale

For the example above, this would be calculated as below: 

Range = (5 - 1) / 3 = 1.3333

This number corresponds to the range that will be mapped to each output. We can then build the following table: 

Range (Original Scale)Output rating (Target Scale)
1 - 2.331
2.34 - 3.662
3.67 - 53

 

Explanation 2 (useful when working with Custom Weights)

The default normalizing of ratings will always take the number of points being used by the rating scale to be adjusted and divide it by the number of points used by the rating scale that it is adjusting to. Like this:

Where:

X = target value you wish to convert.
A = Minimum value of the current rating scale.
B = Maximum value of the current rating scale.
C = Minimum value of the target rating scale.
D = Maximum value of the target rating scale.

As an example, if employee John Doe has received a rating of 2.3 on his PM form and we would like to know what would be the value of this rating on a 3-star rating scale, we would apply the formula above like this:

The result will be 1.65. Since each point of the 3-star rating scale is equivalent to 33% of the total scale, each value will be placed according to the following weights:

 

From

To

Weight

Rating

1

1.66

33.3%

1

1.67

2.33

33.3%

2

2.34

3.00

33.3%

3

 

Based on the table above, since 1.65 is less than 1.67, after the conversion the rating for John Doe will be 1 in the target rating scale.

For an easier way of identifying which will be the value of a rating after normalizing it, refer to the spreadsheet “Normalization” attached.

 

Custom Weights

In some cases, the default normalization is not adequate for a company’s internal process and needs. For this reason, the Matrix Grid Report allows you to set up Custom Weights for each placement, so you can manually configure where each rating should fall on the 9-box. By applying custom weights to the Matrix Grid, the calculation of the normalized ratings is changed according to the values supplied in the Matrix Grid configurations.

Custom weights can be defined under Admin Center > Succession > Matrix Grid Reports: Performance-Potential (or Matrix Grid Reports: How vs What, depending on which you wish to apply the custom weights), as can be seen in red below.

In this page, you can set up the specific percentage of each placement. As an example, if we would want to use custom weights to make ratings 2, 3 and 4 fall under “Medium” while rating 1 will fall under “Low” and rating 5 will fall under “High”, this is how we could achieve it:

RatingPlacementCustom Weight
1Low 20%
2, 3 and 4Medium 60%
5High 20%

Like for normalizations, you can refer to the spreadsheet attached to this Article for an easier way of converting and applying Custom Weights to rating scales.


Example: How to use Custom Weights to normalize Performance Ratings?

In this example, we'll consider that the rating is coming from a Performance Form, which is based on a 5 point rating scale. On the other hand, the Performance-Potential Matrix Grid is using a 3 point rating scale for both Performance and Potential axes. Therefore, the rating coming from the Performance Form (1-5 rating) will be normalized to fit into the Matrix Grid Rating Scale (1-3 rating). 

Original Rating Scale (from the Performance Form):

Target Rating Scale (Matrix Grid Rating Scale): 

The Standard Normalization process would convert the 1-5 rating to the 1-3 rating as indicated below:

range = (Max Original Scale - Min Original Scale) / Number of Options from Target Scale

In this case: range = (5 - 1) / 3 = 1,3333

Original rating

Target rating

1 - 2.33

1 

2.34 - 3.66

3.67 - 5

3

The requirement from the customer, however, could be to achieve the following mapping:

Original rating

Target rating

1 - 1.8

1 

1.8 - 4.2

4.2 - 5

3

To calculate the Custom Weights, we'll need to identify which percentage of the total range (5 - 1 = 4) corresponds to the range of each rating output.

For example, the range of the first rating output is 0.8 (1.8 - 1). So, we'll need to identify what's the percentage that 0.8 corresponds to in relation to the total range of 4.

Original rating

Target rating

Individual Range

Percentage

1 - 1.8

1 

0.8

20

1.8 - 4.2

2.4

60

4.2 - 5

3

0.8

20

Once we have the percentage calculated for each target rating, we can configure the Matrix Grid Report with the Custom Weights.

Go to Admin Center > Succession > Matrix Grid Reports: Performance-Potential or How vs. What > Enable "Custom Weights" > input the calculated percentages > Save the changes.

 

Average of all found ratings

If you want the system to calculate the average across all ratings and use that instead of just the latest rating for your matrix grid reports, it is necessary to enable the option "Use average of all found ratings, instead of latest" in Admin Center > Succession > Matrix Grid Reports: Performance-Potential (or Matrix Grid Reports: How vs What, depending on which you wish to apply the custom weights):

The system uses the date range entered for the matrix report to collect ratings and calculate the average. If the date range is undefined and empty, all ratings are calculated to get the average value.

When we using average rating, the system renormalizes each rating the user has, using the below equation:

renormalizedRating = (rating - oldScaleMin)/(oldScaleMax - oldScaleMin)*(newScaleMax - newScaleMin) + newScaleMin

WHERE:

  • rating = rating number

  • oldScaleMin = minimum value of rating scale

  • oldScaleMax = maximum value of rating scale

  • newScaleMax = maximum value of rating scale

  • newScaleMin = zero*

*NOTE: According to the configuration of Performance-Potential, in a rating scale of 1 to 3, the system uses the new scale of 0 to 3, as to map to 3 ratings, it should have 4 scales. That is why the system switches to 0-3, instead of 1-3.

And then, averages all normalized ratings, using the below equation:

averageRating = (Sum of renormalizedRatings) / (number of ratings)

Example of renormalization and average ratings

It is necessary to calculate the normalization of each rating for each scale (e.g. Performance, Potential). In the example below, we are calculating the normalization for only one scale.

Considering:

  • Rating scale of 3-points

  • User A has ratings "2", and "3";

  • User B has ratings "1" and "3".

Rating Calculation for User A

renormalizedRating = (rating - oldScaleMin)/(oldScaleMax - oldScaleMin)*(newScaleMax - newScaleMin) + newScaleMin

renormalizedRating "2" = (2 - 1)/(3 - 1)*(3 - 0) + 0 = 1/2*3 = 1,5

renormalizedRating "3" = (3 - 1)/(3 - 1)*(3 - 0) + 0 = 2/2*3 = 3,0

averageRating = (Sum of renormalizedRatings) / (number of ratings)

Avergare Rating = 1,5 + 3 / 2 = 2,25

Rating Calculation for User B

renormalizedRating = (rating - oldScaleMin)/(oldScaleMax - oldScaleMin)*(newScaleMax - newScaleMin) + newScaleMin

renormalizedRating "1" = (1 - 1)/(3 - 1)*(3 - 0) + 0 = 0/2*3 = 0

renormalizedRating "3" = (3 - 1)/(3 - 1)*(3 - 0) + 0 = 2/2*3 = 3,0

averageRating = (Sum of renormalizedRatings) / (number of ratings)

Avergare Rating = 0 + 3 / 2 = 1,50

 

Considering above average, user A will be plotted for the calculated axis, in the third box, and User B in the second box, as the system always rounds up the calculation.

For quick reference on converting the source rating to your matrix grid scale, please download the Normalization Calculator below: 

Related Content

Bridge KBA: 3025156

Related Documents

Succession: Implementation and Administration - Custom Rating Distributions Using Custom Weights

SAP KBA: 2341508 - Working with v12 Matrix Grid Reports