Color Codes

The Perceptyx platform leverages a color coding system in the Trend and Demographic Crosstab Reports to highlight the results of the scaled survey questions, to provide visual guides for easier analysis of the data. Each color coding option serves a different purpose in the report – whether it be helping the viewer identify what scores are higher or lower than others, helping understand significant changes in trend, or helping understand whether a certain group’s scores are significantly higher or lower than the company overall (to name a few). 

This article walks through:

• Color Codes in the Trend Report 
• Color Codes in the Demographic Crosstab Report

Color Codes in the Trend Report 

The Trend Report offers four color coding options:

• Absolute 

• Relative 

• Significance 

• Outliers 

Please note that color coding will be available at the category level only if all items in a given category have a trend score for the current listening event and for the prior listening event which is being compared to. 

Absolute

The Absolute color coding option colors cells based on their absolute “favorability” score, broken into five equal buckets from 0 to 100 (0-19, 20-39, 40-59, 60-79, 80-100). As a reminder, “favorability” scores represent the percent of respondents who responded “Agree” or “Strongly Agree” to a question. Absolute color coding can be applied to all columns of trend data within the Trend Report. However, you can partner with your Program Manager and request that the color coding not be applied to the oldest trend comparison in the report (located in the column furthest to the right, on your screen).

Relative

The Relative color coding option calculates percentile rankings based on the favorability scores for each question and/or category within the report (it highlights the range of scores). These percentile rankings are independent calculations for each row (i.e., question or category), so the favorability scores and color coding for one question have nothing to do with the percentile rankings for a different question.

The percentile rankings, and different color coding options, are broken out into five buckets: 1st - 20th Percentile, 21st - 40th Percentile, 41st - 60th Percentile, 61st - 80th Percentile and 81st - 100th Percentile.  

If there is only one trend comparison, and therefore only two total columns of data in the report, each question/category will always have one score colored dark blue and the other will be red (unless the score is unchanged from the prior listening event). If there are two or more trend comparisons, it is possible for other percentile rankings (and therefore color coding options) to be displayed depending on where a trend score falls relative to the other scores for a single question/category. If there is very little variation in favorability scores across the trend comparisons, it’s likely all cells for that question/category will be dark blue or red. If there is greater variation across the trend comparisons, other percentile rankings/colors are likely. 

With this color coding, it is important to remember to analyze the data one row at a time (i.e., one question or one category). Additionally, please note that the percentile rankings/colors are not indicative of the full story for a given question/category. For one question, a cell could be dark blue for the current listening event, meaning the favorability score is higher than the prior listening event. But, it could still be the lowest rated item in the listening event, and therefore still an area of opportunity. 

Significance

The Significance color coding option highlights whether scores have gone up or down (and whether those are significant or not) compared to the prior listening event. The significance calculations are based on a two-proportion Z-test, calculated at a 95% confidence interval. This is a statistical formula that uses the score, change, and sample size to determine the significance of change.

For larger groups, smaller changes in scores may be deemed statistically significant while smaller population sizes may experience the same change in score but those changes may not be deemed statistically significant, due to the n-size.

Please note that the oldest trend comparison displayed in the report (located in the column furthest to the right, on your screen), will not have any color coding. The significance color coding for each cell is based on the prior listening event only. For example, let’s say the current listening event happened in 2024 and there are two trend comparisons: one for 2023 and another for 2022. The color coding for the 2024 data will  be based only on the change from 2023. And the color coding for the 2023 data will  be based only on the change from 2022. Because there is no trend comparison prior to 2022, that data can not be color coded. 

Outliers

The Outliers color coding option highlights cells if the change in score from one listening event to the next is above or below predetermined thresholds. Similar to the Significance color coding, each cell only compares to the prior trend comparison. Unlike the Significance color coding, this option does not take n-size into account.

Color Codes in the Demographic Crosstab Report

The Demographic Crosstab Report uses six color coding options:

• Relative (Compared to each other)

• Absolute (Compared to absolute threshold)

• Outliers (Compared to [overall] / [weighted average])

• Trend (Significance of change)

• Trend (Actual change)

• Internal Comparison (Significance of difference)

Relative (Compared to each other) 

The Relative (Compared to each other) option compares scores in or across groups, and highlights the spread of scores. It compares the range of scores generated for a particular item or category across all demographic options and indicates what percentile rank a score falls in. This is the same type of color coding as the “Relative” option in the Trend Report. 

These percentile rankings are independent calculations for each row (i.e., question or category), so the scores and color coding for one question have nothing to do with the percentile rankings for a different question.

The percentile rankings, and different color coding options, are broken out into five buckets: 1st - 20th Percentile, 21st - 40th Percentile, 41st - 60th Percentile, 61st - 80th Percentile and 81st - 100th Percentile.

When only two options are being compared (e.g., male vs. female), high is the 100th percentile (dark blue) and low is the 1st percentile (dark red). Therefore, Relative color coding is more helpful to use for demographics with more options as it gives a wider range of scores. Additionally, please note that the percentile rankings/colors are not indicative of the full story for a given question/category. For one question, a cell could be dark blue, indicating that particular group rated higher than other groups in the demographic being analyzed. However, that question at the company overall level may be the lowest rated item, and therefore the question as a whole is still an area of opportunity. 

Absolute (Compared to absolute threshold) 

The Absolute color coding option colors cells based on a range a particular score falls into. For favorability scores, that range is broken into five equal buckets from 0 to 100 (0-19, 20-39, 40-59, 60-79, 80-100). Absolute color coding is useful to see the ranges in which scores fall for different groups within a demographic. 

Outliers (Compared to [overall] / [weighted average])  

The Outliers color coding option highlights cells if a particular demographic group’s score is above or below the view/filter overall, based on predetermined thresholds. The default thresholds will only highlight scores that are at least +/-15 points from the overall group score. You may see the parenthetical text for this color coding option labeled as “compared to overall” or “compared to weighted average” but those two views are the same if there is demographic data available for all respondents in the current view/filter. If some demographic data is missing from the respondent datafile (e.g., the organization only has tenure data for 90% of the employees), then the overall group score and the weighted average will be slightly different. The weighted average is calculated only based on the data available within a given demographic cut.

Outlier color coding is helpful to detect scores that differ greatly (above or below) from the overall score and is most useful when comparing data across many groups (e.g., business units or locations). However, keep in mind that this option does not take n-size into account. So, analyze outliers accordingly – particularly for groups with small n-sizes. 

Trend (Significance of change) 

The Trend (Significance of change) color coding option highlights whether scores have gone up or down (and whether those are significant or not) compared to the prior listening event. The calculations use a T-test when comparing mean scores, and a two-proportion Z-test, calculated at a 95% confidence interval, for comparing favorable scores (most customers use favorable scores). The Z-test is a statistical formula that uses the score, change, and sample size to determine the significance of change.

For larger groups, smaller changes in scores may be deemed statistically significant while smaller population sizes may experience the same change in score but those changes may not be deemed statistically significant, due to the n-size.

Similar to other reports, a trend score will be available at the category level only if all items in a given category have a trend score for the current listening event and for the prior listening event which is being compared to. 

Trend (Actual change)

The Trend (Actual change) color coding option highlights cells if the change in score from the current listening event to a prior event is above or below predetermined thresholds. The default thresholds will only highlight scores that have changed by at least +/-15 points from the prior listening event. This color coding is useful to quickly see where the biggest increases/decreases have occured across various demographic groups. Unlike the Significance color coding, this option does not take n-size into account.

Similar to other reports, a trend score will be available at the category level only if all items in a given category have a trend score for the current listening event and for the prior listening event which is being compared to. 

Internal Comparison (significance of difference) 

The Internal Comparison (significance of difference) color coding option follows the same format/statistical calculations (Z-test) as the trend significance color coding. But, instead of comparing a group’s scores over time, various demographic groups’ scores are colored depending on whether they are statistically significantly different from the group/organization’s overall score. If a score is significantly higher than the comparison, the score is colored dark blue. If a score is significantly lower than the comparison, the score is colored red. If a score is higher or lower compared to the comparison but the difference is not statistically significant, light blue or yellow coloring is used. This type of color coding is useful for a quick overview of which groups are statistically different from the comparison group’s scores.

Please note, in the “Internal Comparison” dropdown menu, a user has the ability to update the group being compared to. The overall group/company is the default, but a group from the current demographic being viewed can be selected. For example, if the “Department” demo is being analyzed, a user could select the “Finance” department and compare all other departments to the Finance department’s scores.