A practical guide to calculating Likert-scale percentages and weighted mean scores accurately, with worked examples using illustrative data.
AC 2.1 is a practical data-analysis task. You are required to work with the pulse-survey data provided for Camellia and calculate two types of results: the percentage of respondents selecting each Likert-scale option for each survey statement, and a mean score for each statement.
This criterion is different from the earlier discussion-based ACs. The priority is mathematical accuracy, a clear method and results that can be used in the later interpretation of the survey. The aim of this resource is therefore to teach the calculation process using separate illustrative examples rather than calculate the figures from your assessment dataset.
A Likert scale records responses across ordered categories. A typical five-point scale might run from Strongly Disagree to Strongly Agree, although the exact wording and coding used in your assessment data should always be checked before calculations begin.
For each statement, the raw data tells you how many respondents selected each response category. Percentage calculations make those frequencies easier to compare, while a numerical coding of the categories can be used to calculate a mean score.
Before doing any arithmetic, confirm the total number of valid responses for each statement and identify how the response categories are coded. Check whether there are blank or missing responses and whether the same number of people answered every statement.
This matters because the denominator used for a percentage should normally reflect the valid responses relevant to that calculation. Using an assumed total when responses are missing can produce an inaccurate percentage.
The basic formula is:
Percentage = (number of responses in the category ÷ total valid responses) × 100
For example, imagine a separate practice survey in which 40 employees answer a statement and 10 select Strongly Agree. The percentage selecting Strongly Agree would be:
(10 ÷ 40) × 100 = 25%
The same calculation is repeated for every response category for that statement.
Suppose an illustrative statement receives 40 valid responses:
The corresponding percentages are 10%, 15%, 20%, 30% and 25%. Together they total 100%.
This example is deliberately separate from the Camellia assessment data. Apply the same method to the figures in your own spreadsheet.
For a statement in which every valid response falls into one of the listed categories, the percentages should total approximately 100%. This provides a useful error check.
A total such as 99.9% or 100.1% can occur because individual values have been rounded. A materially different total may indicate an incorrect denominator, omitted category or calculation error.
A mean score provides a single numerical summary of the responses to a statement. To calculate it from Likert categories, each category is assigned a numerical value according to the coding used in the dataset.
For illustration only, a five-point scale might use 1 for Strongly Disagree, 2 for Disagree, 3 for Neither Agree nor Disagree, 4 for Agree and 5 for Strongly Agree. You should use the actual coding specified for your assessment rather than automatically assuming this sequence.
When you have the frequency for each response category, calculate the mean by multiplying each response value by the number of respondents who selected it. Add those products and divide by the total number of valid responses.
Weighted mean = Σ(response value × frequency) ÷ total valid responses
This is a weighted mean because a response value occurring more frequently contributes more to the final score.
Using the same practice data and an illustrative 1-to-5 coding:
(1×4) + (2×6) + (3×8) + (4×12) + (5×10) = 138
There are 40 valid responses, so:
138 ÷ 40 = 3.45
The illustrative statement therefore has a mean score of 3.45. This calculation demonstrates the method only and does not use Camellia’s assessment figures.
A mean score only has meaning when you know what the numerical values represent. If a higher number represents stronger agreement, a higher mean generally indicates responses towards the agreement end of the scale. If the coding is reversed, the interpretation reverses as well.
Never interpret a mean merely because it is numerically high or low without checking the scale. Also check whether any survey statements are reverse-worded, because their meaning may differ even when the numerical coding is identical.
Percentages show the distribution of responses. They allow you to see whether responses are concentrated in one category or spread across several categories.
A mean condenses the responses into one number, which can make comparisons across statements easier. However, two statements can have similar means while having very different distributions. For example, one may have many neutral responses while another has employees split between positive and negative extremes.
This is why it is useful to retain both forms of analysis rather than relying on the mean alone.
A spreadsheet can reduce repetitive arithmetic and make the calculation process easier to audit. Percentage formulas can divide each category frequency by the relevant response total, while a weighted mean can be calculated from the category values and frequencies.
Whichever software you use, check at least some calculations manually. A spreadsheet formula can be syntactically valid while referencing the wrong cells, and copying a formula across a table can reproduce the same error many times.
Use a consistent rounding approach throughout the analysis. Percentages may be presented to whole numbers or a chosen number of decimal places, while mean scores are often easier to compare when shown consistently, for example to two decimal places.
Avoid changing rounding conventions between statements. Keep the unrounded values in the spreadsheet where practical and round the displayed result, because repeatedly rounding intermediate calculations can introduce small discrepancies.
Before creating charts or interpreting results, check that:
These checks matter because errors at AC 2.1 will flow directly into the charts and interpretation required afterwards.
AC 2.1 establishes the numerical foundation. The next stage is not simply to reproduce these numbers in a different format. Charts should make important patterns easier to see, and interpretation should explain what those patterns may indicate.
Keeping a clear calculation table at this stage will make it easier to choose suitable visualisations and compare statements in AC 2.2.
The numerical examples on this page use an invented practice dataset so that the method can be demonstrated without completing the learner’s assessment calculations. Use the formulas and checking process to calculate the results from your own Camellia spreadsheet and retain evidence of your working as required by your assessment instructions.
Return to the 5CO02 resource directory to explore all assessment criterion study guides.
Specialist support for CIPD Level 3, Level 5 and Level 7 learners, with guidance from HR professionals.
