What are the different methodologies for calculating scores?
With scores, you can summarize the data in a Likert scale question into a single number. We have four preset scoring methodologies, which you can choose from: Top Two Box, Top/Bottom Delta, Weighted Average, and Top/Bottom Ratio. This article explains how each of them is calculated, and what they're generally useful for.
These scoring methodologies are applicable for 1-4 or 1-5 scales. In the examples below, I will use a 1-5 scale.
Top Two Box
This scoring method shows the percentage of people who give the top two ratings on a scale.
Calculation: % of 4-rating + % of 5-rating
This is the most common scoring method, appropriate for many use cases. It generates a clean and concise score, as we receive a comparable score across all brands. Using this method cuts the white noise of both ‘I don’t know’ and ‘neutral’ responses, but in the process also cuts out the low ratings.
Top/Bottom Delta
This scoring method shows the difference between how many people gave the top 2 ratings vs the bottom 2 ratings.
Calculation: (% of 4 rating + % of 5 rating) - (% of 1-rating + % of 2-rating)
This is similar to how NPS scores are calculated, and it considers both the positive and negative balance of responses. This methodology focuses only on the opinions on brands, also cutting out the white noise of both ‘I don’t know’ and ‘neutral’ responses.
Weighted Average
This scoring method multiplies the percentage of respondents for each score by the score value, and sums them up.
Calculation: (% of 1-rating x 1) + (% of 2-rating x 2) + (% of 3-rating x 3) + (% of 4-rating x 4) + (% of 5-rating x 5)
The weighted average considers all scores - negative, neutral, and positive. It is a scoring system that is consistent with other brand evaluations, e.g. Uber rating, Amazon reviews, and therefore provides a readable 1-5 score. It can be a blunt tool, as it doesn't capture nuance, like the distribution across different scores and shifts in those over time.
Top/Bottom Ratio
This score looks at the percentage of responses with the top two ratings, out of the total non-neutral ratings.
Calculation: (% of 4-rating + % of 5-rating) / (% of 1-rating + % of 2-rating + % of 4-rating + % of 5-rating)
This scoring methodology gives brands an objective and balanced analysis and is useful for an index where you need to compare many brands of different categories against each other. It considers both the positive and negative balance to brand responses and also focuses only on the opinions on brands, cutting the white noise of the ‘I don’t know’ and ‘neutral’ responses. Considering only active reactions to brands means that awareness of a brand would not skew the data.