About

Aggregating ratings without proper statistical treatment introduces systematic error. A simple arithmetic mean ignores the fact that not all ratings carry equal credibility: a panel-expert score and a casual user score contribute identically, which distorts the composite R̅. This calculator computes both unweighted and weighted means, reports population standard deviation σ, median, and mode so you can detect skew and outliers before publishing a final score. It maps numeric results to letter grades using configurable scale brackets.

The tool assumes ratings are drawn from a bounded ordinal scale (default 0 - 10). Weighted mode uses the formula ∑w_ir_i∑w_i. If all weights are equal the result collapses to the arithmetic mean. Limitation: ordinal data treated as interval data is an approximation; interpret sub-decimal differences with caution.

Formulas

The arithmetic (unweighted) mean of N ratings:

R = 1N N∑i=1 r_i

The weighted mean, where each rating r_i carries weight w_i:

R_w = N∑i=1 w_i ⋅ r_iN∑i=1 w_i

Population standard deviation:

σ = √N∑i=1 (r_i − R)²N

Where: r_i = individual rating value, w_i = weight assigned to rating i, N = total number of ratings, R = arithmetic mean, σ = population standard deviation.

Reference Data

Letter Grade	GPA Equivalent	Percentage Range	Rating (0-10 Scale)	Descriptor
A+	4.0	97 - 100%	9.7 - 10.0	Exceptional
A	4.0	93 - 96%	9.3 - 9.6	Excellent
A−	3.7	90 - 92%	9.0 - 9.2	Very Good
B+	3.3	87 - 89%	8.7 - 8.9	Good
B	3.0	83 - 86%	8.3 - 8.6	Above Average
B−	2.7	80 - 82%	8.0 - 8.2	Satisfactory
C+	2.3	77 - 79%	7.7 - 7.9	Average
C	2.0	73 - 76%	7.3 - 7.6	Below Average
C−	1.7	70 - 72%	7.0 - 7.2	Marginal
D+	1.3	67 - 69%	6.7 - 6.9	Poor
D	1.0	60 - 66%	6.0 - 6.6	Barely Passing
F	0.0	0 - 59%	0.0 - 5.9	Failing
Common rating scale benchmarks
-	-	-	4.0 / 5	Amazon "good" threshold
-	-	-	7.0 / 10	IMDb "decent" threshold
-	-	-	60 / 100	Metacritic "mixed" cutoff
-	-	-	3.5 / 5	Yelp median restaurant
-	-	-	4.2 / 5	Google Play Store median

Frequently Asked Questions

Use weighted average when rating sources differ in reliability or sample size. For example, if one critic reviewed 50 products and another reviewed 3, assigning higher weight w_i to the experienced critic reduces noise. If all raters are equally credible, weights are unnecessary and both formulas yield the same R.

Two products can share the same mean of 7.0 but have vastly different distributions. A standard deviation σ of 0.3 indicates consensus; σ of 3.1 indicates polarization (love-it-or-hate-it). Always report σ alongside the mean to avoid misleading conclusions.

You must normalize all ratings to the same scale before averaging. Convert a 5-star rating to a 10-point scale by multiplying by 2, or to percentage by multiplying by 20. This calculator uses a configurable maximum scale value. Set it to match your data, or normalize inputs before entry.

The median is robust to outliers. If 9 reviewers give 8.0 and one gives 1.0, the mean drops to 7.3 but the median remains 8.0. For small sample sizes (30 ratings), median often better represents typical user experience.

The numeric average is converted to a percentage of the configured maximum scale, then mapped to standard US academic brackets: 97 - 100% → A+, 93 - 96% → A, and so on down to F below 60%. The bracket thresholds are fixed to the US system.

Yes, especially if that outlier carries a high weight. A rating of 1.0 with weight 10 among nine ratings of 9.0 each weighted 1 yields a weighted mean of 5.26 instead of the unweighted 8.2. Always verify that weight assignments reflect genuine credibility differences, not arbitrary values.