Sorensen-Dice Coefficient
Also known as: Dice coefficient, Czekanowski index, F1 similarity
Sorensen-Dice coefficient, also called Dice coefficient or Czekanowski index, measures the similarity between two sets or samples based on presence and absence of attributes. Introduced independently by Thorvald Sorensen (1948) and Lee Dice (1945), this index ranges from 0 (completely dissimilar) to 1 (identical). It is particularly well-suited for binary presence-absence data and is the symmetric counterpart to the Bray-Curtis dissimilarity for abundance data.
Key highlights
- Intuitive for presence-absence data; easy to interpret in ecological or linguistic contexts
- Symmetric: S(A,B) = S(B,A), treating both sets equally
- Ranges from 0 to 1; values are easy to compare and communicate
- Works well with sets of different sizes and with rare attributes
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Sorensen-Dice is ideal for binary presence-absence data and for abundance data where relative proportions matter more than absolute counts. Use it in ecology for species composition comparison, in text analysis for document similarity, and in any domain comparing sets of discrete attributes. It is more intuitive than Jaccard for data where sets of different sizes are common, and less sensitive to rare attributes than Bray-Curtis.
Strengths & limitations
- Intuitive for presence-absence data; easy to interpret in ecological or linguistic contexts
- Symmetric: S(A,B) = S(B,A), treating both sets equally
- Ranges from 0 to 1; values are easy to compare and communicate
- Works well with sets of different sizes and with rare attributes
- Not a true metric; does not satisfy the triangle inequality
- Less suitable for continuous data; requires discretization or thresholding
- Sensitive to the choice of presence-absence threshold; small changes can affect results
- Does not account for attribute importance or relationships between attributes
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
How does Sorensen-Dice differ from Jaccard similarity?
Both measure set similarity, but Sorensen-Dice counts shared attributes twice in the denominator (2*intersection / sum of sizes), while Jaccard counts them once (intersection / union). For identical sets, both equal 1. For different sets, Sorensen-Dice produces larger values than Jaccard, making it less strict.
Can Sorensen-Dice be used with weighted or abundance data?
Yes. Weighted Sorensen-Dice exists for abundance data: it sums the minimum abundances as the intersection and the sum of both abundances as the union, similar to Bray-Curtis. Use this variant when abundance matters, not just presence-absence.
What is the relationship between Sorensen-Dice and F1 score?
Sorensen-Dice is equivalent to the F1 score (harmonic mean of precision and recall) when applied to classification: if a set represents predicted labels and another represents true labels, Sorensen-Dice equals the F1 score. This connection bridges ecological and machine learning terminology.
When should I use Sorensen-Dice over Bray-Curtis?
Use Sorensen-Dice for binary presence-absence data or when you want a symmetric, easy-to-interpret measure. Use Bray-Curtis when comparing abundance data (counts, proportions) where magnitude differences matter, not just set overlap.
Sources
- 1.Sorensen, T. (1948). A method of establishing groups of equal amplitude in plant sociology based on similarity of species content and its application to analyses of the vegetation on Danish commons. Biologiske Skrifter, 5, 1-34.
- 2.Dice, L. R. (1945). Measures of the amount of ecologic association between species. Ecology, 26(3), 297-302.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 3). Sorensen-Dice Coefficient. ScholarGate. https://scholargate.app/decision-making/sorensen-dice-coefficient