On-line statistics

Nearest Neighbour Sorting

Working out Søerensen's Quotient or Peterson's Index (or any other similarity value) tells you how alike two samples are (approaching 0 = dissimilar, approaching 1 = nearly identical). However, when you have sampled more than two sites and wish to investigate similarities and differences between them, how can you present your data to show the whole picture? When only a few sites have been sampled a simple table showing the similarity value for all possible pairwise combinations may be sufficient. For example for six samples:

When many samples have been taken the best way to present the data is as a 'cladogram'. Nearest neighbour sorting is a crude but quick method of clustering sites by similarity. The highest value in the table (0.84) shows that the two most similar sites are 1 and 2. The next highest (0.75) shows that sites 3 and 4 are similar. The next highest value is (0.74) and indicates a similarity between 6 and 1; site 6 can therefore be clustered with sites 1 and 2. The results can be presented as:

---

Descriptive Stats

Diversity Indices

Comparisons

Correlations

Regression