Viewpoint: trend assessment by similarity--a demonstration.

Abstract

Methodology for assessing trend in range condition is still evolving. This paper demonstrates use of Dice's community similarity coefficient, 2a/(2a + b + c), with communities present at 3 times and a notional community as a goal. Coefficients range from 0 (indicating a complete lack of similarity) to 1 (indicating complete similarity). Similarity is classed as low (0 - 0.25), moderate (0.26 - 0.50), high (0.51 - 0.75), or full (0.76 - 1). Study of time-goal coefficent graphs is suggested for deciding whether trend is up, down, or static. Defining goals and lack of statistical tests are major limitations. The goal concept and use of data standardization are discussed.