Assortative mating is defined as non-random within genotype or phenotype mating. Such non-random patterns of mating are of significance as they lead to population divergence and promote speciation. Due to its significance to population genetics as well as ecology it has been studied in a wide range of species for a traits such as age, behavior, phenology, genotype, ecotype, visual (color) differences among others. While most assortative mating is considered to be positive, few cases of negative assortative mating or diss-assortative mating have been recorded.
While ecologists, have no trouble defining assortative mating around phenotypes, geneticists seem to be very picky. Even the third edition of the book "Evolution" by Ridley defines assortative mating as "Tendency of like to mate with like. It can be for a certain genotype (e.g., individuals with genotype AA tend to mate with other individuals of genotype AA) or phenotype (e.g., tall individuals mate with other tall individuals)."
On the other hand in the book "An Introduction to Population Genetics" by Rasmus Nielsen and Montgomery Slatkin define assortative mating as "A mating structure in which pairs of individuals that are (genetically) similar to each other mate with higher probability than expected under random mating."
Is the "genetic" basis of the non-randomness implied or even necessary? Does the lack of a genetic basis make the phenomenon less interesting? The recent review/meta-analysis by Kirkpatrick's group seems to suggest that these and many more questions could help explain trends in the process of species formation.