Methods of Time Trends Analysis Applied to Contaminant Levels in Canadian Atlantic Cod (Gadus morhua)

Publisher Website
Google Scholar
Abstract
Similarities and differences between multiple linear regression (MLR) and analysis of variance (ANOVA) approaches which have been used as two seemingly different methodologies in time trend studies are explained. Difficulties encountered in choosing an appropriate MLR model are presented, and need for identifying outlying observations in a MLR analysis is emphasized. Univariate analyses such as MLR and ANOVA are of restricted scope and may even be misleading when employed on data sets containing several contaminants. Multivariate structures of these data sets have distinct advantages over a univariate model. The use of multivariate analysis of covariance (MANCOVA) for time trends investigation is illustrated with Canadian Atlantic cod (Gadus morhua) data. Significant linear time trends are shown, although these did not explain all variations between years.

This publication has 1 reference indexed in Scilit: