IGN 17-6
Put down that ANOVA! Using regression-based designs to deal with spatial heterogeneity

Thursday, August 8, 2013
101E, Minneapolis Convention Center
Caitlin E. Hicks Pries, Climate Sciences, Lawrence Berkeley National Laboratory, Berkeley, CA
Margaret S. Torn, Earth and Environmental Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA
Arctic landscapes are rife with fine scale spatial heterogeneity. Within a couple meters you go from walking on crunchy lichens to sinking into soft moss as water laps your ankles. Heterogeneity reduces the statistical power of ANOVA-based experimental designs. Therefore, causes of this heterogeneity, like permafrost thaw and water table depths, must be measured and incorporated into multiple regressions that include ANOVA treatments as fixed effects. An even better approach is to change the experimental variable “continuously” instead of repeating treatments at fixed levels. This approach allows empirical data to be more easily incorporated into models.