Experimental Design CONTINUED

continuous

variables measured on a continuous numerical scale (e.g. heights, habitat coverage, species richness)

categorical

variables with one or more unique categories (e.g. sex (male/female), trophic level (producer, herbivore, carnivore), etc.)

dichotomous

type of categorical variable where there are only two possible categories or levels (e.g. sex (male/female))

single regression

for each replicate, measure both the predictor (independent) and response (dependent) variables

treatments

in ANOVA, different categories of predictor (independent) variables

replicates

in ANOVA, within each treatment multiple observations are made

single-factor design

ANOVA design where each treatment represents variation in a single predictor (independent) variable or factor

multi-factor design

ANOVA design where treatments cover two or more different factors (independent variables)

main effects

in two-way ANOVA, additive effect of factor 1 would be measured as the response of the treatment group at each factor 1 level averaged over the responses to factor 2 levels (which is more important)

additive effect

two factors/treatments combine so that the total combined outcome equals the simple sum of their individual effect acting alone

interaction effects

high factor 1 levels and high factor 2 levels might have a synergistic effect (combined result greater than the sum) you might not see in a single-factor experiment

block

an area or time period over which conditions are relatively homogenous. conditions within them should be more similar than between them

nested

ANOVA design that has sub-sampling within the replicates. number of replicates doesn’t change, but can take multiple measurements across each one

factorial

layout design of a two-way ANOVA. two or more factors are tested simultaneously in one experiment. treatments are fully crossed or orthogonal

split-plot

ANOVA design where treatment is applied to an entire block. similar to a random block, but has a second treatment added

circularity

both the randomized block and repeated measures designs have this assumption for the within-subjects factor

between-subjects factor

corresponds to the whole plot

within-subjects factor

corresponds to different times

randomized block

variance of the difference between any pair treatments in the block is the same

repeated measure

variance of the difference of observations between any pair of times is the same. multiple observations on the same replicate are collected at different times

BACI (before after control impact)

NOT an ANOVA design. measurements are taken before and after a treatment (having an event and looking at the control before the event and the impact after the event)