Biology 2244A/B Lecture Notes - Lecture 12: Type I And Type Ii Errors, Ernst & Young, Confidence Interval
28 Feb 2017
School
Department
Course
Professor
Document Summary
Completely randomized design: no real sampling going on they constructed the units themselves. Three treatments diff levels of defoliation. 1 explanatory variable 1 factor anova: balanced design equal number of replicates don(cid:495)t require this to run anova. No srs bc there was an experimental community but there is randomization: the treatment is a level or levels of explanatory variables, can have combination. Y-bar = mean species richness (mean of the variable we are looking at) Don(cid:495)t want matched pairs in each sample: srss. Each one has to be an srs. When you are working with experimental units you can get away with using a randomized design instead of srs only in anova: pops are normal. Cant use 50+ the clt because we are not dealing with sample distribution of sample means: constant variance. Each of the normal dists has the same variance (sd) Requires 2 degrees of freedom that define you.
