- Type Parameters:
DataType - Type of data measured from each subject.
- All Superinterfaces:
- Cloneable, CloneableSerializable, NullHypothesisEvaluator<Collection<? extends DataType>>, Serializable
- All Known Implementing Classes:
- AnalysisOfVarianceOneWay, FriedmanConfidence, MultipleComparisonExperiment
title="Randomized Block Design",
public interface BlockExperimentComparison<DataType>
- extends NullHypothesisEvaluator<Collection<? extends DataType>>
Implements a null-hypothesis multiple-comparison test from a block-design
experiment. In these experiments, there are some number of "subjects"
assigned to each "treatment". Typically, the null-hypothesis of a
block-design experiment is that the treatments have no effect on the
measurements of the subject.
For example, one treatment may contain 10 subjects who are given nothing
but water. Another treatment may contain 8 subjects who are given pure
corn syrup. A third treatment may contain 12 subjects given liquefied lard.
The experimenter then measures the change in the subjects' weight from the
beginning to the end of the experiment. The null-hypothesis may be that
the treatments (water, corn syrup, lard) has no effect on the weight gain
of the subjects.
Please note that block-experiment comparisons, such as ANOVA and Friedman's
test, typically DO NOT indicate which treatment is statistically
significantly different from the others, just that a difference exists.
You must run a MultipleHypothesisComparison test to determine which
treatments are different from the others.
- Kevin R. Dixon
ConfidenceStatistic evaluateNullHypothesis(Collection<? extends Collection<? extends DataType>> treatments)
- Evaluates the null hypothesis for the given block-design treatments
treatments - Collection of treatments for the block-design experiment, where each
- The confidence for the null hypothesis.