COMMERCE 2QA3 Study Guide - Final Guide: Probability Plot, Variance, Dependent And Independent Variables
28 Jun 2018
School
Department
Course
Professor
Design of Experiments and Analysis of Variance (ANOVA)
Distinguish between an observational study and a statistical experiment.
Observational Studies
They are retrospective if they study outcomes in the present, using historical records
•
They are prospective if they identify subjects in advance, and collect data as events occur
•
These studies are conducted using pre-exiting data collected without any particular design.
Randomized Comparative Experiments
Experimenter actively and deliberately manipulates factors to specify treatment
1.
Experiment assigns subjects to treatments randomly
2.
An experiment is a study where the experimenter manipulates attributes (factors) of the study participants and
observes the consequences. Factors are manipulated by being set to particular values, levels. Levels are
assigned or allocated to individuals in the study. A combination of factor levels assigned to a subject are the
subject's treatment. Two features distinguish experiments from other investigations:
Randomize
1.
Assign treatment to subjects randomly.
Control
2.
Control sources of variation other than the control factors by making conditions as similar as possible.
Control treatments are special treatment classes that mark the baseline - the group that receives it is
the control group.
Replicate
3.
Repeated observations at each treatment are replicates. An experiment is balanced if the replicates is
the same for each treatment combination.
Replication is when you repeat entire experiments for different groups of subjects in different
circumstances or at different times.
Block
4.
Grouping or blocking subjects according to uncontrollable factors may affect response. Those factors
are blocking factors, and their levels are blocks.
Example
Sex, ethnicity, marital status.
Apply the principles of experimental design.
Completely Randomized Designs
Each of the possible treatments is assigned to at least one subject at random. In the simplest completely
randomized designs, subjects are assigned at random to two treatments.
Randomized Block Designs
When there is a blocking factor, you randomize the subjects to treatments within each block.
Design of Experiments and Analysis of Variance (ANOVA)
October 26, 2017
4:31 PM
Statistics Page 1
Document Summary
Design of experiments and analysis of variance (anova) Distinguish between an observational study and a statistical experiment. These studies are conducted using pre-exiting data collected without any particular design. They are retrospective if they study outcomes in the present, using historical records. They are prospective if they identify subjects in advance, and collect data as events occur. An experiment is a study where the experimenter manipulates attributes (factors) of the study participants and observes the consequences. Factors are manipulated by being set to particular values, levels. Levels are assigned or allocated to individuals in the study. A combination of factor levels assigned to a subject are the subject"s treatment. Experimenter actively and deliberately manipulates factors to specify treatment. Control sources of variation other than the control factors by making conditions as similar as possible. Control treatments are special treatment classes that mark the baseline - the group that receives it is the control group.
