Experimental Design Flashcards
Stratified sample
Group homogeneous groups then take a random sample from each group
Simple random sample
Randomly pick from everyone (need a list of everyone)
Cluster sample
Group by location then randomly select some locations and use everyone in that location
Systematic sample
Pick a random place to start and sample every 4th person after that
(Don’t need sampling frame)
Completely randomized
Randomly assign treatments to experimental units
Blocked design
Group experimental units into homogeneous groups them randomly treatments in each group
Matched pairs
Group two like experimental units together an have one do one treatment and one do the other
Have one do two treatments compare the results
Difference between experimental design and observational study
Experimental design you do something to the units and observation you don’t
Increasing sampling size
Decreases variation
Randomization reduces
Bias, but it does not eliminate
Stratification reduces
Variation
What is a confounding variable
Something that will effect results of experiment
Booking is a type of experimental design, it happens
Before you assign treatments
Having more then one treatment group does not mean the experiment is
Blocked
The principles of design
Control randomization replication
Placebo
Fake treatment (does nothing
Blinding is when
Experimental unit doesn’t know
Double blind is when
Experimenter and experiment unit do not now
Well designed experiment only shows
Cause and effect
Observational study and correlation does not
Show cause and effect
In simple random sample
Every sample has same chance of being poked
Types of bias
Voluntary response Non response Under Coverage Convenience sampling Response bias
What is an experimental unit
The single individual to which the treatments are assigned
What is a factor
It is the explanatory variable
What is a level
It is a specific value for the factor
What is the response variable
It is what you measure
Statistical advantage of stratified random over simple random is
It reduces variation
The use of an extraneous variable to control the placebo effect is (important or not important in the design of the experiment)
Not important