MIDTERM 1 Flashcards
WHAT ARE THE 4 METHODS FOR COLLECTING DATA?
- DESIGN EXPERIMENT
- OBSERVATIONAL STUDY
- SAMPLING SURVEY
- CLINICAL TRIALS
What is the process of doing an experimental design?
- specify the treatment, response, experimental unit
- assign treatments to the unit
- observe the response, the causal effect
- need techniques to control the uncontrollable factors
What is an observational study?
- treatments are simply observed as belonging to the unit
- may not be causal
- causal effect may not be clear, may need specific methodology to find the causal effect
- retrospective and prospective
What is a sampling survey?
- a physical population of units
- we select a sample from the population, collect data from the sample
- sample is assumed to be representative of the pop.
- estimate the population features using the sample
- we can detect association, but not a causal effect
What are clinical trials?
- similar to a design experiment
- for humans, many treatments allocations are not feasible
- specific designs
What are the objectives of a statistical experiment?
- determine which factors are most influential on the response
- determine where to set the influential factors so that: 1. the outcome is close to the designs nominal values ? and 2. the effects of the uncontrollable factors are minimized
What are the 3 key elements of a stats experiment? describe them
- Experimental unit - the unit that you can apply the treatment to
- Treatment options - what options we apply to the units
- Response measurement - what is the outcome of interest?
Review: Hotelling’s elegant design
ok
When the problem involves data that are subject to experimental errors ….
we must use statistical methodology for the analysis
Design: How we collect the data and Analysis: …
need to ensure that all assumptions are valid and that we collect data in a manner that ensures analysis assumptions are valid
What are the 3 basic principles for stats exp. design ?
- randomization 2. repitition 3. blocking
What is randomization?
when we allocate the treatments in a completely randomized way and the order in which experiments are performed is random
When we use randomization, what can we do? 4
- make assumptions about independence of obs
- make each possible assignment equally likely
- avoid systematic error/bias
- aveerage out the effects of extraneous factors
What is replication?
- Repitition of the basic experiment - change the unit each time
Why is replication important? 2
- allows us to estimate the experimental error
- gives us a more precise estimate of the effect of interest
replication > repeated measurements (estimate the measurement error)
what is blocking?
a technique that is used to improve the precision with which comparisons of the factors of interest are made
what is the goal of blocking?
create a set of relatively homogeneous experimental design conditions so the experimental units are are homogeneous as possible
why is blocking used? what does it reduce?
the variability from nuisance factors
what are nuisance factors?
they are factors that have effects on the response
