Unit 3, Ch 4 Flashcards
Surveys and Experiments, Designs, blablabla
Identify confounding variable
- How it’s associated with expl var
- How it’s associated with response var
Level
Different value of a factor
Factors and Levels
Factors: category
Levels: value for each category
Ex. Factor: fertilizer, water
Ex. Levels: For fertilizer -> 0 lbs, 20 lbs
For water -> 0 cups, 5 cups
4 Principles
R,R,C,C
RaReComCon???
Randomize - select randomly, least bias
Replicate - use multiple subjects for consistent results
Control - keep variables constant
Comparison - compare 2+ treatments
Blinding (2 types)
Single: subjects don’t know what treatment they have
Double: testers and subjects don’t know what treatment they have
Random Assignment (3 steps)
Label - label the subjects you’re randomizing
Randomize - RNG, hat draw, etc
Assign - put subjects in corresponding levels
Placebo effect
A false treatment (subjects may think it’s giving the same effect and can cause psychological changes)
Flowchart
- Subjects
- Randomization
- Group 1, group 2, etc…
- Treatment 1, treatment 2, etc…
- Compare
- Underneath, write how subjects will be assigned (LRA)
Block design
Groups of similar subjects
- not necessarily a factor being studied. Splits factors so different effects can be considered
Stratified vs Block
Stratified: survey
Block: experiment
Matched Pairs Design
Type of block. Pairs of similar units
Ex.
1st subj -> trtmnt 1 -> trtmnt 2
2nd subj -> trtmnt 2 -> trtmnt 1
Sampling
SRS
Every group of given individuals has the same chance of being chosen
Ex. 2500 people, want to choose 250.
Choosing 200 males and 50 females is NOT SRS!!! It’s not truly random (ex. getting 250 males and no females, or 100 males and 100 females, etc)
Sampling
Systematic
Picking by intervals
Ex. Choosing every 5 students to be part of experiment
Sampling
Cluster
Split into groups -> choose group -> sample everyone from each group
ALL from SOME
Sampling
Stratified
Split into groups -> SRS from each group
SOME from ALL
- Can be trait based
BIASED Sampling
Voluntary Response
sample involves indvls who choose to participate
BIASED Sampling
Convenience Sampling
sample involves indvls who are easy to reach
Undercoverage Bias
part of population less likely to/can’t be chosen
- may not represent true variety of population
(Sampling process)
Response Bias
inaccurate response due to wording/connotation or external influences
- forces responders to pick specific side
Ex. Guns kill millions of animals a year. Do you support the 2nd amendment?
Another ex. a teacher asks if students like his/her class. Students may feel inclined to respond positively (not honest answer)
Non-response bias
part of population doesn’t respond/can’t be reached
(Response process)
Observational Study
observes + measures variables of interest
Experiment
imposes treatment to measure response
Treatment
conditions imposed on subject
Experimental unit
what’s being tested on
- if it’s a human, it’s called a subject
Is random sample generalizable?
yes
What does random assignment ensure?
Treatment actually causes change in response var (varied subjects = more uniform result)