AP Stats Unit 4 Collecting Data

Types of bias

• Undercoverage bias

• Nonresponse bias

• Response bias 

Types of samples (bad and random)

• Convenience sample

• Voluntary response sample

• Simple random sample

• Stratified random sample

• Cluster random sample

• Systematic random sample

Types of experimental design

• Completely randomized

• Randomized block design

• Matched pairs design

Elements of a well designed experiment

1. Comparison - 2 or more treatments

2. Random assignment to treatment

3. Replication - more than 1 in each treatment group

4. Control - keep other variables constant

What random sample vs random assignment to treatment allows us to do

Random sampling allows you to generalize conclusions to the population that the sample came from (if voluntary/convenience sample, can only generalize to experimental units) while random assignment allows us to conclude causation between explanatory and response variables

confounding variable 

a variable that is related to the explanatory variable and influences the response variable

low bias vs low variability

a sampling method with low bias produces estimates close to the true value (not above or below), while a sampling method with low variability produces estimates to each other

Undercoverage bias

When some people are less likely to be chosen/more likely to be missed

Nonresponse bias

When people cannot be reached or refuse to answer

Response bias

When there is an issue in the data gathering method (people feel pressured to lie, there are leading questions in the survey, etc.)

convenience sample vs voluntary response sample

sample where people who are easiest to reach are chosen vs sample where people who choose to participate are chosen

simple random sample

sample where individuals are labeled then immediately selected with random number generator

stratified random sample

the population is split into strata (groups of individuals with shared attributes) and an SRS is chosen from each strata

homogenous sample vs heterogenous sample

strata vs cluster

cluster random sample

a heterogenous group is randomly selected from the population that is meant to represent the population

systematic random sample

a starting point is randomized and the rest of the sample is chosen from an equal interval

completely randomized design

simplest form of experiment (no splitting into groups) where sample is assigned treatment randomly right off the bat

randomized block design

subjects are separated into blocks and then randomly assigned treatments within each block (within the block, there is a shared characteristic, e.g. gender)

matched pairs design

subjects are paired (block of 2) by a similarity and then each randomly assigned to a treatment (diff from the other) OR every subject receives both treatments and the order of treatments is randomized (you are your own pair)