sampling methods
A subset of the population (N) chosen in a way that every sample size (n) has the same chance of being chosen
SRS
difference between sample values and population values, just because you are not measuring all subjects in the population
sampling error
Expected, attempt to minimize error by using random sampling techniques
two main types of variables
qualitative and quantitative
what does it mean?
indicates the strength of the linear relationship
DOUBLE
1. Either the subject or the experimenter does not know whether the subject received the treatment or control
2. Both the subjects and the experimenter do not know if they subject received the treatment or control
1. single blind experiment
2. double blind experiment
stratified sampling
divide the population into strata, and randomly select subjects from each strata
something was done wrong
non-sampling error
explanatory variable: ?
response variable: ?
explanatory = x
response = y
what is the range
-1 to 1
An experiment in which subjects are randomly assigned to either a treatment group or a control group
Ex. Drug treatments (placebos for control)
Completely randomized experiments (aka randomized comparative experiment)
cluster sampling
Divide the population into clusters/groups that make sense for the study. Choose an SRS of full clusters and gather data from every subject within the chosen clusters
sample subjects provide inaccurate information
response bias
qualitative variable for which the raw data are group or category names, that don’t necessarily have a logical ordering
ex. eye color, country of residence
categorical variable
strong correlation values
0.8 to 1; -0.8 to -1
examples of observational studies
observation (duh)
surveys (strongly disagree, disagree, neutral, agree, strongly agree)
arranging the sample in an order that makes sense, and choosing every k^th subject
systematic sampling
sample subjects cannot be contacted or choose not to respond/participate
non-response bias
qualitative varaible that has a logical order/ranking
ex. highest educational degree earned, t-shirt size
ordinal variable
weak correlation values
0 to (not including) 0.5; 0 to (not including) -0.5
randomized block design
1. Block the groups by a common characteristic that could affect response to treatment
2. Within each block, randomly assign subject to treatment or control group
3. Then compare responses within blocks
RANDOM DOUBLE
what is a lurking variable? what is a confounding variable?
lurking: variables you are not interested in, but could affect the response to treatment (hidden, sneaky). They create the impression that the variables you are studying are related when they may not be
confounding: When you can’t tell which variable (or a combination of the two) had an effect on the outcome (ex. lawn example)
bonus: if someone feels strongly about the topic of interest, they are more likely to respond than those who are uninterested
volunteer or self-selected sample
discrete can only be in whole numbers (Ex. number of siblings) while continuous can be any number (ex. weight)
moderate correlation values
0.5 to (not including) 0.8; -0.5 to (not including) -0.8
darwin, cooper, quinn, sybil
honorable mention to dixie and pickles