STATS 252- statistical studies & sampling methods

STUDY
PLAY
observational unit
who or what is being studied
variable
characteristic of the observational unit
what are the two types of variables
quantitative and categorical
what are the two types of categorical variables
ordinal and nominal
quantitative variable
numerical value with units (ex: how many..?)
categorical ordinal
when the values of a categorical variable have an intrinsic (number) order (ex: class year)
categorical nominal
categorical variable with no order (ex: eye color)
population
the entire group of subject we want to learn about
sample
the smaller part of the population that you gather data from
response variable
dependent variabel; outcome or result of study
explanatory variable
independent variable; variable that we think explains the changes
confounding variable
outside variable that is correlated to both response and explanatory variable (ice cream and murders example)
correlation does not mean...
causation
parameter
characteristic used to describe a whole population
statistic
characteristic used to describe the sample
response bias
"leading questions" -worded to force a certain type of response; dishonest response to maintain image
volunteer response
those who respond chose to do so and tend to have stronger opinions
undercoverage
population is not evenly represented; created by sampling who ever is "convenient"
non-response bias
not everyone will respond to survey request
simple random sample (SRS)
each person in sample is assigned a number, numbers randomly generated to create sample size, then surveyed; results will vary with different samplings
stratified sampling
divide population into similar groups (strata) and then take a random sample from each group
cluster sampling
divide population into clusters (usually geographically based) then take SRS in each cluster
systematic sampling
create interval "K", select number between 1-K to start, pick every "K"th person
response bias
wording the question in a way that inflicts a certain response