Sampling and Experiments Flashcards
sample
a representative subset of a population
population
the entire group of individuals which about whom we intend to learn
bias
a SYSTEMATIC failure of a sample to represent a population
voluntary response bias
bias from individuals who choose to not be in the sample
nonresponse bias
bias from individuals in the sample who fail to respond
convenience sample
a sample of individuals who are easy to survey; this sample typically is biased
undercoverage
this results from any sampling method that under-represents a segment of a population
response bias
anything in a survey that causes subjects to respond differently from the truth
systematic random sample
Example: randomly survey every 12th person in line at a movie theater
cluster sample
a random sample of several diverse groups (and each group is representative of the population)
stratified sample
a random sample taken from each of several homogenous (alike in some way) groups
sampling frame
the “list” from which a random sample may be drawn (it may not be a real list)
sampling variability
the variability that naturally occurs from sample to sample
simple random sample (SRS)
a random sampling method where each subset of the population has an equal chance of being selected
statistic
a calculation from a sample
parameter
a numerical characteristic from a population
census
a survey that contains the entire population
strata
a group of individuals that are alike in some important way
cluster
a group of individuals that are very diverse, and who represent the population well
multistage sanple
a sampling method that combines several sampling methods to obtain a sample from a population
observational study
a study in which no treatments are assigned to subjects; they are simple observed
retrospective study
a study in which data from the past is collected on subjects for analysis
prospective study
a study in which subjects are followed and data is collected in the future
experiment
a study in which treatments are assigned to subjects and responses are analyzed
factor
a variable whose levels are set by an experimenter (to see if there is an effect on the response variable of interest)
level
the value that an experimenter chooses for a factor
experimental units (subjects)
the individuals on which an experiment is performed
treatment
the different levels of a factor
three principles of good experimental design
control, randomization, replication
control group
a treatment group that gets no treatment (or a placebo)
blinding
when an individual does not know which treatment is given to a subject
double blind
when neither the subjects nor the evaluators know which treatment is given to subjects
placebo
a treatment that is know to have no effect
blocking
a design where an experiment is conducted in each of several groups that are alike in some important way (Example: groupings by gender, by age, by race, etc.)
confounding
this happens when the levels of one variable are hopelessly mixed together with the levels of another variable (making it impossible to determine which levels are truly affecting the response variable of interest)
a representative subset of a population
sample
the entire group of individuals which about whom we intend to learn
population
a SYSTEMATIC failure of a sample to represent a population
bias
bias from individuals who choose to not be in the sample
voluntary response bias
bias from individuals in the sample who fail to respond
nonresponse bias
a sample of individuals who are easy to survey; this sample typically is biased
convenience sample
this results from any sampling method that under-represents a segment of a population
undercoverage
anything in a survey that causes subjects to respond differently from the truth
response bias
Example: randomly survey every 12th person in line at a movie theater
systematic random sample
a random sample of several diverse groups (and each group is representative of the population)
cluster sample
a random sample taken from each of several homogenous (alike in some way) groups
stratified sample
the “list” from which a random sample may be drawn (it may not be a real list)
sampling frame
the variability that naturally occurs from sample to sample
sampling variability
a random sampling method where each subset of the population has an equal chance of being selected
simple random sample (SRS)
a calculation from a sample
statistic
a numerical characteristic from a population
parameter
a survey that contains the entire population
census
a group of individuals that are alike in some important way
strata
a group of individuals that are very diverse, and who represent the population well
cluster
a sampling method that combines several sampling methods to obtain a sample from a population
multistage sanple
a study in which no treatments are assigned to subjects; they are simple observed
observational study
a study in which data from the past is collected on subjects for analysis
retrospective study
a study in which subjects are followed and data is collected in the future
prospective study
a study in which treatments are assigned to subjects and responses are analyzed
experiment
a variable whose levels are set by an experimenter (to see if there is an effect on the response variable of interest)
factor
the value that an experimenter chooses for a factor
level
the individuals on which an experiment is performed
experimental units (subjects)
the different levels of a factor
treatment
control, randomization, replication
three principles of good experimental design
a treatment group that gets no treatment (or a placebo)
control group
when an individual does not know which treatment is given to a subject
blinding
when neither the subjects nor the evaluators know which treatment is given to subjects
double blind
a treatment that is know to have no effect
placebo
a design where an experiment is conducted in each of several groups that are alike in some important way (Example: groupings by gender, by age, by race, etc.)
blocking
this happens when the levels of one variable are hopelessly mixed together with the levels of another variable (making it impossible to determine which levels are truly affecting the response variable of interest)
confounding
