Sampling and Surveys Vocab Flashcards
population
all of the individuals from whom we want data
census
collects data from every individual in the population
sample
a subset of individuals in the population from where we collect data
sample survey
stidy collecting data from a sample of the population.
1) decide population
2) find what want to measure
3) choose how pick sample
convenience sampling
selects individuals who are easy to reach
Issues with convenience sampling
1) undercoverage
2) response bias (who is asking influence)
3) nonresponse bias
bias
a method that systematically favors certain outcomes over another
Voluntary response sampling
participants voluntarily chose to join the sample
Issues with voluntary response sampling
1) voluntary response bias (could have strong opinions)
2) undercoverage
3) non response bias
Random Sampling
Uses chance to determine who is included in the data set
Simple Random Sampling
Every group of n individuals has an equal chance of being selected (ideal)
Issues with SRS
1) non response bias
2) most ideal
3) hard to conduct
How to phrase SRS
Randomly assign [subject] numbers [1-3000]. Use a random number generator to select [100] people, ignoring repeats.
menu-probability-randseed; menu-prob-randint(lower, upper, trials)
Sampling without replacement
Once an individual is selected, they can not be selected again
strata
groups of people who share characteristics thought to be associated with tested variable
Stratified Random Sampling
Divide population into strata and combine subsamples
Issues with Stratified Random Sampling
1) non response
2) very hard to conduct/time consuming
How to phrase Stratified Random Sample
Stratify by [group]. Randomly assign [subject] numbers [1-3000]. Use a random number generator to select numbers [25] from each [group].
Groups of blubber fish 20-move down 20 on 0.25-0.5cm and get fish #37.
cluster
group of individuals in a population that are near each other
cluster sampling
Select sample by randomly choosing clusters
How to phrase cluster
Randomly select [5] [group] and survey every [subject] in those [groups]
Issues with cluster sampling
1) undercoverage
2) nonresponse
Systematic Random Sampling
chose every [10th] seat in a group of [800] for a sample size of [80].
How to phrase Systematic Random sampling
Randomly assign [subject] numbers [1-3000]. Use a random number generator to select numbers [1-10]. Survey every [30th] [subject].
Issues with Systematic Random sampling
1) non response
2) time consuming
undercoverage
part of the population is not accounted for in the sampling
non response
individual refuses to participate or can not be contacted
response bias
systematic pattern of inaccurate responses (based on interviewer, or voluntary response)
observational study
observes individuals and measures variables without influencing results. Only confirms correlation, not causation. Ex: smoking, volcano, natural
response variable
measures outcome of a study
experimental study
Experimenter controls variables. Can show causation.
explanatory variable
helps explain or predict changes in response variable (x)
confounding
variable influences independent and dependent variables (extra variable other than what testing for).
experiment
imposes treatments on individuals and compares with a control group to measure impact.
placebo
treatment without any active ingredient (control)
treatment
specific condition applied to individuals in a experiment
experimental unit
object treatment is randomly assigned
subject
word for when experimental unit is a human being
factor
An explanatory variable that is manipulated and may lead to a change in response variable
level
the degree to which the explanatory variable has changed (0.5ga, 1ga, 2ga)
control group
baseline to compare effects of other treatments. Could be a placebo, active treatment, or no treatment
placebo effect
belief that react to treatment, even when received inactive treatment (unknowingly).
double-blind
neither subject nor administrator know what the variables are/ if it is a placebo
single blind
either the participant or the administrator knows the variables/ if it is a placebo
Random assignment
experimental units are assigned treatments by chance
control
keep all other variables constant in experiment
replication
Ensuring a large sample size so that differences between treatment groups can be observed
completely randomized design
sizes of groups, treatment received, and response variables completely random
block
group of participants that are known to be similar (found after decide sample)
randomized block design
random assignment of treatments to participants is decided within each block
matched pairs design (2 students for each block)
Subjects similar characteristics selected into pairs (similar SAT score both take math test dif locations and compare)
matched pairs design (student serves as own block)
Subjects similar characteristics selected into pairs (each student takes test twice, one in each location)
sampling variability
random samples produce different estimates (larger= more accurate)
statistically significant
When results cant be explained by chance alone. Only study that can draw conclusions from.
scope of inference
1) random selection individ allow inference about population
2) random assignment of individuals to groups allows inference cause and effect
