module 2 Flashcards
what is the point of sampling design
- to have good rep of a stat population
bias
an over- or under-estimate of some value from an average sample compared to the statistical population
4 goals of an ideal sampling design
- all sampling units are selectable/can be included
- selection of sampling units are unbiased
- selection must be independent, not influence other selection
- all samples are possible (all samples from sampling units are options)
what is an easy approach to avoid bias?
- assign all sampling units a number and find your sample by randomly generating a set of numbers
Sampling independence
- selection of one sampling unit doesnt influence the probability that other sampling units are selected
observational study def, goals, limitation
- researchers have no influence
- characterize smth abt an existing stat population
- cannot make statement of causes, only correlation/association
why cant observational studies give the cause?
- there is no way of knowing if there are other variables that weren’t measured that affect the outcome
- confounding variables
response variable
- the response you are interested in
explanatory variable
- the factor you investigate
confounding variables
- unobserved variables that affect the response variable
When the relationship between an explanatory variable and response variable is thought to be driven mostly by a confounding variable, the relationship is called ________.
- spurious
5 observational study designs
- simple random survey
- stratified survey
- cluster survey
- case-control survey (outcome is known, other differences are observed)
- cohort survey
simple random survey
- sampling unit randomly selected from stat population
stratified survey
- taking potential influence into account
-break stat population into strata and sample each strata - sampling units are selected from within predefined groups
cluster survey
- removes heterogeneity in stat population
- create groups (clusters) where non-relevant heterogeneity is contained
- the cluster is the sampling unit and the observation unit is nesting inside the cluster
- one stage=data collected from all observation units in a cluster
- two stage= randomly selected within a cluster
case-control survey
- outcome is known, other differences are observed
- 1st group= case, contains sampling units with particular responsive variable
- ## 2nd group= control, sampling unit but no responsive variable, biased
cohort survey
- follow sampling units overtime
- simple random survey but over time
strata
- subgroup within stat population
retrospective vs prospective studies
- retro=outcome is known ex case control
- pro=outcome is not known ex cohort
cross-sectional vs longitudinal studies
- cross-sectional: one point in time only
- longitudinal: many points in time
experimental study def and goal
- involving stat population, investigator controls variable(s)
- goal: study the effect of one (or more) manipulated variables on one (or more) response variables
manipulated variables are aka ____
factors, one level doesn’t have it the second does
t or f: the only difference between experimental and observation studies is that the explanatory variable is manipulated by the researcher
f: that sampling units are randomly assigned to each level in each factor for experimental
________ is the cornerstone of experimental studies
Replication
Replication
- the idea that treatment (manipulation) will be repeated many times to see how repeatable a measured outcome is
- the number of sampling units
error in experimental studies
- pseudoreplication
- observation units are analyzed rather than the sampling units
common design elements of experimental studies
- control treatment (reference treatment to compare against the treatment levels that alter the explanatory variable)
- blocking (control for variation among the sampling units that is not of interest to the researcher)
- placebo
- blinded (sampling unit doesn’t know which treatment they are exposed to
sham treatment