Unit 3 - Collecting Data Flashcards
Samples may not always represent a ———
population
population
all the individuals we want to study
sample
subset of a population used to infer about the population
When are samples used?
Population is too large to study
Other limitations
A sample can be generalized to its respective population if …
1- randomly selected
2- representative
3- selected specifically from that population
confounding factors
Any number of factors that makes it challenging to
observational study
survey without imposing treatments
can’t conclude cause & effect
retrospective OS “looking back”
examining past data
prospective OS “looking future”
examining data as it progresses
experiment
different treatments imposed on different groups
well-designed experiment- can conclude cause & effect
census
assess all individuals in a population
best way
random sample
easier way
if perform correctly, can be representative of population
simple random sample
every group groups have an equal chance to be selected
Assigning numbers to individuals or groups etc. and using a random number generator to generate random numbers and subsequently, random individuals
cluster random sample
1- population is divided into CLUSTERS
2- SRS taken of each clusters
stratified random sample
1- population divided into STRATA based on SIMILAR characteristics
2- SRS taken of each strata
Cluster vs Stratified
Cluster- HETEROGENEOUS grouping
Strata- HOMOGENEOUS grouping
Systematic random sample
1- RANDOM starting point
2- sample taken at fixed PERIODIC interval
Biased
measure of accuracy
Variation
measure of precision
Non-random sample are ….
biased
simple random sample advantages
unbiased/ HIGH accuracy
easy to explain
easy to perform
SRS disadvantages
logistical issues
MODERATE variability
Cluster random sample advantages
unbiased/ HIGH accuracy
easy to perform
Cluster random sample disadvantages
homogeneous clusters result in HIGH variability
stratified random sample advantages
unbiased/ HIGH accuracy
LOW variability
stratified random sample disadvantages
hard to implement
bias
systematic tendency to FAVOR CERTAIN RESPONSES over others
undercoverage
part of population has REDUCED CHANCE of being included
nonresponse
individual chosen for sample DON’T respond
convenience samples
bbabs
