Sign in to save your progress, vote, and build your own decks.Sign in
Survey Methods
Survey Epi Final
70 cards·by asolivan
epidemiologysurvey methods
Systematic departure is due to?
low validity
Low reliability is related to
random variation
what is the difference between a reliable and a valid survey
a valid survey is accurate, a reliable survey is consistent
reliability is not subject to measurement error
false
what are three different ways to test reliability
test-retest, alternate form, internal consistency
what are other forms of reliability
inter and intraobserver reliability
test-retest reliability has respondents complete a survey at two different time points
true, this examines stability of responses
what is intraobserver reliability
the tendency of a single observer to get different results using the same test on the same
person
how do you use alternate form reliability
use differently worded forms to measure the same attribute
problems with test-retest
individuals become familiar with the items and answer according to memory - inflating
reliability estimates
problems with alternate-form reliability
the two items may not address the same aspect or behavior
what is the split halves method
adminster one item type to each half and compare the two halves
what is internal consistency reliability applied to
not to a single item, but groups of items that are thought to aspect different aspects of the
same concept
strategies if internal consistency is low
add more items or re-examine exisiting items for clarity
what is interobserver reliability
how well two evaluators agree in their assessment - use correlation coefficient
what is the Kappa statistic used for
evaluates how much better then random chance the agreement between two tests/observers
what are the four forms of validity
face, content, criterion, construct
what is face validity
cursory review to untrained individuals
what is content validity
subjective measure of how appropriate the items seem to a set of reviewers with some knowledge
of subject matter
what is concurrent criterion validity
assess instrument against a gold standard (like a lab test)
what is predictive criterion validity
asses the ability of instrument to forecast future events, behavior, attitudes, or outcomes
(SAT score vs performance in college)
what is convergent construct validity
implies that several different methods for obtaining the same information about a given
trait or concept produce similar results
what is divergent construct validity
the ability of a measure to estimate the underlying truth in a given area, but not correlated
closely with a specific trait
When are Focus Groups useful
for qualitative research
what is the goal of cognitive laboratory interviews
get information about how the respondent undertstood the questions and how they decided to
answer them
what type of techniques are cognitive lab interviews
think-aloud, detailed probing, go through Q twice once with discussion
what is the goal of pre-testing
find out how well the data collection protocols and survey work in real conditions
Quantitative method for pretesting
have respondents and interviewers grade the question etc.
field pre-testing
data collection similar to survey, particiapants should be similar to intended population
what is pilot testing
walk-through of entire study, should only differ to study in scale
what type of pupulation do you use for pilot test
represenative of target sample selected how you would in actual study
what is a unipolar rating scale
response alternatives from poor to excellent
what is a bipolar response alternative
range from strongly disagree to strongly agree
for very complex emotional issues you may want to have 2 middle points
true
you should always have balanced scales
true unbalanced lead to bias
what does univariate analysis examine
distribution of cases on only one variable at a time
what is the aim of univariate analysis
describe the populate
what does bivariate analysis examine
distribution of cases on one dependent and one independent scale
what is the aim of bivariate analysis
comparison and explanation
what does multivariable analysis examine
the distribution of cases on one dependent and more than one independnet variable
what is the aim of multivariate analysis
explanation and predictive models
What does multiple imputation methods do
multiple sets of randomly generated missing data
what are maximum liklihood methods
account for a variable in the covariate distribution
what is weighted estimating equation methods
use weights for regression analysis obtained through statistical modeling of missing data
what is the ecological fallacy
results observed at a group level might not be relevant at the individual level
what is the sampling frame
the actual list of sampling units from which the sample is selected
what is the observation unit
the exact element from which data is collected-not always the unit of analysis
what type of error is sampling error
random
what is probability sampling
every unit has a known non-zero probability of selection
a sample frame can be patients who attend a clinic
yes, as long as it is documentable
what is simple random sampling
each element is selected from the sample frame independently, randomly, and without
replacement
what is systematic sampling
uses a random start and then uses a sampling ratio
what are the issues with systematic sampling
periodicity, and there is a known probability and no research discretion
what is the sampling error of a stratified sample
there is zero sampling error for the variable by which stratification occurred
what are the benefits of stratified sampling
reduces variability, reduces overall sampling error, increases efficiency
what is the best sampling method if there is no sample frame
multistage sampling
how to decrease sampling error of multistage samples
increase sample size, increase homogeneity of units being sampled
what is snowball sampling
selected participant refers other participants
what is critical case sampling
select cases that are essential to acceptance or assessment
what is quota sampling
interviewers select sample that yields the same proportions as the population
proportionson easily identified variables
what is typical case sampling
selectcasesthatareknown beforehand to be useful and not to be extreme
what type of sampling are focus groups
non-probability, potentially convenience
when is non-probability sampling a good idea
with hard to reach populations, pretesting
what is the sampling error
variation around the true population measure
how do you decrease sampling error
increase homogeneity of sample, increase sample size
the lower the variability
the higher the power
what direction is the bias of non-response
unclear
if non-response is independent of the outcome variable
the chance of bias is lower
what is a wave analysis
analyze data for persons who respond initially compared to those who respond after
follow-up, assume that non-responders are more like f/u
analyze worse/best case scenarios
to calculate the pattern you would need in nonresponders to reverse the results