exam 2 Flashcards
population
group of interest
sample
subset of the population that is chosen for the study
purpose of sampling
do not need and cannot possibly study entire population
goals of sampling
external validity, economic sample, representative of the population
sampling error
extent to which the sample is not representative
biased sample
some parts of sample are overrepresented, some underrepresented
nonprobability sampling
easier than probability sampling
convenience sampling
choosing people for your sample based on proximity and willingness… easiest/quickest way to sample
quota sampling
similar to convenience sampling, except you take steps to ensure that a certain number of people from a particular group are included in your sample
purposive sample
choose certain pockets of people to be in your sample that, in the past, have reflected the views of the views of entire population
probability sampling
better. everyone has equal probability of being in sample
error of estimation (margin of error)
degree to which the data obtained from the sample are expected to vary from the population. the smaller the error, the better the sample. function of three things: sample size, population size, variance of data
simple random sampling
type of probability sampling in which you choose a random set of people from the whole population tot be your sample. you need a sampling frame– the name and contact info for everyone in your population
not commonly used
systematic sampling
type of probability sampling in which you choose “every so many” individuals for the sample. do not need a sampling frame
stratified random sampling
divide population into strata; choose randomly from each stratum. need a sampling frame
cluster sampling
sample clusters of population first then work your way to the sample. do not need a sampling frame
threats to representativeness
- response bias - problem of nonresponse
- selection bias - sampling procedures produce biased samples. automatic in nonprobability samples can happen with probability samples
- misgeneralization - attempting to generalize results based on an unrepresentative or biased sample
descriptive research
used to describe the characteristics or behaviors of a given population in a systematic and accurate way
relies on external validity
two ways to collect data
- watch and record (observation and physiological measures)
- surveys (questionnaires and interviews)
naturalistic observation
observing an individual group in their natural habitat. this is unobtrusive and external validity is important
how to achieve unobtrusive observation
hide, habituation, participant observation, group infiltration
behavioral recording
observing behavior and recording “what” is happening
narratives
unstructured notes containing a full description of everything the subject says or does. typically from video or audio tapes.
field notes
more concise descriptions of behavior
checklists
structured description containing a tally of specific behaviors. behaviors to record are decided in advance, strong use of operational definitions
temporal measures
when and how long a behavior occurs
latency measures
elapsed times between 2 behaviors, or event and behavior
duration measure
how long a behavior lasts
rating scales
for measuring quality or intensity of behavior
physiological measures
measuring internal processes that are not readily visible
surveys/self-report measures
includes questionnaires and interviews, a survey is as only good as its questions, how you implement a survey can influence how you interpret results
cross-sectional survey
sample consists as a “cross-section” of the population. one group tested at one time point. no major drawbacks but you must be valid and reliable
successive independent sample surveys
2 or more samples of respondents answer the same questions at different points in time.
drawbacks: two samples might not be comparable and variance may change
two samples surveyed one time each
longitudinal survey
a single sample of respondents is questioned more than once. assesses how responses change over time.
drawbacks: attrition and time
internet survey
develop an online questionnaire and administer it thru website or specialty software.
advantages: inexpensive, many subjects, no data entry
drawbacks: biased and limited sample, must follow directions
correlational research
research that focuses on the degree to which two variables are related. predicts outcomes but doesn’t imply causation
correlation coefficient
indicates the degree to which two variables are related. always a number between 0 and 1, sign only indicates direction, can only use with a linear relationship
positive correlation
as variable A increases, so does variable B and vice versa
negative correlation
as variable A increases, B decreases and vice versa
factors that distort correlations
- restricted range- restrict ages you are looking at
- outliers - online outliers exaggerate a nonexistent correlation
offline outliers deflate or reduce existing correlation
effects of unreliable measures
lower correlations
introduce wrong variability
creates noisy data which obscures the relationships between variables
noisy data
data w a large amount of additional meaningless information
coefficient of determination
(pearson’s r)^2
determines how much of the variability in factor A can be explained by B
measure of systematic variance
on ratio scale
third variable
connects two other variables or gives an explanation
ex. more fire hydrants and more crime are linked to urban areas/high density housing
multiple-regression analysis
shows the relationship between two variables while holding another measured variable constant. helps rule out alternative explanations that may possibly account for the correlation. increases the validity of the study.
limitations of multiple-regression analysis
can only rule out variables that you measure
internal validity is now better, but still to low to make causal claims
predictor variables
what factors may be related to the behavior of interest?
independent variable
criterion variable
what is the behavior of interest that is changing?
dependent variable
longitudinal designs
can provide evidence for temporal precedence by measuring the same variables in the same people at several points in time
one sample surveyed multiple times
cross-sectional longitudinal design
investigates covariance
one sample surveyed one time
autocorrelation longitudinal design
investigates stability of traits or behaviors over time
cross-lag correlation longitudinal design
investigates which variable comes first (temporal precedence)
limits of longitudinal desgin
low internal validity
haven’t ruled out alt. explanations and possible third variables
temporal precedence
establishing that one variable comes before another.
acquiescent response style
individuals are most like to agree with a statement whether its true or not
correlations satisfy….
covariance
covariance
measure of relationship between two random variable
multivariate longitudinal correlational study
measure two variables at (at least) two different time points
unreliable data typically results in
weaker correlations than what really exists
pattern and parsimony
researchers investigate causality by using a variety of correlational studies that all point in the same causal direction
mediator
explanatory variable, used to show how A may be causing B
why are they related
meow
meowww
criteria for cause
covariance
temporal precedence
internal validity
