Chapter 5 & 6 Flashcards
Sample and sampling
.a subset of participants from a population
.process researcher selects sample of participants in a study
Probability sample
the likelihood that any particular individual will be selected for a sample can be specified
.simple random sample is a type of probability sample
simple random
every possible sample of the desired size has the same chance of being selected from pop
.every individual has equal chance of selection for sample
what’s another way to test assess generalizability of findings without resorting to using a simple random sample?
just by replicating a study on other samples of participants that differ in age/education/socioeconomic status/geographic region/personal or psychological characteristics…
Representative sample
.can draw accurate, unbiased estimates of characteristics of the larger population
.can only do this if the sample is REPRESENTATIVE
.still note this is very time consuming and not what is commonly used in psychological research!
sampling error
fact that individuals selected for the sample differ from characteristics of the general population
error of estimation (margin of error)
the statistical degree to which a sample differs from a population it is from
.ONLY meaningful if what he have is a probability sample, otherwise error of estimation is meaningless
Error of estimation is a function of three things:
sample size, population size, variance of data
The greater the variability in the data, the more difficult it is to..
estimate the population values accurately. So larger the variance, larger the sample needs to be to draw accurate inferences about the population.
systematic sampling
take every nth individual for the sample
stratified random sampling
.divide the population first into subgroups or strata (singular stratum)
.like dividing into men and women, or diff age ranges
.then individuals randomly sampled from each strata
proportionate sampling method
.cases sampled from each stratum in proportion to their prevalence in the population
.e.g. if we know 55% in city are demos, and 45% are repubs, we would want to make sure our two strata reflect that ration
cluster sampling
.researcher first samples groupings or CLUSTERS of participants
.clusters often based on geography or particular institutions (schools)
multistage cluster sampling
sample large clusters, then sample smaller clusters from within the large clusters, then sample even smaller clusters until finally obtain sample of participants
.advantages are that we don’t need a sampling frame of the entire population! just a list of institutions or something.
.second, easier to contact participants that share something in common like living area or school
nonresponse problem
failure to obtain responses from individuals selected from a sample, or further responses after initially surveyed
misgeneralization
researcher generalizes sample results to a population that differs from the one from which the sample was drawn
.think of the landon roosevelt elections
nonprobability sample
researchers have no way of knowing the probability that a particular case will be chosen for a sample
.most research uses this
convenience sample
includes participants that are readily available (like undergrads!)
.not usually a prob because we’re just looking at relationships between variables and not looking to make a statement about population as a whole
quota sample
researcher takes stepes to ensure certain kinds of participants are obtained in particular proportions (like equal amount of men and women)
purposive sample
researchers use past research findings or judgment to decide which participants to include in the sample, trying to choose respondents who are typical of the population they want to study.
economic sample
one that provides researchers with a reasonably accurate estimate of the population at reasonable effort and cost. Not too large, but just big enough to get a small margin of error!
Power
Ability of a research design to detect any effects of the variables being studied that exist in the data.
.particular studies might be powerful enough to detect strong effects but not weak ones, depending on their sample sizes
.larger samples needed when expected effects are weaker
Descriptive research
describe characteristics or behaviors of a given population in a systematic and accurate fashion.
.not designed to test hypotheses, but rather provide info about the physical, social, behavioral, economic, or psychological characteristics of some group of people
Cross-sectional survey design
A single group of respondents - a “cross-section” of the population - is surveyed.
successive independent samples survey design
two or more samples of respondents anser the same questions at different points in time
.e.g. different samples each time. like looking at percentage of americans who say they attended religious services in the past week over decades
probs with successive independent samples survey design
samples may change drastically demographically, ethnically, socioeconomically year over year! so can’t always trust conclusions
longitudinal or panel survey designq
single group of respondents questioned more than once
demographic research
concerned with describing and understanding patterns of basic life events/experiences such as birth/marriage/divorce/employment/migration/death.
.e.g. why people have number of children they do, socioeconomic factors that predict death rates
epidemiological research
used to study occurrence of disease/death in different groups of people. psychologists like because can document occurrence of psychological problems like depression, alcoholism, child abuse, shizophrenia, etc, and see if they have any effects
simple freq distribution. grouped?
displays number of times each score obtained
groups them from 1-9, 10-19, and so on… can give us relative freq like what percentage scored in each group?
diff between histogram and bar graph?
histogram the bars touch each other and the data is continuous (think normal dist or something)
bar graph the data does not touch each other, data ordinal or categorical
Confidence interval and error bars
CI, 95% confident successive samples will fall within the CI for example
Error Bars are the I-shaped vertical lines extending from tops of bars in bar graphs.
So 95% samples would fall within the error bars.
Positively skewed vs negatively skewed
looks like is approaching the graph if positive (yeah i like this!)
looks like running away if negative (get me outta here!)
Z-score
indicates how far from the mean in terms of standard deviations the score falls.
.outliers fall below -3 or above +3
.one std dev above and below captures 68% or responses