1/39
Looks like no tags are added yet.
Survey
the process of polling, or surveying, an identified group of respondents on topics of interest - people's attitudes, perceptions, intentions, behaviors.. etc
What is the central goal of most survey research?
to provide sample estimates of population values that are as accurate as possible.
What is the core requirement for random assignment?
that each person participating in the study has the same chance of being assigned to a specific experimental or control condition as any other person.
Sampling or Selection
how respondents got into the sample.
goals of random assignment versus sampling procedures
random assignment is necessary to achieve high internal validity for making causal inferences. Sampling procedures are concerned with external validity to allow generalizing results to a population of interest.
Census
an investigation involving all the potential units that could be included from the target population.
sample
a subset of eligible units in a population
a sample if preferred to the use of a census because...
within some reasonable degree of sampling error, the sample will approximate the results that would have been obtained had a complete census been taken.
sampling error
the expected typical discrepancy between the estimate calculated from a sample the value that would have been obtained if the entire targeted population (census) had been used.
sampling frame
a listing of the population of interest from which some members of a population are drawn and used as the sample for a study.
random sampling
a random mechanism is used to obtain a sample intended to be representative of the underlying population.
random sampling AKA
probability sampling , because the probabilities of a member of a population being included in the sample may be determined, although probabilities of member selection may or may not be equal.
non-random sampling
a non-random mechanism is used to obtain a sample from a population.
non-random sampling AKA
non probability sampling, the probability that members are selected from a population cannot be determined, usually because a sampling frame listing of potential respondents or locations is unavailable.
simple random sampling
every member of the population in question has an equal (and nonzero) probability of being selected every time a unit is drawn for inclusion in the sample.
how is the probability of selection in simple random sampling calculated?
the sampling fraction is calculated by dividing the number of units to be included in the sample by the total number of units in the population.
epsem designs
equal probability of selection method. examples: simple random sampling, systematic sampling, and proportionate stratified sampling
systematic sampling
sampling every predetermined nth member from a population.
stratified sampling
the population is divided into theoretically meaningful or empirically important strata before members are randomly drawn from each stratum ( or sub population ) and used for the sample.
oversampling
purposefully collecting extra data from a small underrepresented group so you have enough information to study them accurately.
why stratify?
to offset the selection of unusual or non-representative group of respondents from a subpopulation.
cluster sampling
geographic locations are randomly sampled, and all members from the selected clusters are used for the sample.
multistage sampling
clusters of locations are randomly sampled from a geographical sampling frame, and then participant units within each cluster are randomly sampled as well.
probability proportional to size sampling (PPS)
ensures that the likelihood of selection in a cluster or multistage sample is the same for all potential sampling units (or respondents) no matter the size of the cluster from which they are drawn.
two-phase sampling
all respondents who might have been chosen in ANY sampling method take the survey
panel survey
a pre-specified sample of respondents (a panel) is surveyed repeatedly, over time.
non-random sampling
respondents are obtained in any way possible, so that an adequate sample size and sufficient statistical power is achieved.
convenience sampling
the sample consists of people most readily accessible or willing to be part of the study.
Snowball Sampling
initially sampled participants are asked to contact and recruit other in their social network.
quota sampling
members are sampled non-randomly until a predefined number of respondents for each of a specified set of subgroups is achieved.
sample size
the decision regarding the size of the sample involves trade-offs, most of which concern the complementary issues of cost and precision.
general sampling weight
used to calibrate the demographic characteristics of the sample to a known population.
random digit dialing (RDD)
the use of a random number generator to develop lists of telephone number, has solved issues associated with the disappearance of the landline but has practical deficiencies of its own. (some numbers aren't always assigned to residential dwellings)
telephone vs face-to-face
both have different strength - economy, efficiency, likelihood of valid responses given the issue involved- but strengths have evolved as time passed. the choice of method will depend largely on available resources and the context of the study.
internet surveys
respondents can be contacted by email or through the Web and directed to a set where they can complete a survey instrument online.
missing data
respondents sometimes answer a survey only partially, skipping over certain questions they do not want to answer.
missing completely at random (MCAR)
occurs if the probability of missing responses in a variable is unrelated to any other variable in the study.
missing at random (MAR)
occurs if the probability of a missingness in a variable is not random, but its missingness may be fully explained by the other measured variables in the dataset.
not missing at random (NMAR)
occurs if a variable is not missing at random and its missingness is not explainable by other measured variables in the dataset (but is explicable by other unmeasured variables) least desirable.
Possible solutions to missing data.
financial incentive, assuring confidentiality