Sampling Basic Terms Research units subjects participants Population

Sampling

Basic Terms • Research units – subjects, participants • Population of interest (all humans? ) • Accessible population – those you can actually try to sample • Intended sample – those you select for participation • Actual sample – those from whom you actually obtain data

Proximal Similarity Model • Donald T. Campbell • To whom can you generalize your results? • To the extent that the population is similar to the sample, generalization should be good. • Typical Sample in Psychology is – Students in Introductory Psychology – Laboratory Animals

Simple Random Sampling • Definition of a random sample • How to obtain one – Sampling frame – a list of all the members of the target accessible population – Each member assigned a random number – Sort by those random numbers – Select n units from the N members • Sampling fraction = n / N • Assumption that sampling fraction = 0

Stratified Random Sampling • Divide population into strata (nonoverlapping homogeneous subgroups) • Sample nj subjects from each stratum • Proportionate stratified random sampling • Disproportionate stratified random sampling

Proportionate Stratified Random Sampling • You sample the same proportion from each stratum • For example – 10% of all freshmen at ECU – 10% of all sophomores at ECU – 10% of all juniors at ECU – 10% of all seniors at ECU – 10% of all graduate students at ECU

Disproportionate Stratified Random Sampling • Some strata have relatively few members • But you want to get a sufficient number of subjects for each stratum • So you sample a larger proportion of those strata with fewer members • For example, nondegree students or doctoral students.

Cluster Random Sampling • Sampling across a wide geographic region. • Divide the population in clusters – for example, counties in North Carolina. • Randomly sample clusters. • Gather data on all target subjects within each randomly sampled cluster. • For example, all city managers in the selected counties.

Multi-Stage Random Sampling • Combine two or more techniques • Example – Randomly select 100 classes (clusters) at ECU. – From each class, randomly select 5 students.

Nonrandom Sampling • Convenience Sampling – get what you can without a lot of hassle – Stand outside of Rawl and try to recruit anybody who comes by • Purposive Sampling – convenience sampling but where you have inclusion/exclusion criteria – For example, subject must be African. American and not live in North Carolina

Nonrandom Sampling • Modal Instance Sampling – you define the “typical” member of the population and then recruit only such members – ECU: 18 year old female resident of North Carolina • Expert Sampling – recruit only persons who are known to expert in some domain – Designing a survey on social aggression, recruit experts to judge potential survey items.

Nonrandom Sampling • Proportional Quota Sampling – convenience sampling, except you want subgroups represented in same proportions they are in the target population. – ECU: 30% freshmen, 30% sophomores, 20% juniors, 20% seniors.

Nonrandom Sampling • Non-proportional Quota Sampling – convenience sampling, except you have specified (nonproportionally) how many subjects you want in each subgroup

Nonrandom Sampling • Heterogeneity Sampling – you want to have adequate numbers of people in each of two or more groups with disparate opinions. – For example, those who thought the world would end this year, and those who did not – There a lot fewer of the former, so you would need sample a larger proportion of them.

Nonrandom Sampling • Snowball Sampling – Identify people who meet your inclusion criteria (for example, lifeguards) – Ask them not only to complete your survey, – But also to send it on to other similar persons they know and ask them to complete it. – Birds of a feather flock together.