Y học - Chapter 6: Selection of research participants: sampling procedures

This is considered highly important in social and behavioral research Three basic questions to consider: 1. Are the research participants appropriate for the research question? 2. Are the research participants representative of the population of interest? 3. How many research participants should be used?

ppt23 trang | Chia sẻ: thuychi11 | Ngày: 18/01/2020 | Lượt xem: 219 | Lượt tải: 0download
Bạn đang xem trước 20 trang tài liệu Y học - Chapter 6: Selection of research participants: sampling procedures, để xem tài liệu hoàn chỉnh bạn click vào nút DOWNLOAD ở trên
Chapter 6 Selection of Research Participants: Sampling ProceduresSubject Selection and SamplingThis is considered highly important in social and behavioral researchThree basic questions to consider:1. Are the research participants appropriate for the research question?2. Are the research participants representative of the population of interest?3. How many research participants should be used?Technical Sampling TermsPopulation – refers to an entire group or aggregate of people or elements having one or more common characteristics Sample – a small subgroup of a population of interest thought to be representative of that populationSampling – the process of selecting a subgroup or sample of the populationSampling Frame – the accessible population or collection of elements from which the sample is actually drawnRandom Processes in ResearchRandom SelectionThe purpose is to enable the researcher to generalize the results to a larger population. Thus, the researcher is concerned about the “representativeness” of the subjects in the sampleRandom AssignmentThe purpose is to enable the researcher to assume that groups are “equivalent” at the beginning of the study. This adds control to a study; it has nothing to do with the selection of the sampleSample Selection MethodsProbability SamplingSampling techniques in which the probability of selecting each participant is knownUtilizes random processes, but does not guarantee the sample is representative of populationEstimates of sampling error are possibleNon Probability SamplingSamples are not selected at randomDifficult to claim sample is representative of populationIntact groups, volunteersSample Selection MethodsProbability SamplingSimple random samplingStratified random samplingSystematic samplingCluster samplingNon Probability SamplingPurposive samplingConvenience samplingSimple Random SamplingWith simple random sampling, every member of the population has an equal probability of being selected for the sample. Also, the selection of one member of the population does not affect the chances of any other member being chosen (equal and independent)Sampling with replacement vs. sampling without replacementUsual procedure:Fishbowl techniqueTable of random numbersComputer generated samplingStratified Random SamplingA stratified random sample is one obtained by separating the population elements into non-overlapping sub-groups, called strata, and then selecting a simple random sample from each strataNo sampling unit can appear in more than one strataA stratified sample will assure representation from each strataThe number of sampling units drawn from each strata depends upon the size of the sampling frame as well as each strata and whether the researcher wishes to maintain the same proportionality that is present in the populationSystematic SamplingAn alternative to simple random sampling in which the sampling units are selected in a series according to some predetermined sequence. The origin of the sequence should be controlled by chanceThe researcher will choose 1/kth of the sampling frame with k being any constant. The first sampling unit is randomly selected by the investigator. Thereafter, every kth unit in the sampling frame is chosenSimple random sampling is to be preferred, but systematic sampling is a practical and useful approximation to random sampling that is easier to performCluster SamplingCluster sampling or area sampling is a simple random sample in which each sampling unit is a collection, or cluster, of elements (e.g., classrooms, schools, counties, city blocks)The sampling unit is the “cluster”Cluster sampling is an effective design when (1) a good frame listing population elements is not available, (2) the removal of elements from their cluster unit is not possible, or (3) it is impractical to conduct simple random samplingThe first task is to delineate or specify the clusterNon Probability SamplingThe probability that an element will be chosen is not known, with the result being that a claim for representativeness of the population cannot be madeThe researcher’s ability to generalize findings beyond the actual sample is greatly limitedBut it is less expensive and less complicatedConvenience sampling and purposive sampling are common examplesPurposive SamplingWhen members of the sample are purposively selected because they possess certain traits that are critical to the studyLimited generalizabilityExample: Selecting elite athletes for a biomechanics studyConvenience SamplingRefers to selecting research participants on the basis of being accessible and convenient to the researcherOften involves use of volunteersLimited generalizabilityExample: Using fellow graduate students as research participantsSample SizeRegardless of size, the crucial factor is whether or not the sample is representative of the population, thus how the sample is selectedPoints to consider regarding sample size:Nature of the studyStatistical considerationsVariability of populationNumber of treatment groupsPractical factorsNature of the study Descriptive, correlational, or experimentalDescriptive and correlational studies typically should have more research participants Experimental studies often employ fewer research participants Statistical considerationsHow do you want to analyze the data? What statistical application will be used? Power of the statistical testPower is the probability that the test will reject the H0 when, in fact, the H0 is falseIn general, the larger the sample size, the more power of the statistic being usedGenerally N=30 is minimum needed to meet assumptions of many statistical proceduresVariability of populationSample size is inversely related to sampling errorThe larger the sample size, the smaller the sampling error and the greater likelihood that the sample is representative of the populationLittle variability – small sample will sufficeHigh variability – sample size will be largerNumber of Treatment GroupsWhen samples are divided into smaller groups to be compared, it is important that the subgroups are of adequate size Should be more concerned with “cell size” than total sample sizePractical FactorsAvailability of research participantsCostsTimeComments About SamplingDescriptive and correlational research are vitally concerned about the representativeness of the sample, usually necessitating larger sample sizes and more attention given to the sampling procedureExperimental studies can often get by with small sample sizes, as long as internal validity is maintainedIn practice, volunteer research participants are involved in a good portion of research. Be aware of the potential of systematic error being introduced in the studyRandom AssignmentThe purpose is to establish “group equivalency” before the introduction of the independent variableTwo basic methodsIndependent groups designRepeated measures designIndependent Groups Design Each research participant is randomly assigned to one of the various treatment groupsEach subject participates in only one groupRepeated Measures Design Subjects participate in more than one group (treatment condition)In the simplest example, each research participant would be assigned to each level of the independent variable and then is measured after receiving the treatmentCounterbalancing is often used to control for possible order effect
Tài liệu liên quan