LES12

nullSamplingSamplingPopulationSampleSampling vs. censusSampling vs. censusSampling much cheaper Applicability/destruction (i.e. fireworks) Precision Sampling may be more precise than census due to nonsampling errors Accuracy Sampling, if properly selected, is sufficiently accurate in most casesStages in samplingStages in samplingDefining the target populationSelect a sampling frameDetermine if a probability or nonprobability sampling method will be chosenPlan procedure for selecting sampling unitsDetermine sample sizeSelect actual sampling unitsConduct fieldworkMeasurement errorMeasurement errorTotal errorRandom sampling errorRespondent errorNonresponse errorResponse biasSystematic error (bias)Administrative errorInterviewer cheatingInterviewer errorSample selection errorData processing errorUnconscious misrepresentationDeliberate falsificationSocial desirability biasAuspices biasInterviewer biasExtremity biasAcquiescence biasNon-sampling errorSampling errorSampling errorPopulationSampling frameSampleResponsesProbability & Nonprobability samplingProbability & Nonprobability samplingNonprobability sampling convenience sampling who happens to be there judgmental sampling Vejle represents the whole of Denmark quota sampling ie. equal number of M + F snowball sampling ”pass the word” Probability sampling simple random sampling systematic sampling i.e. every 5th person stratified sampling proportional/ disproportional (i.e. targets groups have more weigth) cluster i.e 10 people in 4 cities Probability samplingProbability samplingx x x x x x x x x x x x x x x xx x x x x x x x x x x xx x xxx x x y y y y y x x y y y y x x y y y y x x z z z z x x z z zz x x z z zz z z x z z zz zxxxxxxxx xxxxxx xxxxxxxx xxxxxx xxxxxxxx xxxxxx xxxxxxxxxx xxxxxxxxxx xxxxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x xSimple random samplingCluster samplingStratified samplingSystematic samplingSampling designSampling designAccuracy required Resources Time Advance knowledge of population Subsequent analysisSample sizeSample sizeHeterogeneity of population Acceptable error (absolute) Confidence level (relative) (Interrelated) Sample sizeSample sizeCodingCodingCode: values assigned to the individual operational items when measurement actually takes place (precoding/postcoding) coding sheet precoded questionnaire computerised interviews scanners Rules for codingRules for coding Codes should be intuitive and reasonable (e.g. when reflecting magnitudes and ranking) Mutual exclusive Not too many options Code book Reliability and data entry tools Editing and cleaningEditingEditingRecoding/checking simultaneously or afterwards Obviously wrong response Recall Missing values Random/average