2022熊猫学术Research Methodology
Marketing Research &Information:Lecture 7 SamplingAaker, Kumar, Day (2007) Ch 14Bradley (2007) Ch 5Saunders, Lewis & Thornhill (2007)Research Methods for Business StudentsThe Research ProcessIdentify problem/info needResearch objectivesDecide on your approach(methodology), and formulate a planResearch Method &Data Collection ToolSample MethodCollect informationAnalyse the dataPresent the findingsSamplingn Choosing the type of people to surveyn Locating themn Deciding how to select the individualsinvolvedn Deciding how many to surveyn Identifying how representative theinformation collected isStages of sampling1. Examine the objective of the study (PURPOSE)2. Define the people of interest (POPULATION)3. Find suitable source for the population members4. Decide on the sampling type and approach(PROCEDURE)5. Decide on the sample size6. Proceed with the fieldwork7. Correct sampling errors ready for reporting(PUBLICATION)What is sampling?n The process of obtaining information from asubset of a larger groupn Populationn The entire group of people (universe) about whominfo is neededn Censusn A collection of data obtained from every memberof the populationn Samplen A subset / segment of all the members of apopulation of interestSamplingn For what reasons might it be necessaryto select a sample rather than collectdata from the population?Sample DesignProbabilitysamplen A member has apositive, calculableprobability of beingchosenn Response rateimportantn Is expensiven ObjectiveNon-probabilitysamplen Uses human judgmentn Subject to errors thatcannot be measuredn Usually cheaper thanrandom methodsn SubjectiveSampling Methodsn Probabilityn Simple random samplingn Systematic samplingn Stratified random samplingn Cluster samplingn Non-probabilityn Convenience samplingn Judgement samplingn Quota samplingn Snowball samplingSampling Process1. Define the population of interest2. Determine whether to sample orcensus3. Select the sampling frame(probability only)4. Determine sample size5. Choose a sample method6. Implement1. Define the population ofinterestn Specify the characteristics of the individualsfrom whom or about whom information isneeded to meet research objectivesn Bases for defining the Population of Interestn Geographic arean Demographicsn Usagen Awareness2. Determine whether to sampleor censusn Difference between a sample and apopulation?n Choice depends on size of population,3. Select the sampling framen Sampling frame is a list of populationitems/elements/ units. The sample isselected from this list.n OR the frame is a specific procedure forgenerating such a listn The population must be defined – e.g.production employees; car ownersn NB take care when using phonedirectory as a listGood frames aren up to daten completen affordablePoor frames aren oldn incompleten inappropriateSampling FramesCommon SourcesCommon frames includen the Electoral Registern the Postcode Address File (PAF)n Telephone directoriesn Subscriber / membership recordsn Customer recordsIdentify a suitable samplingframe for each of the following:n A study to identify how company directors ofmanufacturing firms where there are >500employees think a specified piece oflegislation will affect their companies?n A study to identify how employees atCheltenham Gardens Ltd think the proposedintroduction of compulsory Saturday workingwill affect their working lives?How sample size isdeterminedn Sample size is generally agreed beforefieldwork, but in some projects it may bedetermined during fieldworkn The sample size for any study depends onn the purpose, in terms of required precisionn the size and nature of the population under study.n Procedural aspects such as time, budget &resources availablen Publishing aspects, in terms of importance ofresultsHomogeneous?The important thing ishow homogeneousthat population israther than its fullsizeFor 1 million peoplewho are identical weonly need a sample ofone person. If there are200 different people weneed to interview all200.Sample size decisionsMain ways to decide the sample size:n by calculationn by using "accepted" industrystandardsn by budget (time or money available)n by "building" analysis cells5. Determine sample sizen Factors in selecting sample size:n Variability in the populationn Required level of confidencen Required limits of accuracyn Likely level of non-responsen Requirements for analysis of subgroupsn Practical factors – e.g. budgetsDetermine sample sizen Governed by:n Confidence levelsn Desired margin of errorn Types of analyses plannedn Size of population (see end of notes for tableof sample sizes)n Expected response ratesExample calculation ofresponse raten Estimate the sample size required, e.g.210n Out of the questionnaires returned (e.g.129, deduct the voids, e.g. 8).Responses total 121n 121 as a percentage of 210 = responserate, i.e. 57.6%Example calculation of samplesize when response rate is knownn If you estimate that you will get 25% ofyour questionnaires returned, and youknow that the desired sample size forthe population is 360, the actual sampleyou should contact is:360 x 100 = 144025Probability samplingn Probability sampling- every member of thepopulation has a known likelihood of selectionn Therefore can only be used when some sortof a list identifying members of thepopulation is available, e.g. staff list;customer list; accounts (as in customeraccounts) databasen AKA representative samplingProbability sampling methodsn Simple random samplingn free from biasn uses a sampling frame, sample generated bygenerating list of random numbersn Systematic samplingn Choose every nth item after a random startn Advantagesn Easy to use and cost effectiven Disadvantagesn Could be biased if there is a regular pattern to thepopulation which coincides with the sampling methodn Not completely randomExample calculation of asystematic samplen There are 750 people employed by anorganisation. With a 5% confidence level, 254employees should comprise the sample. Whatis the sample fraction (i.e. how many namesshould be skipped each time the sample isbeing selected from the list of employees)?750/254 = every third name should beselected (approx)n see Saunders, M. Lewis, P. & Thornhill, A.(2007:212)Probability sampling methodsn Stratified random samplingn Variation on random sampling. Thepopulation is divided into strata /categories. Random samples taken fromeach strata.Probability sampling methodsn Cluster sampling: non-random samplingmethod that involves selecting one definablesubsection taken to be representative of thepopulation in questionn Advantagesn Good alternative to stratified sampling ifsatisfactory sample frame not in existencen Inexpensive to operaten Disadvantagesn Potential for considerable biasProbability samplingn Advantagesn Sampling error can be calculatedn Results can be projected to the wholepopulationn Researcher can be sure of obtaining resultsfrom a representative cross section of thewhole populationn Disadvantagesn Expense, time consumingNon-Probability samplingn Non-probability sampling – sampleswhere specific elements of thepopulation have been selected in a nonrandommannern Used where there is lack of an availableframe to select the sample from, e.g.coffee drinkers.n The process is otherwise similarNon-Probability samplingmethodsn Convenience sampling – where samples aredrawn at the convenience of the interviewern Advantagesn Able to gather substantial amounts of data fromhigh traffic locationsn Quick & easy to implementn Disadvantagesn Potential for considerable biasn Misrepresentation of populationNon-Probability samplingmethodsn Judgement (purposive) sampling – where theinterviewer uses an educated guess as towho should represent the populationn Advantagesn Viable for focus groupsn Useful if researcher has knowledge about thepopulationn Disadvantagesn Subjective, some members of the population willhave smaller chance of selectionNon-Probability samplingmethodsn Quota sampling – interviewers are toldto interview all the people they meet up#p#分页标题#e#to a certain quotan Advantagesn Cheap and easy, no sampling frame necessaryn Disadvantagesn Bias – cannot make a valid estimate ofsampling errorQuota SamplingMost commonmethod,typically usesn Agen Sexn Social GradeTypes of QuotaInterlockingn (interrelated)n Strata are linkedn Greater controlover interviewern Difficult taskn RepresentativeNon-interlockingn (independent)n Strata unconnectedn Simple taskn Likely to skew toco-operative peoplen Less usedNon-Probability samplingmethodsn Snowball sampling – new respondentsselected based on referrals from initialrespondentsn Advantagesn Good for low incidence populations,reduces research costsn Disadvantagesn Bias – respondents may be reluctant togive referralsNon-probability samplingn Advantagesn Cheaper & quickern Acceptable where accuracy is not critical –i.e. exploratory researchn Disadvantagesn Unaware as to what extent the samplerepresents the populationn Sampling error cannot be calculated, soresults cannot be accurately projected ontothe whole populationn Qualitative researchalways uses nonprobabilitymethods ofsamplingn So subjectivity andhuman judgement isalways involvedQual & quantsampling differencesRepresentative?n Qualitative samples will NOT berepresentative of the populationn Many quantitative samples ARErepresentative of the populationn But some are drawn to examine extremes,non-users, or users of above-averagequantities. Samples may be selected tounderstand minorities.n The end-sample may be representative ofthese groups, but not the overall populationSample sizesn In Qualitative research sample sizedetermination is more subjectiven In theory sample sizes should not be fixedfirmly at the start of the project. Theoverriding idea is that new cases should beselected until the data brings nothing new.n In practice a methodological compromise ismade and most proposals set a certainnumber of groups and depths. This thenallows budgets to be controlled andtimetables to be developedSourcesn Aaker, D. Kumar, V. & Day G. (2007)Marketing Research.9th Edn. NewJersey: Wileyn Bradley, N. (2007) Marketing ResearchOxford:OUPn Saunders, M. Lewis, P. & Thornhill, A.(2007) Research Methods for BusinessStudents, Harlow: Pearson Education
