Y học - Chapter 9: Descriptive research

Chapter 9 Descriptive ResearchOverview of Descriptive ResearchFocused towards the presentGathering information and describing the current situationMay or may not involve hypothesis testingAnswers the question “What is?”Many types of research fall within this classificationTypes of Descriptive ResearchSurveyDevelopmentalLongitudinal approachCross-sectional approachCase StudyCorrelationalNormativeObservational (a.k.a. qualitative)ActionCausal-comparative (a.k.a. ex post facto)Survey Research MethodologySurvey research is the most common type of descriptive researchInvolves questioning techniques for data collectionSurvey methodology consists of asking questions of a (supposedly) representative sample of the desired population at a single point in time. The persons of whom the questions are asked are called survey respondentsThe most difficult part of conducting a survey is writing the questionsCensusA survey which obtains responses from the entire population is called a censusGeneral Survey Data Collection MethodsInterviewsQuestionnairesInterview MethodsPhone interviewCommon in marketing research Not used much in HHP Personal interviewApplicable if the sample is small and accessibleStructured, semi-structured, or unstructured interviewRecording of informationFocus group interviewInterview groups of peopleRequires skilled facilitatorAn interview schedule or guide contains the questions to be read to the respondent during an interview as well as a place for recording the answersQuestionnairesA questionnaire is a self-report instrument that is generally mailed or handed to the respondent to complete with no help from the researcherAdministered questionnaire Respondents are directly given the questionnaireDistributed questionnaire Questionnaire is mailed or electronically deliveredMajority of survey research in HHP uses a questionnaire as data-collection techniqueQuestionnaire DevelopmentComposing Questions – each question should have three important attributes: focus, brevity, and simplicityThe questions should focus directly on the issue or topic relevant to the information needs of the studyThey should be as short or brief as possible while still conveying the meaningThe questions should be expressed as simply and clearly as they can beQuestionnaire DevelopmentQuestions should be worded so they are not ambiguous, misleading, or biasedDouble-barreled questions should be avoidedUse simple sentences where possible and complex sentences only when essentialLevel of wording (vocabulary) should be appropriate for the intended respondentsQuestionnaire FormatOrganization of Questionnaireappearance and layout is importantlength (no. of questions) is a major factorin general, longer questionnaires have a poorer return rateplace easy to answer questions firstsensitive questions should be near the enditems should be placed in a logical orderdemographic informationrecommend placing demographic questions at endResponse FormatClosed-ended (structured)standard answers providedeasy to code and scorefacilitates answering sensitive questionsmay make for a long questionnairegenerally considered easy for respondentOpen-ended (unstructured)respondent may answer as they chooseexploratory; allows detailed responsepreferable for complex questionsmay be difficult to code and scorerequires more time and effort of respondentAppropriateness of QuestionnaireValidity and reliability of the questionnaire should be determined before it is administeredUse of a pilot studyRevise questionnaire as neededQuestionnaire DistributionControlling costs bulk mail rateslength of questionnaire (i.e., weight of mailing)High return rateself-addressed, stamped envelope (SASE)postage-paid, business reply envelopesgood mailing listfollow-up strategyCover letterAppropriate timing of mailingCover Letter GuidelinesAll mailed surveys should include a cover letter explaining the following:who is conducting the surveywhat is the purpose of the surveywhy it is important for the respondent to answerinducements for the respondent, if anyhow is the confidentiality of the respondent being protectedbasic instructions for completing and returning the questionnaireCover Letter Guidelines – 2Personalize cover letter if possible (mail merge)this has been shown to increase the return rateCorporate or institutional letterhead and printed envelopes should be used for the cover letter and mailing envelopesMailing GuidelinesThe typical survey questionnaire should be mailed in a standard number 10 business envelopefirst class postage stamps on the mailing envelope will result in the highest return ratebulk mail obtains the lowest return rate, is slower, but is less expensive, and can be over 3 ounces without extra postage . . . easily justified for large mailingsReturn Mail GuidelinesThe self-addressed return envelope should be smaller than the mailing envelope so that it can be inserted in the mailing without foldingusually a no. 9 size business envelopereturn postage should be providedaffixing 1st class postage stamp results in the highest return ratea postage paid business-reply return envelope is more economical and is an acceptable alternativeSurvey Research ErrorsInformation obtained from a survey depends on the questions that are asked, on how the respondents react to the questions, and on what respondents choose to reveal about themselvesResearchers are somewhat limited in their ability to assess the quality of the findings since there is usually no way of checking the accuracy or truthfulness of the responsesGathering meaningful information is difficult because survey research is particularly susceptible to two types of errors:Nonsampling errorsSampling errorsNonsampling ErrorsNonsampling errors have several sources including any differences in the way the researcher and the respondents interpret questions, the inability or unwillingness of the respondents to provide correct or honest information, mistakes made when recording or coding the responses, and missing responsescontrolling response errors when subjects provide incorrect or false information is particularly difficult since they depend on the behavior of the respondents, something that is beyond the control of the researchercontrolling processing and data collection errors can be largely controlled by the researcherSampling ErrorsSampling errors are those errors resulting from any differences between the data obtained from the sample, and the data that would have been obtained from the complete populationthis is the type of error made from inferring a population characteristic based on a sampleSampling error or margin of error is often reported with survey findingsFor example, a survey reporting that 57% of Iowans favor abortion with ± 3.5% margin of error really means that we can be 95% confident that the true population value lies within the ranges 53.5% to 60.5% (57% ± 3.5%)