Data editing ( In research methodology )

  • View
    6.370

  • Download
    1

Embed Size (px)

DESCRIPTION

 

Text of Data editing ( In research methodology )

  • 1. Data editingIN RESEARCH METHODOLOGY
  • 2. Data editing
  • 3. Data integrity refers to the notion that the data file actuallycontains the information that the researcherneeds to provide the decision maker data integrity extends to the fact that the datahave been edited and properly coded Any errors harm the integrity of the data.
  • 4. Editing Fieldwork often produces data containingmistakes How long have you lived at your currentaddress? 48 What is your age? 32 years This answer contradicts the earlier response.If the respondent is 32 years of age, then howcould he or she have lived at the sameaddress for 48 years?
  • 5. Therefore, an adjustment should be made toaccommodate this information The most likely case is that this respondenthas lived at the current address for four years This example illustrates data editing
  • 6. Editing is the process of checking andadjusting data for omissions, consistency,and legibility So, the editors task is to check for errors andomissions on questionnaires or other datacollection forms. When the editor discovers a problem, he orshe adjusts the data to make them morecomplete, consistent, or readable.
  • 7. Field Editing Field supervisors often are responsible forconducting preliminary field editing on thesame day as the interview USES 1.Identify technical omissions such as a blankpage on an interview form 2. Check legibility of handwriting for open-ended responses 3. Clarify responses that are logically orconceptually inconsistent.
  • 8. Field editing is particularly useful whenpersonal interviews have been used to gatherdata. A daily field edit allows supervisors to dealwith some questions by asking interviewers,who may still be able to remember theinterviews, about facts that may allow errorsto be identified and perhaps corrected A daily field edit allows fieldworkers to identifyrespondents who should be recontacted to fillin omissions in a timely fashion.
  • 9. For example, if an interviewer did notcorrectly follow skip patterns, training may beindicated. The supervisor may also notice that aninterviewer is not properlyprobing some open-ended responses. Defn:Preliminary editing by a field supervisoron the same day as the interview to catchtechnical omissions, check legibility ofhandwriting, and clarify responses that arelogically or conceptually inconsistent
  • 10. In-House Editing In-house editing rigorously investigatesthe results of data collection The research supplier or researchdepartment normally has a centralized officestaff perform the editing and coding function
  • 11. Arbitron measures radio audiences by havingrespondents record their listening behaviortime, station, and place in diaries The diaries are returned by mail, in-houseeditors perform usability edits in which theycheck that the postmark is after the last dayof the survey week, verify the legibility ofstation call letters (station WKXY could looklike KWXY) If the respondents age or sex is notindicated, the respondent is called to ensurethat this information is included
  • 12. INCONSISTENCY Consider a situation in which a telephoneinterviewer has been instructed to interviewonly registered voters in a state that requiresvoters to be at least 18 years old. If the editors review of a questionnaireindicates that the respondent was only 17years old The editors task is to correct this mistake bydeleting this response
  • 13. In which of the following cities have youshopped for clothing during the last year? San Francisco Sacramento San Jos Los Angeles Other _________ Please list the clothing stores where you haveshopped during the last two months.
  • 14. Suppose a respondent checks Sacramentoand San Francisco to the first question If the same respondent lists a store that has alocation only in Los Angeles in the secondquestion, an error is indicated These answers are obviously inconsistent.
  • 15. TAKING ACTION WHEN RESPONSE ISOBVIOUSLY AN ERROR If solid evidence exists that points to the factthat the respondent simply failed to check LosAngeles then the response to the first question can bechanged A change should only be made whenmultiple pieces of evidence exist thatsome response is in error and when thelikely true response is obvious
  • 16. Many surveys use filter or skip questionsmethod A survey might involve different questions fora home owner than for someone who doesnot own a home if someone indicated that he or she did notown a home, yet responses for the homequestions are provided, a problem isindicated In cases like this, the editor should adjustthese answers by considering all answers tothe irrelevant questions as no response ornot applicable
  • 17. Editing Technology computer routines can check for inconsistenciesautomatically. Thus, for electronic questionnaires,rules can be entered which prevent inconsistentresponses from ever being stored in thefile used for data analysis. These rules should represent the conservativejudgment of a trained data analyst. Some online survey services can assist in providingthis service.In fact, the rules can even bepreprogrammed to prevent many inconsistent
  • 18. For ex:- if a person fills up data; a pop-up windowcan appear requiring the respondent to go back andfix an earlier incorrect response. Electronic questionnaires can also prevent arespondent from being directed to the wrong set ofquestions based on a screening question response.
  • 19. Editing for Completeness Item non response is the technical term foran unanswered question on an otherwisecomplete questionnaire.Missing data results from item nonresponsive. Inmany situations the decision rule is to do nothingwith the missing data and simply leave the itemblank.
  • 20. However, when the relationshipbetween two questions is important,such as that between a question aboutjob satisfaction and ones pay, theeditor may be tempted to insert a plugvalue. The decision rule may be to plug inan average or neutral value in eachinstance of missing data.
  • 21. Editing Questions Answered Outof Order Another task an editor may face isrearranging the answers given to open-endedquestions such asmay occur in a focus group interview. The respondent may have provided theanswer to a subsequent question in his or hercomments to an earlier open-ended question. Because the respondent already had clearlyidentified the answer, the interviewer may nothave asked the subsequent question.
  • 22. If the editor is asked to list answers to allquestions in a specific order, the editor may movecertain answers to the section related to theskipped question.
  • 23. Facilitating the Coding Process While all of the previously described editingactivities will help coders, several editingproceduresare designed specifically to s