Laserfiche WebLink
Task 11: Implement Survey <br /> Interviewing for the survey will be conducted from a centralized and fully <br /> monitored call center (either Davis Research in Calabasas or Pacific Market <br /> Research near Seattle) under the ongoing oversight of full-time, non-interviewing <br /> supervisors. Every interview will be checked as it is completed for accuracy, <br /> clarity, and completeness, and any problem areas will be corrected <br /> immediately. In the event a problem cannot be resolved by recall, the <br /> respondent will be called back for clarification or amplification. Cases with <br /> missing data will not be accepted. <br /> All survey calls will be placed during regular business hours unless a potential <br /> respondent requests otherwise. In order to ensure a representative result, up to <br /> six attempts will be made to reach an eligible respondent at each listing in the <br /> sample. <br /> Task 12: Code and Analyze Data <br /> Coding of the survey's closed-ended questions will be accomplished by the <br /> interviewers as they conduct the interviews. Coding of what we have assumed <br /> for budgeting purposes will be two open-ended questions will then be <br /> undertaken in three stages. <br /> First, one of our coding specialists will select a random sample of ten percent of <br /> the completed interviews and review all of the open-ended responses in order <br /> to develop a codebook for each question. These codebooks will then be used <br /> to code the open-ended questions, with any responses that fail to conform to <br /> the coding scheme being set aside for the possible addition of new codes. <br /> Both the original codebooks and all added codes will be approved by the <br /> President. As a final check the integrity of the coding as a whole, the Vice <br /> President and Data Analysis Manager will review a ten percent random sample <br /> of all of the coded interviews. <br /> The resulting data will be downloaded into SPSS. Preliminary tabulations will then <br /> be run using SPSS for Windows and manually reviewed for accuracy. In the <br /> event any errors are detected during this cleaning process, they will be <br /> corrected by actually calling the respondent back. When the data are <br /> completely clean, final tabulations and other analyses (means for scaled <br /> 10IPage <br />