Scottish Inpatient Patient Experience Survey 2012 Volume 2: Technical Report

6 Data entry and fieldwork quality control

Introduction

6.1 This section of the report provides detail on the arrangements for data entry and coding and datafile processing.

6.2 In addition it discusses data checks, quality assurance processes and reliability and validity tests undertaken.

Data entry and coding

6.3 Contractors were responsible for data entry and quality assurance of data collected during the Inpatient Survey.

6.4 Prior to fieldwork commencing, a datafile template was sent by the Scottish Government to each contractor. The template was then used by the contractor as the format in which to send data to ISD. In advance of moving data into this format, each contractor used their own internal processes to undertake data entry and coding of the responses.

6.5 With the exception of answers which allowed respondents to provide written comments, the Inpatient Survey contained predefined answer options in questions for respondents to tick as appropriate. Answer options corresponded to codes which were entered into the datafile to indicate which answer had been selected by respondents for each question. A copy of the questionnaire is in Appendix A.

6.6 For questions where a written comment could be provided, guidance was provided to contractors on how to code these answers (see Appendix F). Questions which contained options for respondents to provide a written answer and approaches to coding these answers are summarised below:

• Q30a - Q30f "If there is anything else you would like to tell us about your experience in hospital please write your comment below: (a) Your experience of admission to hospital, (b) Your experience of the hospital and ward, (c) Your experience of care and treatment, (d) Your experience of staff, (e) Your experience of leaving hospital; (f) Your experience of care and support services after leaving hospital.

6.7 In addition to recording predefined codes and coding of comments, contractors were required to capture cases where a patient had ticked two or more answers on any question in the survey and it was unclear what the correct answer was.

6.8 Contractors were also required to record responses where a patient answered a routing question and then continued to answer subsequent questions contrary to the routing instructions within the original routing question.

Datafile processing

6.9 Once data had been collected and input into the required datafile format, contractors were able to establish a raw final datafile containing all survey information including double answers and routing errors.

6.10 In addition to producing a raw datafile containing all information collected during fieldwork, contractors were also required to establish a cleaned version of the datafile by undertaking data cleaning. Data cleaning involved the following activities:

• Removal of double answers and marking of these as 'missing' as the intention of the respondent could not be ascertained;
• Cleaning of routing errors where routing instructions had been incorrectly followed and patients incorrectly provided answers; and
• Removal of any information recorded within free text comments which could be viewed as disclosive to respondents, other patients or staff in order to comply with the Data Protection Act.

6.11 Once a raw and clean datafile had been developed by each contractor, files were transferred to ISD where final merging of contractor files took place prior to final analysis of data.

Quality assurance processes

6.12 Throughout the data entry process and prior to datafiles being sent to ISD, contractors were required to undertake data checks on data entered into final datafiles.

6.13 Checks were made by two team members checking every tenth record to ensure that what was contained within the respondent's completed version of the survey was checked against what had been entered within the data files.

Survey routing errors

6.14 Routing errors were cleaned within the final dataset based on questions being applicable to respondents. For example, where a patient answered a routing question and then continued to answer subsequent questions contrary to the routing instructions within the original routing question.