General CAPI guidance

General checks

  • Is all the content included and in the right order?
  • Is the nesting correct?
  • Are the labels correct?
  • Have soft and hard checks have been removed?
    • Note: checks which lead to something that the respondent would have seen on the computer when completing the questionnaire are kept in.
  • Check the Interviewee, particularly for instances where it needs to be changed to Interviewer or Proxy.  
  • If there are missing spaces in the questionnaire (e.g., if a question literal says “these next questionsare about”), do not add the space in (e.g., between “questions” and “are).
  • Have the question grids been parsed?
  • "SHOW CARD" text us not included in MCS, but is included in BCS, Next Steps, NCDS, and USoc (main). Discuss with the team if SHOW CARD text appears in other questionnaires.

Questions

  • Remove "brackets" questions (see Example 1) - these are usually auto-generated and are a way to narrow down/get to the answer to a question.
  • Has any text which includes key contextual information been concatenated to the question literal (see Example 3 on the Question Items page)?
  • Watch for Month and Year questions that need text, and sometimes instructions, from the Month question concatenated onto the Year question (the order of the questions could be Year and then Month, and the text needs to be concatenated to Month question).
  • Check for text in question literals which can be deleted from the question literal and added as a statement before the question (e.g., a statement such as "The next set of questions are about your health", which applies to a set of questions rather than just one). There doesn't need to be a hard return or line break to split the text off, but the decision on whether or not to split the text off should be obvious. If it isn't, this suggests it should not be split off and so it should be kept in the question literal.
  • Check for instruction text that should be in the question literal. For example, "Please enter month"/"Please enter year" which needs to be in the question literals to make them unique and give them context.
  • Have all the instructions been grouped together if they appear above and below the question literal in the questionnaire?
  • If instruction-type text is on the same line as the question literal, it is added to the question literal. Keep in mind that other things may play into this decision, such as if the text has "INSTRUCTION:" before it.
  • Check the footnotes to ensure there isn't any important information that belongs in the question literal. For example, a condition that controls what question text is displayed to the respondent.

Example 1 (bcs_16_biomed)

Code lists/response domains

  • Are the code list values and categories correct?
  • Are the code list categories in the right order?
  • Is the cardinality correct?
  • Check that Don’t know, Refused and Missing have been added as categories to code lists where these responses are referred to in conditions.
  • We do not fix code lists (i.e., duplicates or labelling that does not follow protocols). However, labelling protocol is followed when inputting questionnaires from scratch (see the Construct a label page).
  • Change the response domains for questions asking about ‘Other’ from Generic text to Other.
  • When a range is specified in the answer to a question, the response domain should be entered as "Range: x-x"

Conditions

  • Is the condition text and logic correct?
  • Are the conditions labelled after the right question/construct? See here for more detailed information on labelling conditions.
    • When multiple conditions reference the same question, are the roman numerals correct? 
  • Are the || and && symbols included the condition logic?
  • Are the constructs inside the condition correctly located in the True/Else branch?
  • Check for condition text that is repeated across conditions - the repeated text can be deleted and the remaining portion of the condition text can be included. 
  • Make sure redundant conditions are deleted (i.e., ones that repeat, so they don’t need to be in twice).
  • Watch for conditions arising from derived variables, and check whether the variables that make up the derived variables come from questions in current questionnaire. If so, add logic to the conditions using the questions in the current questionnaire. Otherwise, leave the condition logic blank.

Loops

  • Are they correctly labelled?
  • Is loop variable appropriate?
  • Are the Min, Max, and loop while fields correctly entered?
  • If text is added to a question that is in a loop in order for it to make sense, make sure square brackets have been used (e.g., “what is your [child number] child’s date of birth?”, for a question which is included in a loop which asks about each child's date of birth).