Interviews

1
Interviews

• Alfred Kobsa
• University of California, Irvine

2
When to chose an interview

• Scope of problem not fully clear, unclear what to
  ask
• Questions abstract, possibly not well understood,
  open-ended
• Mailing lists not available
• Rapid answers not needed
• Specific location needed
• Recording of responses difficult
• Lots of information / holistic view from a few
  users needed (vs. little information from many
  users)
• Narrow geographic distribution of respondents
• Anonymity not required
• Topics difficult to talk about

3
Problems to consider

• Honesty (? check consistency, challenge answers,
  discard data, blacklist potentially dishonest
  user)
• Self-incrimination (? emphasize harmlessness,
  anonymity, impunity)
• Social desirability bias (? phrase questions
  carefully, use several questions for same
  concept, go for facts and not habits, attitudes
  or intentions)
• Self-image (? see above)
• Prestige response bias (? go for facts)

4
Steps

• Identify the objectives of the study (in writing)
• Select the type of interview (structured,
  unstructured, semi-structured, contextualized,
  via phone vs. in person)
• Decide how you will analyze the data (analysis
  software, statistical tests if any)
• Formulate questions (mind double-barreled, double
  negatives, leading or loaded questions, and
  questions with self-image and social desirability
  bias)
• Write questions down
• Test your questions
• Audiotape answers or use note-taker
• Decide on awards for interviewees (if any)