Catilyst · Voxworks
Research notes
Notes on telephone survey methodology, response rates, survey modes and AI voice interviewing, with sources linked.
IVR surveys vs AI CATI: why conversation beats the keypad
IVR made telephone surveys cheap by taking the conversation out. AI voice interviewing puts it back at the same scale, and with it the completions, questions and answers a keypad could never handle.
CATI vs CAWI vs CAPI: how survey mode shapes the data
Every survey mode trades one source of error for another. The useful question is which trade a study is making, and whether it can measure the cost.
How political polls are fielded in Australia and New Zealand
Landline samples gave way to mobiles and online panels, and a polling failure in 2019 gave Australia a disclosure code. Small geographies remain the hardest test of any method.
Phone survey questions: writing for the ear
A telephone respondent hears each question once, at the reader’s pace, with nothing to look back at. Questions written for the page need rewriting for the ear, and an AI interviewer needs the rules written down.
AI voice interviewers for surveys: what the evidence shows
Early studies show that AI interviewers can run a survey. Whether one can run yours depends on who takes part, how answers are captured and what a fair comparison measures.
Survey response rates and nonresponse bias, explained
A response rate says how much of the sample took part. Nonresponse bias depends on who is missing and whether they differ on what the study measures, so the two need separate evidence.
What is CATI? Computer-assisted telephone interviewing
CATI puts a programmed questionnaire behind a telephone interview. How it works, what the software does, where it falls short, and the design decisions that determine data quality.
Running telephone studies?
catilyst is fielding telephone studies with research teams now. Tell us what you are planning.