A moderated focus group — focus

The moderator puts each question to the group; every participant answers (speaking order rotates across questions so nobody speaks first every round); participants see the discussion so far, as in a real group. The moderator closes with a synthesis of themes and disagreements.

Usage

focus_group(
  moderator,
  participants,
  topic,
  questions = NULL,
  n_questions = 3L,
  msg_mode = NULL,
  quiet = FALSE,
  ...
)

Arguments

moderator: An Agent running the group.
participants: A list of Agents.
topic: The study topic (context for everyone).
questions: Character vector of questions. If NULL, the moderator drafts n_questions itself, which is useful for piloting.
n_questions: Number of questions to draft when questions is NULL.
msg_mode: Message construction, "roleflip" (default) or "flat"; NULL uses getOption("LLMRagent.msg_mode"). See conversation().
quiet: FALSE prints the session live.
...: Passed to the underlying LLMR calls.

Value

An object of class agent_focus_group: a list with transcript (tibble: turn, question_id, speaker, text), questions, summary (the moderator's synthesis), and topic. as.data.frame() returns the transcript.

Examples

if (FALSE) { # \dontrun{
cfg <- LLMR::llm_config("groq", "openai/gpt-oss-20b", temperature = 0.9)
fg <- focus_group(
  moderator = agent("Moderator", cfg, persona = "A neutral focus-group moderator."),
  participants = list(
    agent("Maya", cfg, persona = "A 34-year-old nurse, prudent with money."),
    agent("Tom",  cfg, persona = "A 22-year-old gig worker, risk-tolerant."),
    agent("Ines", cfg, persona = "A 58-year-old teacher nearing retirement.")
  ),
  topic = "Attitudes toward a 4-day work week",
  questions = c("What would a 4-day week change in your daily life?",
                "What worries you about it?")
)
fg$summary
} # }