Conversational norms
Also known as: Gricean maxims, Conversational maxims, Logic of conversation
Listeners assume everything mentioned is relevant, so asking changes what an answer means.
What it means
Conversational norms are the tacit rules of cooperative communication that lead listeners to assume a speaker is being informative, truthful, relevant, and clear, and so to infer far more than the literal words say. Grice framed them as maxims: say enough but not too much, say what you believe true, stay relevant, be orderly. The mechanism carries over from talk to questionnaires, instructions, and interfaces, because people presume that whatever has been put in front of them was put there for a reason. This generates effects that look like irrationality but are sound inference from a bad cue: irrelevant details get used because their mention implies relevance, earlier questions set the topic for later ones, and response scales are read as the researcher's picture of normal behavior. Many classic 'biases' shrink once the conversational cue is stripped out. It matters because no question is ever neutral—asking is itself a message.
The maxims and what they license
Grice separated what a sentence literally says from what it implicates. His four maxims — quantity (be as informative as required, no more), quality (say what you believe true), relation (be relevant), manner (be brief and orderly) — are less rules speakers obey than assumptions listeners make. Because I assume you are following them, I read your ordering, your emphasis, even your choice of words as carrying meaning. If you say 'some students passed,' I infer not all did: had all passed, the stronger 'all' was available and unused. The listener does this extra work and the speaker relies on it. Carried into a questionnaire, the respondent treats the researcher as a cooperative speaker and mines the question itself for evidence about what is being asked.
What the evidence shows
The effects are concrete. Schwarz, Strack and Mai (1991) found that asking about marital satisfaction just before general life satisfaction raised the correlation between them from about .32 to .67, because the earlier question made marriage the salient input — until a wording cue telling respondents not to double-count reversed the pattern into a contrast. Schwarz and colleagues (1985) found people report watching more television when a scale's top category is high, reading the range as the researcher's picture of normal. Hertwig and Gigerenzer (1999) argued that much of the Linda 'conjunction fallacy' is people interpreting 'probable' and 'bank teller' as a cooperative speaker intends, not committing a logic error. Across these cases, clarifying the pragmatics moves the numbers.
Where it shows up
It appears wherever a question is authored by someone the answerer treats as cooperative. In survey design, a presupposing item ('How often do you feel lonely?') plants its premise and inflates reports. Clinical intake forms and diagnostic interviews steer recall by naming symptoms whose mention implies relevance. Product and UX surveys quietly leak the team's own worry through the items they choose to include. Courtroom examination exploits the mechanism deliberately: a loaded verb narrows what a witness later 'remembers.' Market research and performance reviews carry the same freight. The newest arena is prompting large language models, which are trained on human dialogue and inherit the cooperative reading — an offhand aside in a prompt gets treated as load-bearing.
Limits and where the account is contested
How much of the classic bias literature is 'really' conversational is disputed. Gigerenzer's group argued that many framing and conjunction effects dissolve once the pragmatics are made clear; Kahneman and Tversky had argued that robust versions survive when the wording is tightened, so the errors are real rather than artifacts of misread instructions. The honest reading is partial: conversational inference shrinks some effects and eliminates a few, but does not abolish the underlying tendencies. The practical lesson is narrower and safe. Because respondents assume every element was placed on purpose, neutral measurement means stripping out cues you did not intend to send: avoid presupposition, randomize question order, offer an explicit 'no opinion' option, and pilot wording to hear what your questions quietly imply.
Examples
Asked how much television they watch, people report more hours when the scale tops out at 'over four hours' than when it tops out at 'over two and a half.'
An interviewer asks a candidate, 'any concerns about the commute?' She infers the commute must be a known problem and begins weighing an obstacle she had not considered at all.
A doctor asks during a routine check, 'have you been sleeping badly?' The patient infers sleep must bear on his symptoms and recalls worse nights than he would have volunteered.
A checkout survey asks, 'What frustrated you most about this purchase?' The wording presupposes frustration, so a satisfied buyer scans for a grievance and names a minor one that never actually bothered him.
A 360-review form asks a manager to 'describe a time this colleague missed a deadline.' Even a reliable colleague's reviewers go hunting for one lapse, because the request implies lateness is worth reporting.
First described in Grice (1975); Schwarz (1994) applied to judgment.
Key references
- Hertwig, R., & Gigerenzer, G. (1999). The 'conjunction fallacy' revisited: How intelligent inferences look like reasoning errors. Journal of Behavioral Decision Making, 12(4), 275-305. doi.org/10.1002/(SICI)1099-0771(199912)12:4<275::AID-BDM323>3.0.CO;2-M
- Schwarz, N. (1999). Self-reports: How the questions shape the answers. American Psychologist, 54(2), 93-105. doi.org/10.1037/0003-066X.54.2.93
- Schwarz, N. (1994). Judgment in a social context: Biases, shortcomings, and the logic of conversation. Advances in Experimental Social Psychology, 26, 123-162. doi.org/10.1016/S0065-2601(08)60153-7
- Schwarz, N., Strack, F., & Mai, H.-P. (1991). Assimilation and contrast effects in part-whole question sequences: A conversational logic analysis. Public Opinion Quarterly, 55(1), 3-23. doi.org/10.1086/269239
- Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Syntax and semantics, Vol. 3: Speech acts (pp. 41-58). Academic Press. brill.com/display/book/edcoll/9789004368811/BP000003.xml
- Schwarz, N., Hippler, H.-J., Deutsch, B., & Strack, F. (1985). Response scales: Effects of category range on reported behavior and comparative judgments. Public Opinion Quarterly, 49(3), 388-395. doi.org/10.1086/268936