SONDAINTEL / FOUNDATIONS
A survey starts with a question. A sound assessment needs design, mathematics, interpretation, and criteria for recognizing what the data cannot tell us.
Explore our foundations01 / METHODOLOGICAL FOUNDATIONS
These are conceptual lenses and families of analysis. Their inclusion does not mean they are all automatically applied to every study.
Kahneman distinguishes fast, intuitive processes from deliberate, effortful ones. This lens helps examine judgment, wording, and cognitive load.
Does the question require recall, calculation, or an immediate impression?
Review response effort, wording, and anchors. A stated response does not directly identify a mental system.
Game theory provides a framework for studying decisions whose outcomes also depend on other agents’ choices.
Which incentives and possible responses belong in the scenario?
State the agents, incentives, and scenario assumptions. The result is not a guaranteed prediction.
A critical assessment of hypotheses requires seeking evidence that could contradict them and stating the limits of conclusions.
What evidence would make us revise this interpretation?
Record the hypothesis and the evidence that would lead us to reject it. A single survey does not establish causation.
Looking only at cases that remain visible can exclude important experiences. The design must consider who was left out of the sample.
Are we only listening to people who stayed as customers?
Document who was invited, who responded, and which experiences may have been left out.
Studying variables together can help investigate relationships and profiles. Model selection depends on the design, variables, and available data.
Is this model appropriate for these data and this decision?
Define variables, usable sample size, and model diagnostic criteria before interpreting relationships.
Nassim Nicholas Taleb studies uncertainty, model error, and extreme events. His work encourages caution with extrapolation and conclusions that depend on fragile assumptions.
Does the interpretation hold when assumptions change?
Compare plausible scenarios and disclose when a conclusion depends heavily on an assumption.
Earlier questions can influence later answers. Reviewing the sequence and testing alternative orders helps investigate this effect in the instrument.
Is the preceding question suggesting what the person should consider?
Record question order and justify changes between rounds. Priming is not an automatic platform index.
Testing comprehension, answer choices, and the flow before collection helps identify problems. Instruments in another language need their own review and pretest.
Do people interpret this question as intended?
Review each translation with the intended audience before treating it as equivalent.
Richard Thaler studies economic decisions with psychological factors in view. His work helps frame hypotheses about preferences, self-control, and choice context.
Can the choice context help explain the gap between intention and behavior?
Distinguish stated intentions from observed behavior and test hypotheses before recommending an intervention.
02 / FROM FOUNDATIONS TO THE STUDY
To assess a conclusion, you need to understand how it was produced. This structure guides the new SondaIntel experience.
Record what we want to understand, for whom, and in which context.
Define questions, participation criteria, and how people will be invited.
Document processing, exclusions, chosen methods, and conditions of use.
Connect conclusions to evidence and state uncertainty, limitations, and required review.
Present the assessment in context and record hypotheses for action in a new round.
03 / TRANSPARENCY IN DELIVERY
One hundred voluntary responses do not automatically represent all customers. Results must include recruitment sources, selection, and participation details.
Conventional margins of error and confidence levels should not automatically be assigned to opt-in samples.
A relationship between variables can guide investigation. Demonstrating the effect of an action requires an appropriate design.
A more sophisticated model only makes sense when it better answers the question and meets its conditions of use.
A HERITAGE THAT EVOLVES
SAES, SAGS, historical instruments, and analyses are part of the collection guiding this rebuild. Modernization requires preserving versions and examining each method’s application in new contexts.
The app demo uses fictional data, counts, proportions, and percentage-point differences. It does not yet run the advanced models presented in these foundations.
Explore the demo assessment