To improve reliability of results, depending on the specific project requirements it is recommended performing one or more of the following data calibration approaches:
- Sample weighting: required if the sample is not representative of the Universe due, for example, to a boosted or to a quota sample. Weighting improves representativeness.
- Current market share adjustment: survey data are weighted so that there is a match between current sample shares and current market shares.
- Approach-based calibration: some models require adjustment as they are known to produce biased results.
- Adjustment for overstatement: respondents tend to over-state their preference to a new product, so Juster calibration is recommended to obtain more accurate shares.
The simulators distributed with R-sw Conjoint allow to apply the above calibrations.