# Conjoint (/docs/analyses/conjoint)



## When to use it [#when-to-use-it]

Use conjoint when each person chose one profile from a set, and each profile is a combination of attributes and levels. The result is a part-worth for each level, and a share for a profile you specify. Best-worst on a list of items, without attributes, is [MaxDiff](/docs/analyses/maxdiff).

## Running it in Tensr [#running-it-in-tensr]

<Steps>
  <Step>
    Analyze → Choice modelling → Conjoint. In chat: “Run conjoint on the choice column.”
  </Step>

  <Step>
    Design names each attribute and its levels, then exports a long file and a wide file. Tensr does not host the survey.
  </Step>

  <Step>
    Analyse marks the chosen profile. MNL is the default. Individual utilities fit one set of part-worths per person.
  </Step>
</Steps>

## Options [#options]

<TypeTable
  type="{
  attributes: {
    type: 'name and levels',
    description: 'Each attribute, and the levels written as a comma-separated list.',
  },
  choice: {
    type: 'column',
    description: 'The profile the respondent chose.',
  },
  estimate: {
    type: 'MNL | individual utilities',
    description: 'MNL is one model for the sample. Individual utilities are one model per person.',
    default: 'MNL',
  },
}"
/>

## Reading the output [#reading-the-output]

MNL reports a part-worth for each level. A higher part-worth is the level people picked more often, holding the other attributes as they were shown. Individual utilities are the same part-worths, one respondent at a time. The choice simulator turns a set of profiles into shares.

## Coming from SPSS [#coming-from-spss]

SPSS has no conjoint item in the base menus. Tensr’s item is Analyze → Choice modelling → Conjoint. The simulator is a separate item, Choice Simulator, under Techniques.

## Related [#related]

[MaxDiff](/docs/analyses/maxdiff) scores items from best and worst. [Van Westendorp and Gabor-Granger](/docs/analyses/van-westendorp) price one product.