Nuove prospettive nella rilevazione di atteggiamenti, opinioni e abilità: il formato di risposta come ipotesi sui dati
The paper starts from the assumption that response format is a hypothesis on the operation of categories, which can be «falsified» by the behaviour of subjects. By using Rasch model for ordered categories, and mainly the key concept of threshold as boundary between two contiguous categories, it is t...
Enregistré dans:
| Udgivet i: | URI:https://journals.openedition.org/qds, |
|---|---|
| Hovedforfatter: | |
| Format: | Article ou chapitre numérique |
| Sprog: | Italien |
| Udgivet: |
Quaderni di Sociologia
2016
|
| Online adgang: | Accès Université d'Orléans et IFPM Accès Université d'Orléans et IFPM |
| Summary: | The paper starts from the assumption that response format is a hypothesis on the operation of categories, which can be «falsified» by the behaviour of subjects. By using Rasch model for ordered categories, and mainly the key concept of threshold as boundary between two contiguous categories, it is treated the case in which the difference between hypothesized, intended order and empirical order of categories can be attributed to a lack of discrimination of one or more thresholds. An example based on Rosenberg’s self-esteem scale shows the strategy to identify and to combine non discriminating categories. Two additional rules are then suggested to improve the strategy: the use of short Likert-type response formats, preferably with no more than five semantically anchored categories, and checks across groups. |
|---|