L’enquête Feti’i e fenua (Enquête Famille, territoire et relations intergénérationnelles en Polynésie française) : Apurement et imputation des données
Collection : Documents de travail
n° 262, 2021, 60 pages
This working paper presents the post-collection statistical operations relating mainly to the cleaning of the data and imputation of missing data. Relatively ignored and rarely documented in the social science literature, these essential operations in the production of quantitative survey data condition the Production and Research File made available to the scientific community for the analysis and promotion of the survey results. During the data preparation operations of the first Family survey collected on the territory of French Polynesia (Feti’i e fenua, Ined-Ispf 2020), this working paper presents the codification of open questions (locations and professions), the data cleaning which is undertaken to eliminate internal inconsistencies between the different modules and the imputation which compensates for non-response. The Feti’i e fenua survey, inspired by the Family and Housing Survey associated to the census carried out in metropolitan France, was collected in 31 islands in the territory from 5,139 respondents with a remarkable response rate of 86%. Users of large quantitative data sources rarely have the details of the choices that prevailed when creating the file made available for analysis. However, the robustness of the analyses depends as much on the quality of the data collection as on the processing necessary to standardize all the responses.
