Are There Notable Insights Regarding the 2009 PeNSE Survey from Complex Mokken Analysis?
Are There Notable Insights Regarding the 2009 PeNSE Survey from Complex Mokken Analysis?
-
DOI: https://doi.org/10.22533/at.ed.394122608017
-
Palavras-chave: Complex Mokken, amostras complexas, nível socioeconômico, pesquisa amostral probabilística, avaliação educacional, estatística aplicada, Psicometria
-
Keywords: Nonparametric IRT, Complex Non-Parametric Mokken Scale Analysis, Complex Sample, Complex Sampling Design, Mokken’s Scalability Coefficients, National Survey of School Health, Status Socioeconomic, Applied Statistics.
-
Abstract: The construction of Mokken scales considering complex samples was unexplored in academic literature prior to 2012. In other words, complex samples were treated as samples obtained through simple random sampling with replacement (SRSWR). This means that the Popularities (Pi), the Scalability Coefficients Hij, Hi and H, and their standard errors were estimated, for instance, ignoring a stratified multistage cluster sampling design. Consequently, this decision can lead to erroneous conclusions. This work illustrates the adequate use of a complex probabilistic sampling design in the Mokken scaling, using Marcia Andrade’s Complex Mokken (2012) statistical software and data from a realistic educational complex survey.
- Marcia Andrade