Research Seminar
Bootstrap estimates of parameter uncertainty in HICLAS analysisJoeri Hofmans & Eva CeulemansKU Leuven | |
| Abstract: | The hierarchical classes (HICLAS) model, proposed by De Boeck and Rosenberg (1988), is an order-preserving Boolean decomposition model for two-way two-mode binary data. In particular, HICLAS reduces the two modes of the data to a few binary components. Furthermore, HICLAS includes hierarchically organized classifications of the elements of the two modes. In practice, HICLAS analysis is often applied to data that are obtained from a sample from a larger population, assuming that the results for the sample may be generalized to the population from which the sample was drawn. However, HICLAS analysis does not provide any estimates of parameter uncertainty due to sampling fluctuations. In this study, it will be discussed how such estimates can be obtained by means of a bootstrap procedure. The performance of the proposed bootstrap procedure was evaluated by means of a simulation study in which the number of components, the sample size, and the amount of noise on the data were manipulated. The results indicate that parameter uncertainty decreases with larger sample size and smaller amounts of noise. Moreover, the bootstrap procedure succeeds in differentiating between stable model aspects that will also be found back in other samples, and less stable aspects resulting from mere sampling fluctuations. |
| Date: | Tue Nov 18, 12:15 pm - 1:15 pm |
| Place: | room 00.60 (Department of Psychology, Tiensestraat 102, 3000 Leuven) |