My main research is on Profile Analysis. Specifically, I am interested in investigating core response patterns of individual participants, utilizing Profile Analysis via Multidimensional Scaling (PAMS) or Profile Analysis via Principal Component Analysis (PAPCA). PAMS and PAPCA aim to (i) identify latent profile patterns in a population and (ii) interpret each individual in a population in terms of these latent profile patterns. PAMS and PAPCA estimate matching statistics between each individual?s observed score profile (that is an array of subtest scores in a given test) and the identified latent profiles. The matching statistics tell us the degree of similarity between the pattern of the observed score profile and the latent profile patterns; these matching statistics are converted into correlation values to enhance interpretation. Each individual is assigned the same number of matching statistics as the number of latent profiles.
To illustrate the power of PAMS or PAPCA, let us examine a typical case. Assume that in some population, two latent profiles are identified: the first profile is labeled as Verbal Skills and the second one as Quantitative Skills. One individual is observed to have a substantial weight on the first latent profile and trivial weight on the other latent profile. From this, we can conclude that the pattern of the person?s profile will be similar to that of the first latent profile. Secondly, we can interpret the person as someone who is strong for Verbal Skills, rather than Quantitative skills. However, if another person has a substantial weight on the second latent profile and trivial weight on the first, then the profile pattern of the person will be similar to the second latent profile pattern and we consider that person as someone who is strong for Quantitative Skills over Verbal Skills. Lastly, if one has substantial weights on both latent profiles, then we expect that the individual is strong for both Verbal and Quantitative Skills. PAMS and PAPCA were originally designed for exploratory analysis, but they can be applied to confirmatory approaches (utilizing Structural Equation Modeling) and longitudinal analyses. In that sense, my current research interest is in Takane?s Constrained Principal Component Analysis and Hwang & Takanes Generalized Structured Component Analysis.
In addition, I am also interested in Correspondence Analysis and have developed the course for correspondence analysis. Someone who is interested in PAMS or PAPCA research may request papers via email: firstname.lastname@example.org or can download them from my electronic reprints page.