RUMM2030 Plus provides a series of ten Interpreting RUMM2030 monographs — free of charge.

These documents contain detailed instruction material to assist with an understanding of the logic behind the conduct of item analyses using RUMM2030, together with an in depth coverage of the different fit statistics and estimation algorithms involved.

  1. Dichotomous data provides an extensive introduction to the fundamentals of the Rasch model and the conduct of a typical Rasch analysis involving dichotomous data.
  2. Polytomous data extends the scope of the Rasch model for item analyses involving multiple category data and the role of item thresholds in assessing anomalies associated with disordered threshold estimates.
  3. Estimation and Statistical Techniques presents details on the pair-wise and principal component [PC] re-parameterisation item estimations, the advantages of the PC method and especially its role in estimations involving missing data and null categories. Also covers the Information Function, the Likelihood-ratio test, the residual statistic distribution and the function of the item-residual correlation matrix in assessing anomalies in the data.
  4. Multidimensionality and Subtests in RUMM examines the role of subscales of a scale, the structure of Coefficient Alpha and the correlation among subscales, and the importance of the proportion of non-error variance common to the subscales.
  5. Quantifying Response Dependence in RUMM addresses the issue of local independence and strategies for identifying conditions of response dependence.
  6. Issues in Measurement presents an overview of the approaches to measures and the central role of a unit in measurement; the Guttman structure; test analysis theories and the notions of reliability and validity; and an historical perspective of Georg Rasch.
  7. Facet Design and Implementation discusses the notion and strategy behind an item-facet design, together with illustrations of the item-facet structure in practice within the context of the Rasch model and the Rasch paradigm .
  8. Linking Analysis in RUMM addresses the situation where persons are drawn from a large range of locations and it is not practical to target any one person across the large range of item thresholds required to define the variable created to measure that person. A special structure is required whereby the individual item subsets, or block, are linked. The aim of this document is to describe the logic of the link design strategy involving response records containing random missing data in the precence of structural missing data.
  9. Random Error Distribution of Ordinal Assessments examines the rationales for theoretical random error distributions of independent replicated measurements, referred to simply as replications, all had in common the principle of minimizing the error relative to the estimate of the true value. These properties of errors are satisfied by unimodal, single-peaked distributions referred to here as strictly unimodal [SU]. The concept of smooth, SU distributions leads to the idea of the requirement of a smooth distribution of ordinal assessments which is central to the context of thresholds in Rasch Measurement Theory.