| DIMPACK - Nonparametric Dimensionality Analysis Package |
| Includes DIMTEST, DETECT, and HCA/CCPROX for Dichotomously Scored Data. For polytomously scored data, see POLY-DIMTEST and HCA/CCPROX. |
These latent dimensionality structure programs previously have been distributed as separate DOS programs. Due to ongoing demand for them, they are now being made available in one package for 32-bit Windows. Essentially no changes have been made in the three main programs, but the Windows interface facilitates selection of program options and provides ready access to help documentation. Printed documentation that is shipped with the CD includes installation instructions and copies of seven relevant journal articles that provide technical background for the methods used.
DIMTEST 2 – Latent Unidimensionality Assessment
DIMTEST 2 is a hypothesis testing procedure that assesses lack of latent unidimensionality for a dichotomously scored educational or psychological test. It does so by assessing the statistical significance of the possible dimensional distinctiveness between two specified subtests: the Assessment Subtest (AT) and the Partitioning Subtest (PT).
DIMTEST 2 operates in either a confirmatory mode, in which case it assesses the user-selected AT that a priori has been judged to be possibly dimensionally distinct from the user- selected PT; or in an exploratory mode, in which case, using cross-validation, it assesses unidimensionality by using a statistically selected AT that is potentially maximally dimensionally distinct from the rest of the test (i.e., PT). DIMTEST 2 is completely nonparametric, requiring neither parametric IRT modeling nor estimation of item response functions.
New Features in DIMTEST 2
DIMTEST 2 is a major upgrade to the earlier DIMTEST, allowing longer tests and larger samples to be analyzed:
- The upper limit on the number of items is 151 (increased from 50)
- The upper limit on the number of examinees is 12,000 (increased from 2,000). 12,000 will almost always provide the statistical power the user needs.
The earlier version of DIMTEST required that the test be divided into three subtests: AT, AT2, and PT, where in addition to the user selected subtest AT discussed above, a bias-correcting AT2 was also needed. Thus, the AT2 requirement resulted in "sacrificing" a portion of the PT subtest to correct for statistical bias. DIMTEST 2 eliminates AT2. Instead of using AT2, AT-based simulated data are used to correct for statistical bias. Research has shown this to result in a more powerful hypothesis-testing statistic that still adheres well to the nominal Type 1 error rate. Furthermore, the elimination of AT2 enables DIMTEST to be applied to shorter tests and to be used with AT's that have larger numbers of items.
One important exploratory capability provided by the old DIMTEST was an automatic procedure, called FAC, by which a linear factor analysis was applied to a training sample to find a single maximally dimensionally distinct AT, which was then tested by DIMTEST on a cross-validation sample. DIMTEST 2 replaces FAC with a new procedure called ATFIND that is more effective at finding a dimensionally distinct AT when multidimensionality holds. Importantly, ATFIND is based on the same basic statistical building blocks as DIMTEST, namely estimated item pair conditional covariances. In particular, ATFIND is based on a combination of DETECT and HCA/CCPROX. Research has shown ATFIND to result in increased DIMTEST statistical power (sometimes dramatically so) while still adhering well to the nominal Type 1 error rates.
The DIMTEST User Interface
DETECT – Dimensionality Analysis
One of the most common types of test formats is one in which the items are grouped around several distinct reading passages or underlying psychological or cognitive constructs—for example, algebra, geometry, and trigonometry in a mathematics test. Using estimated item pair conditional covariances, the DETECT procedure nonparametrically provides a detailed dimensional description of this type of test, which is said to exhibit approximate simple structure.
DETECT begins by using clusters obtained from the dimensionality-sensitive cluster analysis procedure HCA/CCPROX and then uses a customized genetic algorithm to efficiently search through all of the possible item cluster partitions to find the one that maximizes the DETECT statistic. In addition to quickly finding the dimensionally most appropriate grouping of the test’s items and thereby estimating the number of dominant dimensions present, its two resulting statistics provide a summary of the test’s total amount of multidimensionality (equivalently, a measure of the lack of fit by a unidimensional model) and a measure of the degree to which it displays approximate simple structure. This information is often useful from a statistical robustness perspective for those wishing to use IRT methodologies based on unidimensionality. DETECT presumes dichotomous item scoring.
DETECT can be run in either confirmatory or exploratory mode. In confirmatory mode, the user specifies a set of non-overlapping clusters for the test items and DETECT calculates its indices for that clustering. In exploratory mode, the data are split into a training sample and a cross-validation sample. The genetic algorithm searches for the best clusters on the training sample and then the cross-validation sample is used to calculate the DETECT indices.
The DETECT USER Interface
HCA/CCPROX – Hierarchical Cluster Analysis with Dimensionally-Sensitive Proximity Matrices
HCA/CCPROX performs a latent multidimensionality-sensitive hierarchical cluster analysis on dichotomously scored items. This nonparametric procedure is able to quickly cluster the items into progressively larger and larger relatively dimensionally homogeneous groups. It allows the user to examine the test’s dimensionality at a variety of agglomeration levels, ranging from which pairs of items are most closely dimensionally related, to which two-cluster solution best dimensionally summarize the entire test. If the items of a test exist in approximately dimensionally homogeneous item clusters, then there should exist a level in the hierarchy at which the clusters found by HCA/CCPROX will maximally agree with this approximate simple structure. Even if approximate simple structure does not hold, HCA/CCPROX will tend to find dimensionally disparate clusters, and hence its uses are not limited to tests demonstrating approximate simple structure. HCA/CCPROX is useful in tandem with a DIMTEST
assessment of dimensionality, because DIMTEST can test hypotheses concerning HCA/CCPROX selected clusters.
The CCPROX/HAC User Interface
DIMPACK Manual
Manual
DIFPACK’s manual includes instructions for installing the program, technical descriptions of each component program, a step-by-step guide to using the program, and copies of the following references to provide a theoretical background:
1. Stout, W., Habing, B., Douglas, J., Kim, H., Roussos, L., & Zhang, J. (1996). Conditional covariance-based nonparametric multidimensionionality assessment. Applied Psychological Measurement, 20, 331-354.
2. Stout, W. Froelich, A.G., & Gao, F. (2001). Using resampling methods to produce an imporoved DIMTEST procedure. In A. Boomsma, M.A.J. Duijn, & T.A.B. Snijders (Eds.), Essays on item response theory (pp. 357-376). New York: Springer-Verlag.
3. Stout, W. (1987). A nonparametric approach for assessing latent trait unidimensionality. Psychometrika, 52, 589-617.
4. Roussos, L., Stout, W., & Marden, J. (1998). Using new proximity measures with hierarchical cluster analysis to detect multidimensionality. Journal of Educational Measurement, 35, 1-30.
5. Zhang, J., & Stout, W. (1999). The theoretical DETECT index of dimensionality and its application to approximate simple structure. Psychometrika, 64, 213-249.
6. Roussos, L., & Ozbek, O. (2006). Formulation of the DETECT population parameter and evaluation of DETECT estimator bias. Journal of Educational Measurement, 43, 215-243.
7. Jang, E. & Roussos, L. (2007). An investigation into the dimensionality of TOEFL using conditional covariance-based nonparametric approach. Journal of Educational Measurement, 44, 1-21.
System Requirements
- Windows 98, Windows 2000, Windows XP, Windows Vista.
- .NET Framework 1.1
.NET Framework 1.1 does not come with Windows Vista, and may not come installed on all versions of Windows. To see if it's installed, go to 'Control Panel' and look for it in 'Add or Remove Programs' (XP) or 'Programs and Features' (Vista). Newer versions of .NET Framework are not backwards compatible with 1.1. If it's not available, we recommend downloading it and its service pack directly from the following Microsoft sites:
and installing them directly before installing and running DIMPACK.
|
Order Options and Details
|
 |
| |
Special Introductory Bundle
Purchase DIMPACK and DIFPACK at the same time, and receive a 25% discount:
Dichotomous/Polytomous Software Package
This package contains: DIMPACK (Windows) + POLY-DIMTEST (DOS) + HCA/CCPROX (DOS)
Upgrade From Older Programs
You can also upgrade to DIMPACK at a discount if you have previously purchased licenses for any or all of the following DOS programs:
- DIMTEST
- DETECT
- HCA/CCPROX
All 3 products: DIMTEST(DOS) and DETECT and HCA/CCPROX
Any 2 products in any of the following combinations:
- DIMTEST(DOS) and DETECT
- DIMTEST(DOS) and HCA/CCPROX
- HCA/CCPROX and DETECT
Any 1 product: DIMTEST(DOS) or DETECT or HCA/CCPROX
Products
Your Account
Your cart
