| To Log or Not to Log: Bootstrap as an Alternative to Parametric Estimation of Moderation Effects in the Presence of Skewed Dependent Variables |
| | Russell, Craig J. | U. of Oklahoma | cruss@ou.edu | (405) - 325-2458 |
| | Dean, Michelle A. | U. of North Texas | deanm@unt.edu | (940)-565-4487 |
| | When gross deviations from parametric assumptions are observed, conventional data transformations (e.g., log) are often applied with little regard for substantive theoretical implications. One such transformation involves taking log transformations of positively skewed dependent variables. Log transforms were shown to severely decrease expected moderator effect sizes using moderated regression procedures in a simulation. Bootstrap procedures were unaffected by violation of parametric assumptions using untransformed data. Bootstrap results also yield more accurate and interpretable estimates of the parameter of interest, b3 in the equation Y = b0 + b1X1 + b2X2 + b3X1X2 + e. Implications are drawn for applied psychological and management research. |
| | Keywords: Bootstrap,; Compensation,; Data Transforms |
| Finding Patterns in Sequences: Applying Sequence Comparison Techniques to Study Behavior Processes |
| | Fichman, Mark | Carnegie Mellon U. | mf4f@cmu.edu | 412-268-3699 |
| | Studying sequences of behaviors is one way to better understand the
mechanisms which generate behavior. Techniques for sequence comparison
and analysis developed in computational biology can have broad
application to the study of behavioral sequence data. We illustrate
the application of dynamic programming methods from computational
biology for global and local alignment of sequences. Such techniques
can allow us to compare sequences to determine if sequences are
similar. Using data on two person interaction, the application of
these techniques is illustrated. The value of such techniques and
critical decisions that need to be made in applying such techniques are
illustrated and discussed. Such techniques have great promise in
helping understand behavior processes and their ordering
|
| | Keywords: measurement,; sequence comparison,; methods |
| The Group Dynamics Q-Sort in Organizational Research: A New Method for Studying Familiar Problems |
| | Peterson, Randall S. | Cornell U. | randall.peterson@cornell.edu | (607) 255-2997 |
| | Owens, Pamela D. | U. of California, Berkeley | pamowens@socrates.berkeley.edu | (510) 642-5292 |
| | Martorana, Paul V. | Northwestern U. | paulm@nwu.edu | (847) 467-1195 |
| | This paper unveils a new research methodology for the study of group decision making in organizations -- the group dynamics q-sort (GDQ). The GDQ is a 100-item instrument designed to study group process across a wide variety of situations and using a wide variety of data sources. The method combines the descriptive richness of a qualitative approach with the rigor of a quantitative approach by creating a common data language to describe group process across groups, observers, and time. This paper introduces the GDQ in four sections, 1) the development of the group dynamics q-sort, 2) comparisons of the GDQ with other traditional research methods, 3) a demonstration of the method for studying dynamic group processes over time, and 4) summary and other potential applications of the method. |
| | Keywords: Leadership,; Group Decision-Making,; Q-Sort |