Todd M. Davis

Simplifying Ranking Tasks in Survey Research: A Method and Example

In many situations in survey research ranking objects or choices is desired. Ranking often presents many difficulties for respondents so data are incomplete. This paper demonstrates that use of paired comparisons may overcome many such difficulties. The paper also describes the simple computations involved in its use.

On the Stability and Internal Consistency of the Self-Concept and Motivation Inventory: Preschool/Kindergarten Form

The purpose of this present study was to determine the internal consistency coefficients (alpha) and the stability (test-retest) of each of the scales of The Self-concept and Motivation Inventory: What Face Would You Wear? Preschool/Kindergarten Form. Such information is critical to either clinical interpretations or program evaluations whose outcomes are based on SCAMIN test scores. Present results suggest caution against both the clinical application of this instruments scales and their use in longitudinal research in its present form.

The Factor Structure and Internal Consistency of the Self-Concept and Motivation Inventory: What Face would you Wear? Preschool/Kindergarten Form

The Self-Concept and Motivation Inventory: What Face Would You Wear? Preschool/Kindergarten Form has the potential to be of use in measuring the academic self-concept of preschoolers. This study sought to determine if the theoretical, three-factor structure of the SCAMIN could be recovered in a state-wide sample of kindergarten students. In 1986, 5742 kindergarteners from Tennessee schools participating in the Reduced Pupil/Teacher Ratio Project were given the Preschool/Kindergarten form of the SCAMIN. The sample was drawn from 42 school districts in all geographic regions of Tennessee and represented a mix of inner-city, urban, rural, and suburban schools. The intercorrelation matrix of the 24 SCAMIN items was analyzed by the principal components method and extracted factors were rotated to a varimax solution. An initial analysis suggested the presence of an atheoretical factor, and two items which appeared to be unrelated to any obtained factor. Subsequent refactoring of the SCAMIN resulted in the successful recovery of the expected three factor structure. Cronbachs alpha, for the scales Academic Self-Concept; Achievement Investment; and Achievement Needs were .666, .599, and .598, respectively. The results of this factor analysis of the Early Preschool/Kindergarten Form of the SCAMIN suggest both strengths and weaknesses in the factorial composition of the 24 SCAMIN items. Users of the SCAMIN are cautioned against a naive application of this instruments scales.

Detecting subgroups in survey research

A problem often not detected in the interpretation of survey research is the potential interaction between subgroups within the sample and aspects of the survey. Potentially interesting interactions are commonly obscured when data are analyzed using descriptive and univariate statistical procedures. This paper suggests the use of cluster analysis as a tool for interpretation of data, particularly when such data take the form of coded categories. An example of the analysis of two data sets with known properties, one random and the other contrived, is presented to illustrate the application of cluster procedures to survey research data.