Dissertation Defenses

Doctoral Dissertation Announcement

---

Candidate: Christina M. Terenzi

Degree of: Doctor of Philosophy

Department: Psychology

Title: Curriculum-Based Measures in Writing: A School-Based Evaluation of Predictive Validity

Committee:
Dr. Galen Alessi, Chair
Dr. Mary Z. Anderson
Dr. John Austin
Dr. Ruth Ervin
Dr. Margaret McGlinchey

Date: Friday, May 15, 2009 11:30 a.m. - 1:30 p.m.
1509 Wood Hall

Abstract:
Recent research in the area of Curriculum-Based Measures (CBM) in writing has shown that traditionally used metrics, such as total words written and total words correct, may not be the best tools for measuring writing performance, for both secondary and elementary aged children (e.g., Tindal & Parker, 1989a; Watkinson & Lee, 1992; Gansle, Noell, VanDerHeyden, Naquin, & Slider, 2002). Evidence suggests that more advanced measures, such as production-independent measures (e.g., percentage of correct word sequences) may be stronger predictors of student skill level in the area of writing. The present study replicates portions of a recent seminal study and investigates the predictive validity of CBM in the area of writing for the Michigan Educational Assessment Program (MEAP), writing and ELA assessments and the Iowa Test of Basic Skills (ITBS) reading assessment. Participants include 700 fourth grade students in a Midwestern urban school district who completed a three minute writing probe, which was scored for 20 independent variables. Dependent variables include assessments administered in the same year and in years following the administration of the writing probes. Correlations are calculated between each of the independent and dependent variables. Interscorer reliability is calculated, with all variables above .80. Alternate form reliability (n=199) is above .40 for all but two independent variables. Stepwise multiple regressions are run with two sets of independent variables with each of five dependent variables. The independent variables that appear to be the most promising indices for predicting performance on dependent measures include percentage of correct word sequences, correct punctuation marks, and words in complete sentences. Implication of analyses, limitations, and implications for future practice and research are discussed.