Comparing Early Screeners for Accurate Identification of TRI First Grade Students Reading Below Grade Level

• In Event: A Multifaceted Lens: Perspectives and Impacts of an Effective Elementary Teacher Literacy Professional Development Program

Abstract

Early reading screeners, such as Dynamic Indicators of Basic Early Literacy Skills (DIBELS) and Woodcock Johnson (WJ) are often used to identify early grade students whose reading is below grade level (Hoffman et al., 2009). Each set of DIBELS and WJ subtests has its strength and weaknesses (Dombrowski et al., 2019). Yet, research suggests that DIBELS and WJ results are often highly correlated in the domains of alphabetics and decoding (Good et al., 2004). However, limited guidance exists on how to choose between the two tests for identifying students who could use supplemental reading support in early grades.

This study compares DIBELS Phoneme Segment Fluency (PSF) and Nonsense Word Fluency (NWF) with WJ IV Letter Word Identification (LW) and Word Attack (WA) subtests in their costs for and performance in identifying students who are struggling with alphabetics and decoding. The investigation was conducted in three rural schools and 23 children in six first grade classrooms. Table 1 shows the characteristics of these first graders.

All four subtests were administered in winter 2021–2022 in the context of the TRI study described in study 3. Fifteen students who were identified as below reading grade level by their teachers were further assessed following DIBELS and WJ screeners in the winter and spring. A larger battery was also administered, which included additional WJ tests (Passage Comprehension, PC; Spelling of Sounds; SOS); TOWRE Sight Word Efficiency (SWE) and Phonemic Decoding Efficiency (PDE); and CELF Recalling Sentences (RS) to represent a full range of early reading skills (alphabetics/decoding, reading comprehension, fluency, and oral language).

Table 2 compares aspects of the DIBELS and WJ screeners, including their costs, internal consistency, concurrent and divergent validity, predictive validity, and generalizability. DIBELS NWF results were highly correlated with WJ LW and WA, suggesting certain overlap between the subtests (r=.63 and .52). The two subtests within WJ were strongly correlated (r=.72) whereas DIBELS PSF and NWF are less highly correlated (r=.25), which is consistent with previous TRI findings that students who were better readers (i.e., high on NWF) may struggle to fully segment phonemes into individual sounds (i.e., low on PSF).
Overall, WJ screeners had more consistent associations with other baseline tests than DIBELS screeners. WJ screeners strongly predicted most of the post-program outcomes (except for CELF RS and TOWRE PDE), with or without controlling for DIBELS screeners. In contrast, after controlling for WJ scores, DIBELS’ predictive power became very limited, as indicated by bivariate and partial correlations. Despite differences in magnitude, the predictive power of WJ tests was consistent in direction between the two states in the study (NC and SC). In contrast, DIBELS NWF showed a strong positive association in NC but negative association in SC, suggesting its limited generalizability across contexts.

We concluded that compared to DIBELS, WJ subtests, albeit drawbacks involving costs of testing kits and packets, may be more effective for identifying students reading below grade level. Future study will further evaluate their performance with a larger sample and/or within a different context.

Authors

• Robin Wisniewski, RTI International
  Presenting Author

• Yihua Hong, RTI International

• Mary Bratsch-Hines, University of Florida

• Heather Aiken, University of North Carolina at Chapel Hill