Discussion
The high false positive rate (54%; 1 - positive predictive power) of the January ORF Level/May ORF DD classification is similar to other efforts in first grade (e.g., O'Connor & Jenkins, 1999) and remains a vexing problem not solved by the use of LSF measures and attempts to assess LSF growth. It would seem that, in addition to evaluating the power of word-level tasks (both CBM and published measures), a second screening step that incorporated a quick hitting intervention may be one way to reduce the number of false positive cases. It must be kept in mind that a dynamic assessment task used as part of a first grade screen in the O'Connor and Jenkins (1999) study reduced false positive cases but the rate still may be considered high at 47.4%. However, they used a very strict criterion to define reading disability (-1.4 SD) that will necessarily increase the number of false positive cases.
To say that more work needs to be done is both true and unsatisfying to those who wish to implement now. Jim Gallagher once described the frustration that policy makers experience with researchers. He noted that researchers have to be right 95 times out of 100 to make a recommendation whereas policy makers would be ecstatic with a recommendation that holds a 50-50 chance of success. If we accept as reasonable a 50% false positive rate given high sensitivity, it means we are comfortable identifying twice the number of children for secondary interventions than who actually need them. If this is the case then the fall first grade procedures identified by O'Connor and Jenkins (1999) might work (dynamic segmentation, sound repetition, rapid letter naming), as might a January screen on Oral Reading Fluency with a cutoff value of 6 wpm (our 25th percentile value). There are two obvious difficulties with these recommendations that must be understood: The O'Connor and Jenkins battery incorporated a dynamic assessment task, which added 30-35 min to the screen, and the January ORF screen, while efficient, may be viewed by some as too late and as ignoring the importance of growth.
With respect to research there are several avenues that may yield more insight on how to identify children for secondary interventions. The only fairly firm conclusion to be drawn from the classification data I presented is that Letter Sound Fluency is likely not a good variable to pursue in first grade. Thus, as mentioned, comparing word-level tasks across a variety of measurement time points would address questions of efficiency and growth. Before accepting a 50% false positive rate, these approaches to screening need to be evaluated.
Given the number of studies devoted to early identification and their disappointing accuracy results, the conclusion that "growth matters" seems inescapable. After all, that is the definition of a false positive case: the child "grew" despite expectations to the contrary. Traditionally, screening has been conceptualized as a two-part process: screen and outcome defined by individual differences. I recommend we view screening as a three-legged stool that incorporates not only individual differences at the screen and outcome but also development conceptualized as rate of learning. Continuing to search for the right combination of child characteristic measures without attention to learning is likely a bankrupt strategy.
Previous Page | Next Page
(Description of Study) | (References)
