Highlights and Important Distinctions from Research on Screening

How At Risk?

Screening approaches also differ in the degree of risk they specify. For example, screens can classify individual as at risk for not meeting a standard (their screening score does not strongly presage meeting a later criterion), or as very at risk for not meeting standard. The DIBELS system distinguishes between students at risk and students very at risk for unsatisfactory reading outcomes. As Table 1 shows, DIBELS uses different measures to identify risk status at different grades (e.g., ISF at mid-kindergarten; PSF at spring of kindergarten). Students who achieve a "benchmark" score on the various screens are very likely to attain the future criterion standard. Students performing below the benchmarks are not assured of achieving the future criterion of satisfactory reading; by implication, they are at risk. In addition, DIBELS specifies performance levels below which students are very unlikely to achieve the future standard. The latter group could be considered "very-at risk" because their likelihood of attaining the future criterion is slim.

Table 1. Percent of Students Placed in Risk Categories from DIBELS and CMB-ORF (from Good et al. 2001)

Table 1 illustrates the different proportion of students that fell into the two risk levels for the various DIBELS measures, as reported by Good et al. (2001). At mid-kindergarten, DIBELS ISF identified 47% of students as at risk. In contrast, only 7% qualified as very at risk (i.e., fell below the "needs intensive service score of 10 ISF). Across DIBELS measures, on average 51% of the sample were classified as at risk vs. an average of 7% as very at risk. Also noteworthy was the considerable variability in the percent of students classified as at risk (33-71%) or very at risk (3-20%) by the different measures.

In summary, depending on the goal of screening (which dictates the emphasis given to particular levels of risk and criterion performance), students can be variously classified: (1) at risk, or (2) very at risk for (3) unsatisfactory, or (4) very unsatisfactory reading outcomes. It is critical to distinguish between these different emphases when comparing screening approaches. The different conventions used for selecting criterion tests and performance levels make it difficult to compare the results from different screening measures .

Measurement Content of Screening Tests

Reading development follows a series of predictable stages (Chall, 1996; Ehri, 1992; 1998) with successive stages emphasizing different skills. Thus, the specific traits that forecast later reading success vary according to children's reading development. For screening measures to be useful they must be sensitive to the skills that pertain at successive stages and grade-levels. In kindergarten, phonemic awareness, letter knowledge, grapho-phonemic knowledge (letter-sound knowledge), and vocabulary are important building blocks of reading development.

In first grade, children continue to develop phonemic awareness, graphophonemic skill, and vocabulary, but the greatest growth occurs in phonemic spelling, decoding, word identification, and text reading. By second and third grade, reading growth is reflected in the number and type of words individuals can read, the difficulty of texts they can read and comprehend, and the fluency with which these tasks are accomplished. As grade level increases, comprehension of more difficult texts becomes the primary measure of reading development. Screening measures cannot adequately mark individual differences in reading development unless they are sensitive to the different reading skills emphasized at different grade levels. Studies confirm that screening measures valid at one grade are invalid at other grade levels (O'Connor & Jenkins, 1999; Foorman et al., 1998).

Good and Poor Candidates for Screening Content. Research shows that some traits are better than others in forecasting reading success. This is important in weighing the merits of various candidates for the content of screening instruments. Perceptual and motor skills, once considered useful in identifying poor readers, do not appear to hold much promise for screening. In addition, general measures of receptive and expressive language ability do not seem to target very precisely children who will have difficulty acquiring beginning reading skills (Schatschneider, Fletcher, Francis, Carlson, & Foorman, in press). General language measures may be more useful in identifying students who will have difficulty in more advanced stages of reading acquisition where the emphasis is on reading comprehension.

Accuracy vs. Fluency Measures. Two types of performance have been used in screening: some emphasizing accuracy and some emphasizing fluency. Accuracy measures distinguish children according to number or percent of correct responses on tasks (number or percent of words segmented into phonemes, words identified), whereas fluency measures distinguish children according to the number of correct responses per minute. Accuracy measures reveal individual differences in knowledge; fluency reveals individual differences both in knowledge and speed of processing. O'Connor and Jenkins (1999) combined accuracy (phoneme segmentation) and fluency (rapid letter naming) in their screens, but most studies focus on accuracy (e.g., Foorman et al. 1998) or fluency (Good et al., 2001: Speece et al. 2003).

Previous Page | Next Page
(Research Highlights) | (Combined Measures)