https://link.springer.com/article/10.10 ... 997-0023-9
A follow-on study was conducted on first- and third-grade children who were tested on the Word Identification subtest of the Woodcock Reading Mastery Test. Errors were phonetically transcribed and proportions of total errors were assigned to one of four strategy types based upon the type of decoding analysis: whole word, part word, and either probable (legal) or improbable (illegal) phonetic decoding. The type of strategy employed was highly correlated with concurrent reading test scores and predicted 30 to 37% of the variance in word recognition 19 months later. Use of phonetic decoding was a strong positive predictor and whole word decoding a negative predictor at both first and third grade. Part word decoding became a negative predictor at third grade. Nearly all children used more than one strategy and there was a developmental trend to shift to a phonetic strategy. But this shift was not inevitable and had not occurred for 31% of the children at the close of the study. Children who stayed with the most inefficient strategies had significantly higher vocabulary scores and equivalent phonemic processing ability when compared to readers with more efficient decoding strategies.