On modeling

Espin and Deno (1989) compared the effects of modeling (i.e., examiner models accurate reading of the entire word) versus prompting (i.e., examiner says the initial part of a word) on sight word reading. They found that modeling was more effective at increasing accuracy of responding than prompting...

...Word Drill (i.e., having the student repeat a misread word following examiner modeling of accurate reading) has been shown to increase accuracy and fluency rates relative to no error correction procedures, modeling with one response repetition only, and phonic analysis (a prompting strategy for eliciting accurate responding) (Rosenberg, 1986)...

...Shapiro and McCurdy (1989) found that training words in isolation using modeling and prompting did not lead to generalization effects for passage reading...

...One study found that having subjects repeat the entire phrase containing a misread word (examiner modeling + phrase drill) was more effective at improving oral reading fluency in isolation and in the context of passage reading than having the experimenter model accurate reading of the misread word or even having the subject repeat only the error word several times (O'Shea, Munson, & O'Shea, 1984)...

...Although instructing in isolation through modeling and prompting may be the most efficient means to promote accuracy (Grossen & Carnine, 1991), fluency building should probably occur in the natural context of skill use (e.g., in passages) to promote generalization more rapidly

Daly, E. J., III, Lentz, F. E., & Boyer, J. (1996). The instructional hierarchy: A conceptual model for understanding the effective components of reading interventions. School Psychology Quarterly, 11, 369–386.