**2. Methods**

## *2.1. Selection Procedures*

Several databases [i.e., PsycINFO, Education Source, Academic Search Ultimate and Education Resources Information Center (ERIC)] were searched using the terms oral reading fluency, and intervention\*, repeated reading, choral reading, partner reading, assisted reading, readers' theater, and reading di fficulty\*, struggling reader\*, or reading disability\* and elementary, primary, or K-5. Since the National Reading Panel (NRP) report identified fluency as an essential component of reading instruction which led to increased interest in fluency research [5,43,44], only peer-reviewed papers published after the NRP report (i.e., 2000–2019) that discussed fluency interventions' impact on elementary students with reading di fficulties or learning disabilities were identified to be reviewed. Selecting papers that have undergone the peer-review process, which carefully inspects studies for publication, helps to ensure the findings align with the most recent evidence-based practices in fluency instruction. In addition, the references cited in recent meta-analysis and synthesis papers examining oral reading fluency interventions were inspected for applicable papers to review [37,45]. Finally, each article's cited references were searched for relevant papers.

The initial search resulted in 365 studies (see Figure 1). Thirty-two articles were duplicates and therefore immediately removed. The titles, keywords, and abstracts of the remaining 333 studies were reviewed for exclusion criteria. Studies were excluded if the title, keywords, or abstracts clearly identified that the study (1) was a synthesis, meta-analysis, or review paper, (2) included students not enrolled in kindergarten through fifth grade, (3) contained instruction in languages other than English, (4) included students with cognitive, visual, or hearing deficits, or (5) did not target oral reading fluency intervention. Using this exclusion criteria, 199 studies were excluded based on title, keyword, or abstracts, leaving 134 studies for further review. The following inclusion criteria were used for an in-depth analysis of the 134 studies.


**Figure 1.** The Preferred Reporting items for Systematic Reviews and Meta-Analyses Flowchart.

After review, 16 of the 134 studies comprehensively reviewed met the criteria for inclusion; 118 studies were excluded because they (a) included participants who were not described as at risk for or having reading difficulties (3 studies), (b) did not use experimental or quasi-experimental designs with a treatment-control group (63 studies), (c) did not include elementary-age participants (1 study), (d) examined fluency intervention delivered in a foreign language (3 studies), (e) examined fluency intervention delivered at home or in a clinical setting (7 studies), (f) did not focus on the implementation of an oral reading fluency intervention (38 studies), or (g) did not include oral reading fluency of connected text as a dependent measure (3 studies).

#### *2.2. Coding Procedures*

The final 16 studies underwent an inclusive coding protocol by the first author. Twenty percent of the studies were double coded by the first and third authors; 94% interrater agreemen<sup>t</sup> was achieved, and discrepancies in coding were resolved via discussion. The following categories were included in all coding procedures: (a) author and study design; (b) participant information, including if English Language Learners were included in the participants; (c) treatment description, including group size, duration, and other intervention variables such as text level used and number of reads; (e) ORF dependent measures; (f) reading comprehension dependent measures if applicable; and (f) findings. See Table 1 for study information.




*Educ.Sci.* **2020**, *10*,

 52


*Educ.Sci.* **2020**, *10*,52

**Table 1.** *Cont.*


*Educ. Sci.* **2020**, *10*, 52




*Sci.* **2020**, *10*, 52

*Educ.*


**Table 1.** *Cont.*



**Table 1.** *Cont.*

#### *2.3. Calculation of E*ff*ect Sizes*

All 16 studies included in this review provided adequate statistical information (i.e., sample means, pre- and post-intervention group means and standard deviations) that allowed us to calculate the e ffect sizes for the di fferences between the intervention and control groups on oral reading fluency outcomes. Nine of the studies provided statistical information to calculate e ffect sizes associated with reading comprehension outcomes. We chose to use the e ffect size formula by Carlson and Schmidt [62], where the di fference between the intervention and control group pre–post-change mean score was divided by the pooled standard deviation of the intervention and control groups at pre-test. To reduce bias in the estimation of e ffect sizes, we also applied a bias correction parameter (that took into account the sample size the two groups) to the formula, as recommended in Morris [63]. This formula was chosen because (1) it accounted for pre-test di fferences between the intervention and control groups, (2) used a pooled standard deviation of both groups at pre-test only that reduced the potential of violating the homogeneity of variance assumption, and (3) findings from simulation studies showing the robustness of this method of e ffect size estimation [63]. Based on recommendations by Cohen [64], estimates of 0.20–0.49 were considered small e ffects, 0.50–0.79 medium e ffects, and values above 0.80, large e ffects.
