**3. Results**

The data obtained from the students' readings included measures of literal and inferential comprehension, word recognition automaticity (words read correctly per minute), prosody (qualitative rating of expressiveness using the multi-dimensional fluency scale), background knowledge, and strategy emloyment. Means and standard deviations for the six variables are presented in Table 1.

**Table 1.** Mean and standard deviations for the variables including comprehension and fluency components, strategy, and background knowledge.


The relationship between the comprehension related factors and comprehension itself, was determined by calculating correlations between variables. The correlations are presented in Table 2; all correlations were found to be statistically significant and substantive.

**Table 2.** Correlations of fluency components, strategy, and background knowledge with comprehension components.


*Note*: \*\* *p* < 0.01.

Given the positive and significant correlations among the variables in Table 2, we ran a path analysis using AMOS and Mplus statistical modeling programs. By this analysis, we aimed to determine the relations among the variables in an integrated model of reading comprehension. Those results are presented in the path diagram in Figure 1.

In the path analysis model, fluency, comprehension strategies, and background knowledge were used to predict reading comprehension. The path analysis results revealed that RMSEA, SRMR, TLI, and CFI values were within expected ranges. The fit of the data, then, to the proposed path model was considered good. For the full sample, the model yielded good fit indices. When reviewing overall fit summary indices in the model, the χ2 test yielded a value of 5.752, which was evaluated with 5 degrees of freedom, and had a corresponding *p*-value of 0.331. The χ2/*df* was 1.150. Additionally, the RMSA was 0.027. The TLI was 0.99 and CFI was 0.99. Moreover, SRMR was 0.0132. These fit measures sugges<sup>t</sup> that all of the indices expressed in the path analysis were a good fit to the data [46,47]. In the model, fluency and comprehension strategies made statistically significant contributions to the prediction of reading comprehension (β = 0.46, *p* < 0.001 and β = 0.27, *p* < 0.01, respectively). Background knowledge, however, did not make a statistically significant contribution to the prediction of reading comprehension (β = 0.13, *p* > 0.05). Overall, this model explained 45% of the variance in reading comprehension. Having a relatively high R-squared value in these predictions indicates that this model works well in the Turkish language context.

**Figure 1.** The relations of strategy use, reading fluency, and background knowledge with reading comprehension. The single-headed arrows show standardized regression coefficients and direct effects in the path model. Double-headed arrows represent correlations (covariance). Dotted arrows show insignificant coefficients in the path model. \*\*\* *p* < 0.001.
