Abstract
The simple views of reading (SVR) and writing (SVW) provide useful foundations for the interpretation of psychoeducational achievement batteries. Research has established that oral language, decoding, and transcription explain significant variance in reading comprehension and written composition, respectively. However, the specific task demands of subtests influence the relationships among these constructs. As a result, the degree to which the KTEA-3’s tasks conform to findings associated with the SVR and SVW are not known. These analyses evaluated the degree to which the KTEA-3 oral language and decoding or transcription tests explained variance in reading comprehension, and written expression. Results inform practitioners as to the relative contributions of each component and useful knowledge when determining which battery to use in psychoeducational evaluation.
Keywords
“The simple view of reading is not just for researchers. School psychologists, teachers, and curriculum coordinators will also find it to be a practical and insightful framework for understanding the reading process, pinpointing the sources of reading difficulties, and guiding lesson planning” (Kilpatrick, 2015, p. 46).
Both the simple view of reading (SVR) and (not so) simple view of writing (SVW) provide important frameworks for psychoeducational literacy assessment (Kilpatrick, 2015; Ritchey et al., 2016). Collectively, they divide reading and writing into components associated with oral language and transcription/decoding competency. Because these models are ubiquitous within reading and writing research (Catts, 2018), practitioners should know the degree to which relationships among measures in popular test batteries conform to them. This investigation analyzes SVR and SVW within the Kaufman Test of Educational Achievement, 3rd Edition (KTEA-3; Kaufman & Kaufman, 2014).
The Simple Views of Reading and Writing
Both models originated over 30 years ago, emphasizing a role for phonics in literacy instruction (Catts, 2018). The SVR describes reading comprehension as the mathematical product of linguistic comprehension and decoding skills (Gough & Tumner, 1986). Similarly, the SVW summarizes written composition as a function of ideation (the generation and organization of ideas) and spelling (Juel et al., 1986). Researchers often operationalize both ideation and linguistic comprehension as oral language skills (Foorman et al., 2018; Kim & Schatschneider, 2017).
While the SVR has generally not required major revisions or additions in its explanation of comprehension (Lonigan et al., 2018), researchers have expanded the SVW into what would later be termed the not so SVW (Berninger & Amtmann, 2003; Kim & Schatschneider, 2017). The not so SVW highlights roles for (1) text generation, oral language applied to the generation, and organization of ideas; (2) transcription, or handwriting and spelling, and skills required to convert oral language to writing; and last (3) self-regulation skills, the writer’s ability to monitor their performance, plan their writing, and set goals. The writer coordinates these skills within working memory. Other theorists have highlighted the need for a writing fluency construct to be included among transcription and oral language (Kim et al., 2018). A separate fluency construct does not appear as necessary in the SVR, although there is inconsistency across studies investigating its inclusion (Adolf et al., 20l6; Catts, 2018; Lonigan et al., 2018). Kim et al. (2018) defined text writing fluency as efficient, automatic writing of connected text. They stressed that like composition, text writing fluency would also draw on oral language and transcription skills, mediating the relationship between them and composition.
These frameworks are useful foundations for the clinical assessment of academic skills. They apply to multiple languages, orthographies, and writing systems (Catts, 2018; Graham & Eslami, 2020; Florit & Cain, 2011; Yeung et al., 2017). Further, longitudinal and intervention research supports causal interpretation of the relationships among the models’ constructs. As a result, they can provide links between assessment results and intervention efforts. Graham et al’s. (2012) meta-analysis on writing instruction described significant effects for self-regulation (ES = .50), text structure (ES = .59), and transcription intervention (ES = .55) on writing outcomes. Santangelo and Graham’s (2015) meta-analysis indicated that improved handwriting led to increases in writing quality (ES = .84) and length (ES = 1.33; see also Graham et al., 2000). A meta-analysis for spelling instruction on writing quality suggested more modest, nonsignificant results (ES = .19; Graham & Santangelo, 2014), although perhaps the finding highlights the significant effort that simultaneous use of text generation and spelling can require from developing writers (Berninger, 1999). Nevertheless, in that analysis, spelling instruction improved spelling in composition (ES = .94). Oral language intervention increases reading comprehension, although frequently its effects appear significantly stronger with researcher-developed measures than standardized tests (ES = .68 for researcher developed vs. ES = .09 for standardized measures; Silverman et al., 2020). Decoding intervention also increases reading comprehension (Torgesen et al., 2001), although again, effects are often smaller on standardized measures (Melby-Lervag & Lervag, 2014). Importantly, the components of the SVR demonstrate instructional skill-by-treatment interactions (Conner et al., 2009). Children with lower decoding skills demonstrate greater literacy growth in classrooms providing significant code-based, explicit instruction, while students with lower vocabulary skills require more meaning-focused instruction.
Nevertheless, causal assumptions for SVR/SVW components on comprehension or composition should not be overstated. An increase in reading comprehension could theoretically improve decoding and oral language. After all, when practicing reading comprehension, students simultaneously practice decoding and linguistic comprehension, likewise for composition, text generation, and transcription. To that point, the number of books youth read explains growth in listening comprehension (Hedrick & Cunningham, 2002), a finding suggesting the opposite causal direction to those advocated here. This may be why some researchers have called for direct intervention in reading comprehension through strategy instruction (Melby-Lervag, & Lervag, 2014). Just like the components of the SVR, strategy instruction increases reading comprehension (Berkeley et al., 2010). Yet, Scarborough (2009) noted that reading comprehension develops as language use becomes more strategic. Thus, comprehension strategies could be associated with the oral language side of the SVR. That may explain why strategy instruction appears more effective once students gain a minimum level of decoding skill (Willingham, 2006) and does not necessarily increase the accuracy of decoding (Vaughn et al., 2000).
Implications for Standardized Achievement Batteries
Research on these models describes relationships among oral and written language skills that should hold true for measures within standardized achievement batteries. Both models generally explain a large amount of variance in their respective literacy outcome. When operationalized with latent variables, decoding and oral language frequently explain nearly 100% of variance in comprehension (Lonigan et al., 2018). Transcription and oral language explain a sizeable amount of composition variance (Graham & Eslami, 2020). Kim and Schatschneider (2017) reported these skill areas to explain more than two-thirds of variance in written composition.
The magnitude of these relationships generally changes across grade levels (Catts, 2018). In younger grades, both reading comprehension and written composition appear more related to decoding and transcription skills, respectively (Berninger, 1999; Lonigan et al., 2018; Limpo & Alves, 2013). When students develop automaticity with lower-order skills, then written language appears more associated with oral language. This trend may explain why the factor structure of the KTEA-3 appears to vary based on grade level (Parkin & Frisby, 2018). Nevertheless, despite these cross-sectional trends and the conceptual distinctness between decoding/transcription and oral language tasks, these skill areas tend to explain more common variance than unique variance in their respective outcomes (Lonigan, Burgess & Schatschnieder, 2018), perhaps because skills like reading comprehension do not represent a mere summation of components but a fundamental integration of multiple skill areas that develops through practice (Scarborough, 2009).
The relationship among these variables requires investigation within achievement batteries because their magnitude can vary based on task demands and characteristics of the examinee. The nature of these relationships is likely different within different psychoeducational batteries. Thus, this information can be useful to practitioners when selecting a battery to use in a comprehensive evaluation. For example, Cutting and Scarborough (2006) described differences between the oral language/reading relationships within the Gates-MacGinitie (MacGinitie et al., 2000), Gray Oral Reading Test (Wiederholt & Bryant, 1992), and Wechsler Individual Achievement Test (Wechsler, 1992). Garcia and Cain (2014) reported that the correlation between decoding and comprehension varies based on the way skills are measured. Measures of decoding accuracy may be more associated with comprehension in younger readers, while decoding fluency may be more associated with comprehension in older readers (Language and Reading Research Consortium, 2015). Tests of real word decoding appear more associated with comprehension than nonsense words. Genre (expository vs. narrative) and whether the examinee reads aloud or silently also impacted the degree of relationship. Decoding appears more related to comprehension in narrative stimuli and when reading silently than for expository text and reading aloud (Garcia & Cain, 2014). Consequently, different test batteries may describe different examinees as demonstrating skill deficits (Keenan & Meenan, 2014). While research on variability across test batteries’ written language subtests appears less well-developed, it is plausible that subtests within different batteries may demonstrate different relationships among oral language, transcription, and composition tasks, and as a result, identify different students as requiring intervention.
The Purpose of This Investigation
These models are important to consider when evaluating literacy with the KTEA-3. The battery includes two measures of reading comprehension, one that focuses on discourse-level language and another that focuses on sentence- and word-level language to infer word definitions. These tasks can be summed into a composite. The degree to which decoding and oral language explain comprehension variance has not been reported for this battery, beyond the correlations in the manual. However, practitioners should know if the comprehension tasks (and their composite) are more influenced by decoding or oral language skills (Kilpatrick, 2015). A reading measure more influenced by oral language may not pick up on reading challenges experienced by students with stronger oral language skills. Likewise, a decoding-influenced subtest may mask oral language challenges an examinee presents. The KTEA-3 includes just one measure of higher-order written expression, yet it appears to combine sentence-level mechanical skills such as grammar and syntax with discourse-level text generation, skills necessary to construct the essay at the end of the subtest. Because the task includes many sentence-level items that vary according to grade level, the degree to which higher-order language influences performance on the task is an open question. The battery also includes a measure of writing fluency. The relative influence of transcription and oral language on writing fluency can be helpful for clinicians to know. However, testing it as a true mediator, akin to Kim et al. (2018), may not be possible here, given the limitations of available data (Kline, 2015).
This investigation seeks to answer the following research questions: In the KTEA-3, how much variance in reading comprehension and written expression is explained by decoding and oral language, and transcription, and oral language, respectively? Do oral language and decoding skills differentially influence the battery’s two reading comprehension measures? Do oral language and spelling skills differentially influence written expression and writing fluency? What could be the indirect effects of these skills on written expression if writing fluency is an endogenous variable? Does the amount of variance explained by components change across grade levels? To what degree is explained variance unique to a component, or do components primarily explain shared variance in reading comprehension or written expression?
Method
Participants
Participants come from the KTEA-3 grade norm standardization sample, a group of 2600 students between Pre-K and 12th grade, split across two test forms. The standardization sample is consistent with the 2012 census data across age/grade, sex, parent education level, ethnicity, geographic region, and special group designation. More information can be found in the battery’s technical manual (Kaufman & Kaufman, with Breaux, 2014). Grades 1 through 12 were included in these analyses, although grade 1 was excluded from the SVW analysis, as first grade students do not complete all writing measures. Grades 1 through 8 each include 200 participants, while grades each 9 through 12 included 150 participants.
Instrument
The KTEA-3 is an individually administered battery of academic achievement measures. These analyses focus on its measures of oral and written language. Oral language measures include three subtests. Listening Comprehension requires the examinee to answer questions based on short passages they heard. It demonstrated a mean reliability of .85 in the standardization sample. Oral Expression requires the examinee to formulate a sentence for a picture. It demonstrated a mean reliability of .81. Associational Fluency requires the examinee to name exemplars of two categories each within a brief time limit. It demonstrated a mean reliability of .62. Regarding reading, these analyses included Letter Word Recognition, Nonsense Word Decoding, Reading Comprehension, and Reading Vocabulary. The first two require examinees to read word lists of real and nonsense words. They demonstrated mean reliabilities of .97 and .96, respectively. Reading Comprehension parallels the demands of Listening Comprehension, requiring examinees to answer questions based on passages they read. It demonstrated a mean reliability of .88. Reading Vocabulary requires examinees to determine a word with similar meaning as one within a sentence. It demonstrated a mean reliability of .93. Last, pertaining to writing, these analyses included the Written Expression, Spelling, and Writing Fluency measures. Written Expression requires examinees to write words and sentences necessary to complete a story and compose an essay about that story. It demonstrated a mean reliability of .86. Spelling does not count in its scoring. Spelling requires the examinee to spell words from dictation. It demonstrated a mean reliability of .95. Last, Writing Fluency requires the examinee to quickly write sentences to describe a picture. It demonstrated a mean reliability of .76.
Data Analysis
Separate models for the SVR and SVW were constructed from the aforementioned subtests and modeled in R (R Development Core Team, 2015), with the lavaan package (Rosseel, 2012). Subtest intercorrelations, provided in the KTEA-3 technical manual, were used as input. Each model is illustrated in Figure 1. Models operationalizing the simple views of reading and writing. Note. These models illustrated the configuration of KTEA-3 subtests within the simple views of reading and writing. LWR: Letter Word Recognition; NWD: Nonsense Word Decoding; OE: Oral Expression; LC: Listening Comprehension; AF: Associational Fluency; RV: Reading Vocabulary; RC: Reading Comprehension; SP: Spelling, WF: Writing Fluency; WE: Written Expression.
The SRV was modeled twice. In the first model, latent decoding, oral language, and comprehension factors were specified from the aforementioned subtests. These factors are consistent with the battery’s decoding, oral language, and reading understanding composite scores, although it is important to note that latent factors and summed composites are not equivalent. In the second model, to determine whether the KTEA-3’s reading comprehension tests are differentially influenced by decoding and oral language skills, the Reading Comprehension and Reading Vocabulary subtests were included as single indicator factors with their residual variance constrained to 0.
For the SVW, a latent oral language factor was included in the model, while Spelling, Writing Fluency, and Written Expression were modeled as single indicator factors with residual variance constrained to 0. In the first model, Writing Fluency is included as an exogenous variable, modeling its potential influence on Written Expression. In the second, Writing Fluency is included as an endogenous variable, akin to Kim et al’s. (2018) model. Note that these two models are equivalent and cannot be differentiated based on their fit to the data. While it is possible to calculate the magnitude of components’ direct and indirect effects on Written Expression, interpreting them as true mediation effects may not be appropriate due to concurrent administration of subtests in the standardization sample (Kline, 2015). However, the aforementioned meta-analytic intervention studies provide support for directionality assumed in the model.
These analyses include multiple measures to evaluate model fit (Keith, 2015). They provide χ2 values, where a nonsignificant p value indicates appropriate model fit, the root mean square error of approximation (RMSEA), where values lower than .08 suggest an adequate fit, the comparative fit index (CFI), and Tucker-Lewis index (TLI), where values greater than .90 represent adequate fit.
To determine the relative influence of decoding and oral language on comprehension (SVR), or oral language, spelling, and handwriting on the written expression subtest (SVW), the respective outcome variables were simultaneously regressed on model components for each grade level. To determine the level of common versus unique variance explained by components of each model, the respective outcome variable was regressed sequentially with each component variable entered last in the regression equation via the procedure explained by Beaujean (2014). The unique variance explained by each variable was subtracted from the total variance explained by the model.
Results
In the KTEA-3, how much variance in Reading Comprehension and Written Expression is explained by decoding and oral language, and transcription, and oral language respectively?
Model Fit for Simple View of Reading Across Grade Levels.
Note. RMSEA: root mean square error of approximation; CFI: comparative fit index; TLI: Tucker-Lewis index.
aresidual variance in comprehension constrained to be greater than 0.
bresidual variance in LWR constrained to be greater than 0.
ccovariance between reading comprehension and reading vocabulary constrained to be equal to 0.
dresidual variance in LWR constrained to be greater than .001.

Simultaneous regression of reading comprehension and written expression on model components across grade levels. Note. Due to multicollinearity, the spelling variable was removed in the simple view of writing for grade 12.
Model Fit for Simple View of Writing Across Grade Levels.
Note. RMSEA: root mean square error of approximation; CFI: comparative fit index; TLI: Tucker-Lewis index.
aSpelling removed as a predictor due to multicollinearity. As the two writing models in Figure 1 are equivalent, these fit statistics apply to both.
Do language and decoding skills differentially influence the battery’s two reading comprehension measures?
The relative effects of oral language and decoding skills on both reading comprehension Reading Comprehension and Reading Vocabulary are described in Figure 3, and fit measures are included in Table 1. For grades 8, 10, and 12, the covariance between reading comprehension and reading vocabulary was constrained to 0. While both subtests show the same general trend of which influence appears greater, the magnitude of that trend appears much larger for the reading comprehension subtest. Decoding generally stops demonstrating the greatest influence on performance in the reading comprehension subtest after grade 2. However, it influences reading vocabulary performance across all grade levels to a greater degree, with the exception of grade 11. Variance explained in KTEA-3 reading comprehension and writing subtests. Note. Components simultaneously regressed on outcomes. Coefficients for writing calculated with writing fluency as an endogenous variable. Spelling removed as a predictor in grade 12 due to multicollinearity.
Do oral language and spelling skills differentially influence written expression and writing fluency? What could be the indirect effects of these skills on written expression, if writing fluency is modeled as an endogenous variable?
Indirect Effects of Oral Language and Spelling on Written Expression via Writing Fluency.
Note. Spelling removed from grade 12 model due to multicollinearity.
Does the amount of variance explained by components change across grade levels?
Models of the SVR and SVW generally operate as expected through the cross-sectional grade levels, as described in Figure 2. For the SVR, decoding explained the vast majority of variance in the comprehension factor and its effect declines as grade level increases. Oral language demonstrated the opposite pattern. Grades 7 and 10 tended to deviate from this trend somewhat. Post hoc inspection of latent correlations revealed that both predictors were highly correlated with comprehension in those grade levels. The SVW demonstrated a similar pattern, although not as pronounced as in the reading measures. Oral language generally increased in effect on Written Expression over grade levels, with notable exceptions of fourth and seventh grade. Writing Fluency, used as an operationalization of the handwriting aspect of transcription, never demonstrated a significant effect on the written expression subtest with the exception of second grade and sixth grade.
To what degree is explained variance unique to a component, or do skills primarily explain shared variance in comprehension or Written Expression
Figure 4 describes unique and common variance explained across measures in the reading comprehension factor and Written Expression. As described in other analyses (e.g., Lonigan et al., 2018), the majority of variance explained is common among measures. In both the SVR and SVW, oral language increases its unique variance across grade level in general. Unique and common variance explained within the reading comprehension factor and Written Expression subtest. Note. These charts contrast the percentage of uniquely explained variance in comprehension or written expression by each component of the simple views of reading and writing. Spelling removed from grade 12 due to multicollinearity.
Discussion
Both the SVR and SVW represent useful models for clinicians to consider when assessing reading and writing skills. Not only do they apply to diverse languages, orthographies, and writing systems, intervention research supports causal interpretation of their components on reading comprehension and writing composition. This investigation analyzed the SVR and SVW within the KTEA-3, evaluating whether relationships among measures conform to prior results for these models.
Implications from the SVR for the Interpretation of the KTEA-3
The KTEA-3 normative correlations largely conformed to predictions stemming from the SVR. Its decoding and oral language tasks explained nearly 100% of comprehension variance. Decoding’s influence decreased across grade level, while oral language effects increased. Language became the stronger influence at approximately 3rd grade, consistent with a recent meta-analysis (Garica & Cain, 2014). Lastly, most variance explained by the predictors was common variance, although oral language explained unique variance in many grade levels.
Decoding and oral language appear to influence the reading comprehension subtests somewhat differently. While oral language quickly becomes the dominant influence on performance for the reading comprehension subtest, decoding remains influential on reading vocabulary across most all grade levels. Kilpatrick (2015) warns that some comprehension measures may mask reading challenges in students with oral language strengths, and that may be the case with these measures, at least after grade 4. The reading comprehension subtests appear reliable enough to interpret individually at many age/grade levels (Kaufman & Kaufman, with Breaux, 2014), according to general reliability guidelines (Kranzler & Floyd, 2020). Due to these differing decoding/oral language effects, separate interpretation could be clinically useful. At the same time, interpretation of the score difference between Reading Comprehension and Reading Vocabulary is likely unreliable.
Implications from the SVW for the KTEA-3
Much of the KTEA-3 writing and oral language correlations also conformed to the predictions of the SVW. The amount of explained variance was lower than in the SVR models, although that is likely due to the inclusion of error variance in most of the modeled variables. If error-free latent variables were used, explained variance may be larger. Additionally, self-regulation and working memory skills represent an important part of the not so SVW, and those skills were not included in the models tested here. Their inclusion may also increase explained variance.
As predicted by the SVW, spelling generally decreased in its influence on written expression across grade level, while oral language increased. Writing Fluency, when included as the handwriting aspect of transcription here, did not demonstrate any meaningful effect after second grade. This lack of effect may be due to Written Expression not significantly penalizing handwriting in its scoring, its provision of many items at the sentence-level tapping grammar/syntax, punctuation, and capitalization knowledge, and a lack of time limits on examinees. Assuming Writing Fluency can operationalize handwriting, this suggests that Written Expression should be interpreted as a writing test minimally influenced by handwriting. If handwriting, as a part of the transcription process, impairs an examinee in the classroom, Written Expression scores may appear higher than classroom work samples might suggest, not consistent with them. Of course, spelling is also not penalized, yet it explained Written Expression variance at many grade levels. This could mean that spelling challenges are not as easily controlled for in writing as handwriting. Insightfully, an anonymous reviewer stressed that misspelling may result in grammar or semantic errors that could influence performance on Written Expression.
Given that Writing Fluency demonstrated minimal effects on Written Expression, it is no surprise that there are few indirect effects for SVW components (listed in Table 3) on Written Expression. However, that finding itself does not necessarily mean the subtest does not represent the construct of writing fluency, described by Kim et al. (2018). The reason for nonsignificant, indirect effects may be associated with Written Expression's minimal reliance on fluency in writing. Instead, it is the lack of larger oral language effects on the subtest that suggest Writing Fluency may not represent that construct. Kim and colleagues described writing fluency as a mediator between transcription/oral language and composition and a construct influenced by both of those SVW components. Alternatively, the KTEA-3 authors designed Writing Fluency to be an indicator of transcription skills that required minimal generation of writing content (Kaufman & Kaufman, with Breaux, 2014). If that was the goal, then it should be expected that oral language demonstrated smaller effects on Writing Fluency.
The increase in explained variance by oral language skills suggests that Written Expression requires stronger language skills as students increase in grade level. This finding lends support to the progression of skills targeted by the subtest, as suggested by the test authors (Kaufman & Kaufman, with Breaux, 2014). However, the degree to which performance on the constrained, sentence-level items, compared to the open-ended essay influence scores remains an unanswered question. Conceivably, writing measures that focus more on text-level writing, such as the essay composition subtest of the WIAT-4 (Wechsler, 2020) may tap oral language or text generation skills to different degrees.
Limitations and Future Directions
These analyses are limited in a few ways. As mentioned earlier, they do not represent a full representation of the not so SVW, as self-regulation skills were not included as independent predictors of writing and Writing Fluency may not represent the handwriting component of transcription well. The use of correlations as input with a normative sample also represents a limitation. In higher grade levels (12th grade, specifically), spelling and oral language skills represented redundant predictors. These skills may be less associated in referral samples, and their explanation of comprehension and composition may change.
These analyses provide evidence that the KTEA-3 oral and written language measures conform to many of the implications from the simple views of reading and writing. Practitioners can feel confident that many of those relationships are operationalized in the battery. Because batteries’ task demands may influence the magnitude of these relationships, these results may not generalize to other batteries. Practitioners may benefit from a comparison of these analyses in other popular, comprehensive achievement batteries.
Footnotes
Acknowledgments
Portions of this paper are presented at the Learning Disabilities Association of American annual conference, 2021.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
