Selecting and interpreting reading assessments for struggling adult readers can be difficult, as few literacy assessments are designed for this group. In addition, modeling the relations among developing reading skills for adults may be different from what we might expect from a model of reading skills in children. The research here examines the relation between reading skills in struggling adult readers using test-level and item-level models.

The first study models the relations between reading assessments and reading skills in 624 native English speaking adult struggling readers. A series of confirmatory factor analysis (CFA) models were fit based on a multi-trait/multi-method (MTMM) approach. Results from the series of confirmatory factor models indicate that silent word reading accounts for the most trait-related variance in the overall reading model, and speededness accounts for the most method-related variance in the model. The model results reaffirm patterns in past research which indicated that there is a lower correlation/integration of reading skills than found with typically developing children (e.g., Sabatini, Sawaki, Shore, & Scarborough, 2010). It also expands prior research by indicating that specific skills may operate differently in this population and the results can inform our understanding of the overall reading process for struggling adult readers (e.g., Nanda, Greenberg, & Morris, 2014).

The second study is an item-level confirmatory factor analysis of the Test of Irregular Word Reading Efficiency (Reynolds & Kamphaus, 2007) and the Word Attack and Letter-Word Identification subtests of the Woodcock Johnson III Test of Achievement (Woodcock, McGrew & Mather, 2007) which were administered to 931 native and non-native English speaking adults who struggle with reading. Using item-level CFA models structured based on an MTMM approach, this study examines the extent to which the items of these three tests measure general word reading ability versus test-specific skills (traits vs. methods) when administered to adult struggling readers and how that measurement structure may be equivalent in both native and non-native speakers of English. The findings from this study indicate that while group differences are found, structural measurement invariance holds across native and non-native groups under both weak and strong (metric/scalar) measurement invariance conditions.