Assessments play an important role in the education of all students. They measure skills or knowledge in a standardized way and provide important information for further support (Lai & Schildkamp, 2013). In school performance, for example, they refer to crucial skills, such as reading. Reading is essential for academic success, and early identification of reading difficulties is important for providing appropriate support (Hakkarainen et al., 2013). Reading screenings therefore are well established in practice and can be used in combination with formative diagnostics (Fuchs et al. 2007). Supporting students with learning disabilities (LD) is an important aspect of inclusive education and early identification of reading difficulties is crucial for providing appropriate support (Fuchs & Fuchs, 2011). To measure more reliably, more quickly and thus more efficiently, the use of technology in education has grown significantly in recent years, and the integration of digital tools in assessment practices has the potential to improve the assessment process (Timmis et al. 2016). Adaptive assessments, in particular, present a perspective on the simultaneous testing of a heterogeneous, diverse student body (Stone & Davey, 2011).

The study is embedded in a larger project, in which a digital adaptive reading screening is to be developed freely available for teachers. Therefore, the study investigates the results of two data collections using different versions of the German inclusive reading screening “Les-In”. The initial version of the screening was developed as a paper-based test (Ebenbeck et al., 2022) and administered to a sample of 700 third graders, with 5% of the students having a LD. Results indicated that two out of four subtests did not measure as broad as expected (subtest 2: Mdn_σ=-0.14, M_σ=0, SD_σ=1.70; subtest 3: Mdn_σ=-0.12, M_σ=0, SD_σ=0.66). Also, ceiling effects in the summative scores (subtest 2: n_items=34, Mdn_sum=31 , M_sum=29.30, SD_sum=3.22; subtest 3: n_items=25, Mdn_sum=25, M_sum=23.37, SD=4.08) indicate the missing of harder items. Because of those reasons, the subtests were modified. The modified tasks were then integrated into a digital version of the test on the web-based platform Levumi.de (Jungjohann et al., 2018). The digital test was administered to a sample of 400 second to fourth  graders, with 7% of the students having a LD.

At the student level, the study examines the extent to which reading skills differ between students with and without LD and whether there is overlap between student groups. At the test level, the study examines whether the modification of the two tasks has added psychometrical value and whether the tasks can now more accurately measure students across a broader range of performance. For this purpose, the results of the digital screening are considered both individually and in comparison with the analog version. The results show that the change in tasks had the desired effect, and students with and without disabilities can now be measured more accurately across a broader range of performance (subtest 2: Mdn_σ=0.10, M_σ=0, SD_σ=0.89; subtest 3: Mdn_σ=-0.39, M_σ=0, SD_σ=0.81). The results also indicate that there are differences in the reading performance of students with and without LD, as students with LD show lower reading skills in the tasks. The extent to which the results can be used to expand the digital screening into an adaptive digital screening is further evaluated and discussed. Further discussed is the extent to which inclusive screening assessments should relate primarily to the low-ability domain, and the extent to which the value of diversity of students with and without disabilities and their individual learning levels and reading abilities can and must be considered in the development of assessments.

Two data collections are evaluated and compared. In the first collection, 700 third graders from German elementary and special schools completed the first version of the screening as a paper-pencil test. The screening includes four subtests, namely “Phonological Awareness” (subtest 1; 36 items), “Vocabulary” (certainty of lexical recall; subtest 2; 36 items), “Flash Reading” (speed of lexical retrieval; subtest 3; 30 items) and “Sentence Comprehension” (subtest 4; 60 items). The analog subtests correlate between r=.26 and r=.52 (p<.001). All tasks are processed single-choice. Three subtests have a maximum processing time of five minutes and are therefore speeded tests. The subtest 3 has no time limit but ends when all items are completed. Students complete the screening simultaneously as a class. For subtests 2 and 3, the psychometric evaluations of this survey showed that the performance range is measured too narrowly and therefore the item pools must be expanded to include more difficult items. In the modified screening, subtests 2 and 3 were therefore changed. In subtest 2, more pseudowords were added to the item pool, and in subtest 3, the display duration of the words was varied. The screening’s modified version was digitalized as web-based computer- or tablet test. The digital subtests correlate between r=.34 and r=.55 (p<.001). In the second collection, 400 second to fourth graders of German elementary and special schools completed this digital and modified version of the screening. All students completed the screening on the tablet simultaneously in class. Subtests 1, 2 and 4 had a maximum completion time of five minutes. In subtest 3, all items in the pool were again processed without a maximum processing time. It is examined how the number of correctly and incorrectly solved tasks changed after the modification of the screening. Students with and without LD are considered separately and compared in order to work out their individual abilities. Subsequently, the new version of the screening will be examined psychometrically. For this purpose, the fit to the one-dimensional Rasch model of each task of the screening is examined. Using graphical model tests, unfair items are identified and removed from the item pool. On the item level, the difficulty of the individual items and of the entire test is examined and compared with the analogous version of the screening. In this way, it is determined whether the modifications have had the desired effect on the difficulty.

In conclusion, the results of the study show that the change in tasks had the desired effect, and students with and without LD can be measured more accurately across a broader range of performance with the second Version of in the screening “Les-In”. The results also indicate that there are differences in the reading performance of students with and without LD, as students with LD show in average lower reading skills in the tasks. Nevertheless, there are also students without LD who show very weak reading performance. The goal of the project is to develop a digital adaptive reading screening for inclusive education. In the next step, the results presented in this study are therefore used to develop an adaptive version of the screening with help of various simulation studies. The item pools seem to be suitable for this purpose so far. The exact performance and suitability of the pools will become clear in further studies. The expanding to an adaptive screening would present a perspective on the simultaneous testing of a heterogeneous study body, as thus tests adapt their difficulty depending on the student’s answer pattern. This would lead to a shorter test length while maintaining an accurate measurement. Furthermore, the study also highlighted the potential of using technology in assessment practices to improve the assessment process and provide more efficient and reliable results for all students. Overall, the study emphasizes the importance of inclusive education and early identification of reading difficulties in order to provide appropriate support for all students.

Ebenbeck, N., Jungjohann, J., & Gebhardt, M. (2022). Testbeschreibung des Lesescreenings LES-IN für dritte inklusive Klassen. Beschreibung der Testkonstruktion sowie der Items der Screeningtests" Phonologische Bewusstheit"," Sicherheit im lexikalischen Abruf"," Geschwindigkeit im lexikalischen Abruf" und" Sinnkonstruierendes Satzlesen" in deutscher Sprache. Version 1. Fuchs, L. S., Fuchs, D., Compton, D. L., Bryant, J. D., Hamlett, C. L., & Seethaler, P. M. (2007). Mathematics Screening and Progress Monitoring at First Grade: Implications for Responsiveness to Intervention. Exceptional Children, 73(3), 311–330. https://doi.org/10.1177/001440290707300303 Fuchs, D., & Fuchs, L. S. (2011). Responsiveness to Intervention: Multilevel Assessment and Instruction as Early Intervention and Disability Identification. The Reading Teacher, 63(3), 250–252. https://doi.org/10.1598/RT.63.3.10 Hakkarainen, A., Holopainen, L., & Savolainen, H. (2013). Mathematical and reading difficulties as predictors of school achievement and transition to secondary education. Scandinavian journal of educational research, 57(5), 488-506. Jungjohann, J., DeVries, J. M., Gebhardt, M., & Mühling, A. (2018). Levumi: A web-based curriculum-based measurement to monitor learning progress in inclusive classrooms. In Computers Helping People with Special Needs: 16th International Conference, ICCHP 2018, Linz, Austria, July 11-13, 2018, Proceedings, Part I 16 (pp. 369-378). Springer International Publishing. Lai, M. K., & Schildkamp, K. (2013). Data-based decision making: An overview. Data-based decision making in education: Challenges and opportunities, 9-21. Timmis, S., Broadfoot, P., Sutherland, R., & Oldfield, A. (2016). Rethinking assessment in a digital age: Opportunities, challenges and risks. British Educational Research Journal, 42(3), 454-476. Stone, E., & Davey, T. (2011). Computer‐adaptive testing for students with disabilities: A review of the literature. ETS Research Report Series, 2011(2), i-24.