Student engagement in inclusive classrooms

ABSTRACT

Using large scale survey data, I document substantial differences in behavioural engagement (defined as involvement in academic and social activities, cooperative participation in learning, and motivation and effort) and emotional engagement levels (defined as a sense of belonging and well-being at school) between students with and without special needs in regular classes and show that the impact of engagement, in particular motivation and effort and participation in learning activities, on academic achievement is equally important for SEN and non-SEN students. The results highlight the importance of inclusion initiatives aimed at improving the engagement of SEN students included in regular classrooms.

KEYWORDS:

• Education economics
• instrumental variables
• special needs education
• mainstreaming
• student engagement

Acknowledgements

The usual disclaimer applies. Thanks to Paul Bingley, Mette Gørtz, the participants of an Expert Panel Workshop at SFI, and Torberg Falch and two anonymous referees of this journal for valuable comments and suggestions.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. Some studies, however, found similar levels of school engagement in students with and without SEN (Bardin and Lewis 2008; Wallace et al. 2002).

2. See e.g. Greene (1993, 279ff). Moreover, this bias is well-known to be exacerbated in equations like the ones I use in this study, i.e. estimating the effect on changes in the outcome variable (Griliches 1977).

3. The inclusion rate is the overall share of students educated in inclusive settings.

4. Details about the construction of the indices are documented in Christensen (2016).

5. This is similar to a specification with the difference as the dependent variable, but it allows the coefficient on the lagged value to be estimated instead of being fixed to 1.

6. Note that each student was asked six additional questions: two questions each about two peers, plus a question on the well-being of two peers.

7. Reading tests are also administered in year 2, but this is not relevant for the cohorts included in the survey data.

8. I.e. whether the child lives with both parents or not.

9. The Strengths and Difficulties Questionnaire (SDQ) is a brief behavioural screening questionnaire about 3–16 year olds. The self-reported version that is used in the student survey is suitable for young people aged around 11–16. For the analyses, I use three subscales of the SDQ: emotional symptoms, conduct problems and hyperactivity/inattention. The scores in each scale range between 0 and 10. Students with scores below 6 are generally considered to be within the ‘normal’ range. According to the definition adopted for use in this study, a student has considerable difficulties if he/she scores outside the normal range in one or more of the three subscales. For further information, see http://www.sdqinfo.org/.

10. The following are the shares of students fulfilling the different requirements: students receiving continuously extra support (7%), students recently returned from segregated special needs education (2%), students with an SDQ-score beyond the ‘normal range’ (9%), and students diagnosed with a psychiatric disorder (5%).

11. Another assessment indicates a somewhat lower number: 14–15% (Nielsen and Skov 2016).

12. In this figure, I use data from all five waves of the panel. The time fixed effects are effectively ‘wave fixed effects’ with a set of dummy variables to indicate which survey wave the observation comes from.

13. The two selected questions are the frequency of (1) participation in classroom discussions and (2) being together with children from your class during recess. Note that we have peer-reported data on only 2 out of the 12 items used to construct the engagement indices. Thus, the main analysis (next subsection) cannot be run using IV models with peer-reported variables.

14. A full results table is provided in the appendix (Table A3). Results from models without SES controls are qualitatively similar (Table 6, col 2).

15. Recall that the analysis using peer-reported measures of engagement suggested that these conventional estimates are probably smaller than estimates that take account of measurement error.

16. This estimate represents the mean of a number of calculations using separate coefficient estimates for the two cohorts from Table 6 and using different estimates of learning gains from the literature. E.g. for the younger cohort (year 6), the estimate for motivation and effort is 0.048 SD (Table 6, col. 7), and I use three different estimates of yearly learning gains from the literature: 0.32 SD (Hill et al. 2008; Table 1, grade 5–6); 0.40 SD (Hattie 2012); and 0.049 SD (Luyten, Merrell, and Tymms 2017; Table 2, cohort 6 vs. 5). Thus, the estimated effect of a 1 SD increase in motivation and effort ranges from 10% to 15% of the yearly learning gain in reading for 6th graders. Assuming that the length of a school year is roughly 10 months, this amounts to 1–1.5 months of learning. The corresponding result for the older cohort is larger: 2–3 months of learning. [This calculation is based on the coefficient estimate for year 8 of 0.077 SD (Table 6, col. 8) and estimates of average yearly learning gains from the literatures of 0.26 SD (Hill et al. 2008; Table 1, grade 7–8) and 0.40 SD (Hattie 2012).] These estimates average to about two months of learning for the two cohorts taken together.

17. The outcomes, the reading and math test scores, are measured roughly at the same time as the data collection of the 3rd student survey (in the spring of 2014), while the lagged test scores are all obtained well before the 1st data collection. Thus, pre-determined student engagement measures for use in the estimations are available from the 1st and 2nd survey (in the spring and autumn of 2013).

18. The only exception is for the reading profile area of decoding, where the estimate of participation in social activities is significant and similarly as important as participation in learning activities.

19. Math scores are observed only for the younger cohort (year 6) in this sample.

20. Note that participation in learning activities is not negatively related to test scores among SEN students. Rather, the net effect for SEN students is close to zero (0.078–0.069).

Additional information

Funding

This work was supported by the Danish Ministry of Education.