Diterima
Aleksander Kocaj, Poldi Kuhl, Malte Jansen, Hans Anand Pant, Petra
Stanat
Kutip artikel ini sebagai: Kocaj, A., Kuhl, P., Jansen, M., Anand Pant, H., Stanat, P., Penempatan
Pendidikan dan Motivasi Berprestasi Siswa Berkebutuhan Pendidikan Khusus, Psikologi Pendidikan
Kontemporer (2018), doi: https://doi.org/10.1016/ j.cedpsych.2018.09.004
Ini adalah file PDF dari manuskrip yang belum diedit yang telah diterima untuk publikasi. Sebagai
layanan kepada pelanggan kami, kami menyediakan naskah versi awal ini. Naskah akan menjalani
penyalinan, penyusunan huruf, dan peninjauan kembali bukti yang dihasilkan sebelum diterbitkan
dalam bentuk akhirnya. Harap dicatat bahwa selama proses produksi kesalahan dapat ditemukan yang
dapat mempengaruhi konten, dan semua penafian hukum yang berlaku untuk jurnal yang
bersangkutan.
MOTIVASI PENEMPATAN DAN PENCAPAIAN PENDIDIKAN 1
Aleksander Kocaj1, Poldi Kuhl1, Malte Jansen1, Hans Anand Pant1,2, dan Petra Stanat1
1
Institut untuk Peningkatan Kualitas Pendidikan, Humboldt-Universität zu
Berlin 2Humboldt-Universität zu Berlin
Penulis Catatan
Universität zu Berlin, Unter den Linden 6, 10099 Berlin, Jerman. Telp: +49 30 2093
Penelitian ini adalah bagian dari disertasi doktor. Data disediakan oleh Pusat Data
adalah bagian dari Pusat Penilaian Pelajar Internasional (ZIB). Selama mengerjakan
disertasinya, Aleksander Kocaj juga merupakan rekan pra-doktoral dari International Max
Planck Research School on the Life Course (LIFE, www.imprs life.mpg.de; institusi yang
Abstrak
biasa atau sekolah biasa berkaitan dengan motivasi berprestasi siswa berkebutuhan
pendidikan khusus (SEN). Lebih lanjut, kami memeriksa apakah lingkungan sosial kelas
(yaitu, prestasi rata-rata kelas dan dukungan sosial) menjelaskan potensi perbedaan
penempatan. Kami membandingkan akademik konsep diri dan kenikmatan belajar siswa SEN
di sekolah khusus pendidikan(n = 420) dan sekolah reguler(n = 678) pada akhir 4th. kelas
Sejalan dengan teori perbandingan sosial, siswa SEN di sekolah pendidikan luar biasa
melaporkan konsep diri akademis dan kesenangan belajar yang lebih tinggi daripada teman
sebayanya di sekolah biasa. Meneliti mekanisme yang mendasari, prestasi rata-rata kelas
berhubungan negatif dengan konsep diri akademik siswa SEN dan kesenangan belajar.
Sebaliknya, dukungan sosial yang dirasakan secara individu oleh teman sekelas - tetapi bukan
dukungan sosial rata-rata kelas - secara positif terkait dengan motivasi berprestasi siswa SEN.
Setelah mengontrol prestasi individu dan rata-rata kelas, tidak ditemukan hubungan antara
penempatan pendidikan siswa SEN dan motivasi berprestasi. Hasil penelitian menunjukkan
penempatan dalam konsep diri akademik siswa SEN dan kesenangan belajar.
1. Pendahuluan
Terpisah tetapi setara tidak pernah sama (Brown v. Board of Education, 1954). Namun
bagi siswa berkebutuhan pendidikan khusus (SEN), potensi keuntungan dan kerugian dari
sekolah terpisah yang dikhususkan untuk kebutuhan mereka masih diperdebatkan secara luas.
Bangsa-Bangsa, 2006), beberapa negara telah berkomitmen untuk menerapkan sistem sekolah
yang lebih inklusif dan meningkatkan proporsi siswa SEN di sekolah reguler. Meskipun
sekolah inklusif melibatkan sejumlah fitur organisasi, praktik pengajaran, dan kelompok
siswa, satu aspek utama terkait dengan penempatan pendidikan siswa dengan SEN.
Penempatan pendidikan mengacu pada jenis sekolah (misalnya, sekolah pendidikan khusus
vs. sekolah biasa) dan jenis ruang kelas tempat siswa SEN menerima pengajaran. Ada
perdebatan yang hidup di antara pembuat kebijakan, praktisi sekolah, dan peneliti pendidikan
(Hocutt, 1996; Lindsay, 2007): Haruskah siswa dengan SEN dididik di ruang kelas biasa atau
di sekolah atau ruang kelas terpisah yang secara khusus ditargetkan untuk kebutuhan mereka?
Tren ke arah sistem sekolah yang lebih inklusif dibarengi dengan penelitian intensif tentang
keefektifannya bagi siswa SEN (Bakker, Denessen, Bosman, Krijger, & Bouts, 2007;
Lindsay, 2007; Ruijs & Peetsma, 2009). Penelitian ini belum mengungkapkan hubungan yang
konsisten antara penempatan pendidikan siswa SEN dan aspek perkembangan sekolah
mereka. Berkenaan dengan hasil kognitif, sebagian besar studi menunjukkan tingkat
pencapaian siswa SEN yang lebih tinggi di sekolah biasa dibandingkan dengan rekan mereka
di sekolah pendidikan khusus (Kocaj, Kuhl, Kroth, Pant, & Stanat, 2014; Ruijs & Peetsma,
2009). 1 Namun, temuan tentang perbedaan penempatan untuk psikososial (Bakker et al.,
2007; Pijl & Frostad, 2010) dan hasil motivasi (Elbaum, 2002; Salend & Duhaney, 1999)
adalah
1
Studi yang dikutip tentang efek penempatan pendidikan pada hasil akademik siswa SEN dilakukan
terutama di negara-negara Eropa termasuk Jerman, Belanda, dan Norwegia.
MOTIVASI PENEMPATAN DAN PENCAPAIAN PENDIDIKAN 6
tidak meyakinkan atau cenderung mendukung sekolah pendidikan luar biasa. Akibatnya,
manfaat dan kerugian dari sekolah inklusif bagi siswa SEN tampaknya bergantung pada
perkembangan sekolah mereka (Elbaum, 2002; Grolnick & Ryan, 1990; Möller, Streblow, &
Pohlmann, 2009). Siswa dengan tingkat pendidikan khusus sangat berisiko untuk
mengalami kegagalan akademis daripada rekan-rekan mereka yang tidak memiliki sekolah
(Allodi, 2000; Grolnick & Ryan, 1990). Peetsma, Vergeer, Roeleveld, dan Karsten (2001)
dapat memicu lingkaran setan yang mencegah siswa SEN untuk mengatasi kesulitan belajar
mereka. Misalnya, siswa SEN dapat mengembangkan evaluasi diri negatif dan menganggap
kemampuan mereka sendiri tidak dapat diubah setelah mengalami kegagalan berulang kali di
sekolah (Deci & Chandler, 1986; Eccles, Wigfield, & Schiefele, 1998). Akibatnya, mereka
mungkin mengurangi upaya mereka dalam pekerjaan sekolah dan kegigihan mereka dalam
menyelesaikan tugas-tugas yang menuntut. Hal ini, pada gilirannya, dapat berdampak negatif
pada prestasi sekolah mereka selanjutnya (Deci & Chandler, 1986). Penelitian pada siswa
tanpa SEN mengungkapkan bahwa motivasi berprestasi rendah juga terkait dengan tingkat
yang lebih tinggi dari depresi, pelepasan, dan putus sekolah (Elbaum, 2002; Reyes, Brackett,
Rivers, White, & Salovey, 2012). Oleh karena itu, mempromosikan motivasi berprestasi siswa
Lingkungan belajar berbeda antara pengaturan pendidikan dan perbedaan ini (misalnya,
dalam kurikulum, metode pengajaran, dan komposisi siswa) dapat membentuk motivasi
berprestasi siswa (Urdan & Schoenfelder, 2006; Wang & Eccles, 2013). Meskipun penelitian
sebelumnya mengeksplorasi perbedaan penempatan dalam motivasi berprestasi siswa SEN
(Bear, Minke, & Manning, 2002; Chapman, 1988; Elbaum, 2002; Forman, 1988; Pijl &
Frostad, 2010), para peneliti jarang meneliti karakteristik lingkungan belajar. yang mungkin
mendasari perbedaan-perbedaan tersebut dalam. Tujuan dari penelitian ini adalah untuk
menyelidiki apakah motivasi berprestasi siswa SEN berbeda antara sekolah pendidikan luar
biasa dan sekolah biasa dan untuk memeriksa apakah lingkungan sosial kelas (yaitu,
penempatan.
pendidikan siswa SEN di sekolah pendidikan luar biasa atau di sekolah biasa dapat
mempengaruhi motivasi berprestasi mereka dalam berbagai cara. Para pendukung sekolah
pendidikan khusus berpendapat bahwa pengaturan ini menyediakan lingkungan belajar yang
terlindungi bagi siswa SEN. Ini termasuk kurikulum yang disesuaikan dan kurang menuntut,
umpan balik individual, dan iklim yang kurang kompetitif di kelas (Chapman, 1988; Peetsma
et al., 2001). Dari perspektif person-environment fit, siswa harus lebih termotivasi jika materi
pembelajaran disesuaikan dengan tingkat kompetensinya (Eccles, 2004; Vaughn, Elbaum, &
siswa SEN mengalami perasaan sukses. Selanjutnya, motivasi berprestasi siswa SEN dapat
diharapkan untuk mendapatkan keuntungan dari perbandingan dengan teman sebaya dengan
tingkat kinerja yang sama dan kesulitan yang serupa (Chapman, 1988). Perbandingan kinerja
dengan rekan-rekan yang berprestasi serupa harus mengurangi perasaan inferioritas siswa
SEN dan stres sekolah karena kecemasan evaluatif (Dijkstra, Kuyper, van der Werf, Buunk, &
reguler dapat merasa lebih dihargai sebagai pelajar yang mampu (Bakker & Bosman, 2003).
Hal ini, pada gilirannya, dapat meningkatkan motivasi berprestasi siswa SEN (Vaughn et al.,
1996; Wiener & Tardif, 2004). Menurut pandangan ini, guru di sekolah reguler memiliki
harapan yang lebih tinggi dan lebih menekankan pada kemajuan akademik (Diamond,
Hornstra, Denessen, Bakker, Bergh, & Voeten, 2010). Dengan menjadi bagian dari kelompok
belajar yang dihargai secara positif, siswa SEN di sekolah reguler dapat mengembangkan
motivasi berprestasi yang lebih tinggi melalui berjemur dalam pantulan kemuliaan prestasi
yang dirasakan rekan-rekan mereka (Marsh et al., 2000). Selain itu, siswa SEN di sekolah
reguler dapat memperoleh manfaat dari teman sekelas yang berprestasi lebih tinggi melalui
mekanisme pembelajaran sosial (Barth, Dunlap, Dane, Lochman, & Wells, 2004): Teman
sekelas yang berprestasi dapat berfungsi sebagai panutan bagi siswa SEN dan memulai
perbandingan ke atas terkait dengan peningkatan motivasi berprestasi (Corcoran, Crusius, &
Menurut teori nilai ekspektasi, dua anteseden utama dari motivasi berprestasi siswa adalah
konsep diri akademik dan kesenangan belajar (Nagengast et al., 2011; Wigfield & Eccles,
2000). Konsep diri akademik didefinisikan sebagai evaluasi diri siswa dalam domain atau
subjek akademik tertentu (Marsh, Kong, & Hau, 2000; Trautwein, Lüdtke, Marsh, Köller &
Baumert, 2006). Kesenangan belajar mencerminkan nilai intrinsik dan minat subjektif siswa
dalam tugas-tugas yang berhubungan dengan sekolah. Kedua variabel motivasi ini
mempengaruhi prestasi sekolah dan ketekunan tugas serta pilihan pendidikan dan aspirasi
karir (Eccles et al., 1983; Nagengast et al., 2011; Pekrun, Goetz, Titz, & Perry, 2002; Wang
Dalam sebagian besar studi, siswa SEN di kelas khusus atau sekolah pendidikan khusus
melaporkan konsep diri akademik yang lebih tinggi daripada siswa SEN dalam pengaturan
inklusif (Bear et al., 2002; Chapman, 1988; Crabtree & Meredith, 2000; Renick & Harter,
1989). Bear, Minke, dan Manning (2002) menyimpulkan dalam meta-analisis mereka bahwa
lingkungan belajar yang lebih ketat untuk siswa dengan ketidakmampuan belajar (yaitu, kelas
terpisah) dikaitkan dengan konsep diri akademik yang lebih tinggi. Namun, mereka juga
melaporkan heterogenitas yang cukup besar dalam ukuran efek di seluruh studi yang terkait
dialami. Sebaliknya, meta-analisis oleh Elbaum (2002) tidak menemukan hubungan yang
konsisten antara penempatan pendidikan dan konsep diri akademik siswa SEN. Peetsma dan
menerapkan desain longitudinal: Motivasi berprestasi guru siswa SEN secara signifikan
lebih tinggi di sekolah pendidikan luar biasa daripada di sekolah biasa. Selama dua tahun,
motivasi berprestasi siswa SEN di sekolah luar biasa juga mengalami penurunan yang tidak
begitu mencolok dibandingkan dengan di sekolah biasa. Namun, setelah empat tahun, tidak
motivasi siswa SEN, penelitian jarang mengeksplorasi karakteristik lingkungan sosial kelas
yang berpotensi mendasari perbedaan penempatan tersebut (misalnya, Nusser & Wolter,
2016). Lebih lanjut, mayoritas studi difokuskan pada konsep diri akademik siswa dengan
ketidakmampuan belajar. Meskipun beberapa studi memasukkan aspek lain dari motivasi
berprestasi (Fulk, Brigham, & Lohman, 1998; Grolnick & Ryan, 1990; Peetsma et al., 2001),
penelitian tentang komponen nilai intrinsik — seperti kenikmatan belajar — masih langka.
Sekolah pendidikan luar biasa dan sekolah biasa menyediakan lingkungan sosial yang
berbeda yang dapat berkontribusi pada perbedaan motivasi berprestasi siswa SEN. Penelitian
pada siswa tanpa SEN menunjukkan bahwa motivasi berprestasi dipengaruhi oleh lingkungan
sosial kelas (Patrick, Ryan, & Kaplan, 2007; Urdan & Schoenfelder, 2006; Wang & Eccles,
2013). Dua fitur penting dari lingkungan sosial kelas adalah perbandingan dan
Siswa SEN di sekolah reguler belajar bersama dengan teman sekelas yang berprestasi tanpa
SEN sedangkan siswa SEN di sekolah pendidikan luar biasa belajar bersama dengan siswa
SEN lain dengan kesulitan yang sama. Perbedaan antara sekolah pendidikan luar biasa dan
sekolah biasa pada tingkat prestasi teman sekelas dapat mengakibatkan perbedaan motivasi
berprestasi siswa SEN. Menurut teori perbandingan sosial, siswa membandingkan diri mereka
dengan teman sekelas untuk mendapatkan pandangan diri yang akurat dan untuk
menyesuaikan evaluasi diri mereka (Festinger, 1954; Marsh et al., 2008). Berdasarkan asumsi
ini, Marsh (1987) mendalilkan efek kolam besar-ikan-kecil (BFLPE; Marsh, 1987),
menunjukkan bahwa perbandingan sosial dalam kerangka acuan eksternal adalah sumber
utama konsep diri akademik siswa. Hipotesis BFLPE bahwa prestasi individu berhubungan
positif dengan konsep diri akademik, sedangkan pencapaian rata-rata kelompok referensi
berhubungan negatif dengan konsep diri akademik. Secara khusus, dengan prestasi individu
yang serupa, siswa di ruang kelas (atau sekolah) dengan teman yang berprestasi tinggi (kolam
besar) akan menunjukkan konsep diri akademik yang lebih rendah daripada siswa di kelas
Efek keseluruhan yang negatif dari prestasi rata-rata kelas ini mewakili efek bersih
dari dua proses penyeimbang: asimilasi dan kontras. Di satu sisi, efek asimilasi (juga disebut
sebagai efek pelabelan atau berjemur dalam kemuliaan yang dipantulkan) berkaitan dengan
kecenderungan siswa untuk membentuk konsep diri akademik yang sesuai dengan tingkat
kemampuan yang dirasakan rekan-rekan mereka (Corcoran et al., 2011; Dijkstra et al., 2008).
Menghadiri sekolah atau ruang kelas bergengsi dengan siswa berprestasi — seperti sekolah
reguler untuk siswa SEN — oleh karena itu harus mengarah pada konsep diri yang lebih
positif. Sebaliknya, keanggotaan dalam kelompok belajar yang berprestasi rendah atau
distigmatisasi — seperti sekolah pendidikan luar biasa — harus dikaitkan dengan konsep diri
Di sisi lain, efek kontras (juga disebut sebagai efek kelompok referensi) adalah hasil
dari perbandingan sosial yang menekankan ketidaksamaan antara diri sendiri dan kelompok
referensi lokal (Corcoran et al., 2011). Efek kontras terjadi jika konsep diri akademik siswa
ditingkatkan dengan tingkat pencapaian yang lebih rendah dan menurun dengan tingkat
pencapaian teman sekelas yang lebih tinggi (Marsh et al., 2000). Kelompok referensi yang
berprestasi rendah di sekolah pendidikan luar biasa dapat mengarahkan siswa SEN pada
evaluasi diri yang mereka lakukan sebaik teman sekelas mereka (Renick & Harter, 1989).
Perbandingan dengan teman sekelas yang berprestasi rendah ini meningkatkan konsep diri
akademis mereka. Sebaliknya, siswa SEN di sekolah biasa membandingkan prestasi mereka
dengan teman sekelas yang lebih mampu. Konsekuensinya, konsep diri akademis mereka
Untuk siswa tanpa SEN, bukti empiris dengan jelas menunjukkan bahwa efek kontras
negatif lebih kuat daripada efek asimilasi positif, yang mengarah ke BFLPE negatif secara
keseluruhan (Marsh et al., 2000). Meskipun sebagian besar studi tentang BFLPE berfokus
pada konsep diri akademis, terdapat beberapa bukti bahwa BFLPE juga menggeneralisasi
hasil motivasi lainnya termasuk kenikmatan belajar (Goetz et al., 2004), minat (Köller,
Trautwein, Lüdtke, & Baumert 2006; Schurtz, Pfost, Nagengast, & Artelt, 2014; Trautwein et
al., 2006), dan uji kecemasan (Zeidner & Schleyer, 1999). BFLPE pada hasil motivasi lainnya
telah terbukti dimediasi oleh konsep diri akademik (Köller et al., 2006; Trautwein et al.,
2006).
Patrick et al., 2007; Wang & Eccles, 2013). Dukungan sosial dapat digambarkan sebagai
persepsi siswa tentang kelas yang peduli di mana mereka mengalami afiliasi dan bantuan
oleh guru dan teman sekelas (Urdan & Schoenfelder, 2006; Wentzel, 1998).Individu Siswa
Persepsi dukungan sosial dibentuk oleh lingkungan sosial kelas yang meliputi interaksi
diadik antara siswa dan guru, interaksi kelompok, serta nilai dan harapan pembelajaran
bersama (Connell & Wellborn, 1991; Russell, 2012) ). Selain itu, lingkungan sosial kelas
dapat mempengaruhi motivasi berprestasi siswa di atas dan di atas dukungan sosial yang
dirasakan pada tingkat individu (Connell & Wellborn, 1991; Wentzel, Battle, Russell, &
Looney, 2010). Oleh karena itu, penting untuk mempertimbangkan indikator dukungan
Dukungan sosial mungkin berkontribusi pada konsep diri akademik dan kenikmatan belajar
dengan cara yang berbeda (Forman, 1988; Wentzel et al., 2010): Pertama, ruang kelas di mana
siswa mengalami dukungan dari guru dan teman sekelas harus memberikan kesempatan untuk
berpartisipasi dalam pelajaran tanpa kecemasan. diejek atau diejek (Patrick et al., 2007).
lingkungan belajar yang aman mendukung perasaan percaya diri dan harapan siswa untuk
sukses (Furrer & Skinner, 2003; Patrick et al., 2007). Dukungan sosial dengan demikian
berfungsi sebagai sumber untuk menjaga emosi positif terhadap pembelajaran (Furrer &
Skinner, 2003). Kedua, dukungan sosial dapat mengurangi efek merugikan dari pengalaman
stres terhadap motivasi berprestasi siswa (Allodi, 2000; Wentzel, 1998). Ketersediaan
dukungan sosial yang dirasakan secara khusus berfungsi sebagai penyangga terhadap
peristiwa stres (Demaray & Malecki, 2002; Furrer & Skinner, 2003). Siswa yang merasa
didukung seharusnya tidak terlalu rentan terhadap potensi penolakan atau kegagalan selama
diskusi atau ujian kelas. Mereka mungkin juga menghabiskan lebih sedikit waktu untuk
merenungkan tentang status sosial mereka di kelas dan sebaliknya memusatkan perhatian
mereka pada konten pembelajaran (Song, Bong, Lee, & Kim, 2015).
Siswa SEN di sekolah pendidikan luar biasa dan sekolah biasa mungkin berbeda dalam
persepsi dukungan sosial oleh teman sekelas, yang pada gilirannya dapat berkontribusi pada
perbedaan motivasi berprestasi mereka. Di satu sisi, siswa SEN mungkin mendapat manfaat
dari penempatan di sekolah pendidikan khusus. Karena siswa SEN di sekolah pendidikan
dengan teman sekelas dengan tingkat kesulitan dan pencapaian yang sama, status sosial
mereka mungkin kurang menonjol dibandingkan di sekolah biasa. Hal ini dapat menyebabkan
perasaan keterkaitan yang lebih kuat dan meningkatkan motivasi berprestasi siswa SEN.
Sebaliknya, kesulitan akademis siswa SEN di sekolah reguler ditemukan terkait dengan
tingkat dukungan sosial yang lebih rendah oleh teman sekelas (Bakker et al., 2007; Savage,
2005; Vaughn, Haager, Hogan, & Kouzekanani, 1992). Selain itu, kekurangan dalam
kompetensi sosial (Bauminger, Edelsztein, & Morash, 2005) dan masalah perilaku (Margalit
& Efrati, 1996; Vaughn et al., 1992) mungkin lebih menonjol bagi siswa SEN di sekolah
reguler yang menyebabkan penerimaan dan sosial yang lebih rendah. didukung oleh teman
sekelas. Di sisi lain, pendukung pendidikan inklusif berpendapat bahwa status SEN dapat
memiliki fungsi perlindungan bagi siswa SEN di sekolah reguler (Vaughn et al., 1992). Lebih
lanjut, penelitian menunjukkan bahwa teman sekelas di sekolah biasa memiliki kompetensi
sosial yang lebih tinggi dan menunjukkan tingkat perundungan yang lebih rendah daripada
teman sekelas di sekolah pendidikan khusus (Rose, Monda-Amaya, & Espelage, 2011) yang
harus dikaitkan dengan tingkat dukungan sosial yang lebih tinggi untuk siswa SEN. di sekolah
biasa.
Studi ini didasarkan pada data dari sistem pendidikan khusus Jerman. Jerman telah
menerapkan sistem pendidikan khusus dikotomis yang menempatkan siswa dengan SEN
baik di sekolah pendidikan khusus atau di sekolah reguler (Authoring Group Educational
Reporting, 2016; Powell, 2009; Sansour & Bernhard, 2018). Siswa dengan SEN di sekolah
reguler menghabiskan sebagian besar hari sekolah bersama dengan siswa tanpa SEN
(Authoring Group Educational Reporting, 2016). Meskipun tren menuju pendidikan yang
lebih inklusif disertai dengan bentuk organisasi yang lebih beragam dari dukungan khusus di
pendidikan dimulai dan diawasi oleh otoritas pengawas sekolah bekerja sama dengan orang
tua, guru, dan pendidik khusus (Sälzer, Gebhardt, Müller, & Pauly, 2015). Prestasi sekolah
yang rendah, masalah perilaku, dan indikasi lain yang menunjukkan bahwa siswa mungkin
memerlukan dukungan tambahan untuk mencapai tujuan pendidikan sering dijadikan titik
awal untuk penilaian kebutuhan pendidikan khusus (KMK, 2017; Sansour & Bernhard,
2018). Keputusan tentang penempatan pendidikan didasarkan pada kebutuhan individu siswa
serta sumber daya yang tersedia di sekolah (KMK, 2017; Powell, 2009; Sälzer et al., 2015).
Misalnya, siswa dengan kebutuhan pendidikan kurang khusus lebih cenderung dirujuk ke
Siswa SEN di sekolah reguler menerima akses ke kurikulum pendidikan umum dengan
dukungan tambahan oleh tenaga kependidikan (misalnya, pendidik khusus, asisten pengajar,
layanan keliling) untuk memfasilitasi inklusi di kelas reguler (KMK, 2017; Sansour &
Bernhard, 2018). Guru pendidikan khusus dapat bekerja sama dengan guru reguler selama
pelajaran (misalnya, mengajar bersama, Scruggs, Mastropieri & McDuffie, 2007) atau
dukungan individu diberikan setelah kelas untuk memantau kemajuan belajar siswa SEN dan
menyesuaikan tujuan pembelajaran jika perlu (KMK, 2017 ; Sansour & Bernhard, 2018).
Dukungan kebutuhan khusus di sekolah reguler selama pelajaran bersama lebih umum
2
Perhatikan bahwa sistem pendidikan khusus berbeda-beda di seluruh negara bagian Jerman karena
tanggung jawab ditentukan oleh struktur federal sistem pendidikan Jerman (KMK, 2017). Misalnya, di
beberapa negara bagian Jerman, siswa dengan ketidakmampuan belajar dapat menghadiri sekolah pendidikan
khusus di kelas tiga paling awal sedangkan di negara bagian Jerman lainnya, sekolah pendidikan khusus
dimulai di kelas satu (Sälzer et al., 2015).
PENEMPATAN PENDIDIKAN DAN MOTIVASI PENCAPAIAN 15
Siswa tunagrahita atau bahasa dan siswa dengan gangguan emosi di sekolah
pendidikan luar biasa menerima kurikulum dan tujuan pendidikan yang sama dengan siswa
di sekolah biasa (KMK, 2017). Namun metode dan materi pengajaran disesuaikan dengan
sekolah luar biasa menerima kurikulum alternatif dengan konten pelajaran dan persyaratan
kinerja yang lebih rendah dibandingkan dengan sekolah biasa (KMK, 2017). Di SLB,
dengan sekolah biasa (KMK, 2017). Juga, dukungan terapeutik (misalnya, terapi perilaku
kontras dan asimilasi pada motivasi berprestasi siswa SEN (Chmielewski, Dumont, &
Trautwein, 2013; Trautwein et al., 2006). Pertama, menempatkan siswa dengan SEN ke
dalam jenis sekolah yang berbeda menentukan kelompok referensi sosial mereka karena
mereka jarang berinteraksi dengan siswa dari jalur yang berbeda selama hari sekolah
mereka. Kedua, perbedaan status antara kelompok belajar lebih jelas dan terlihat bagi siswa
(Powell, 2009).
pendidikan siswa SEN dan motivasi berprestasi mereka. Selain itu, kami memeriksa apakah
lingkungan sosial kelas (yaitu, prestasi rata-rata kelas dan dukungan sosial) mendasari
yang lebih tinggi di sekolah pendidikan luar biasa, meta-analisis (Bear et al., 2002; Elbaum,
2002) mengungkapkan variabilitas yang cukup besar dalam besaran perbedaan penempatan.
Penjelasan untuk variabilitas dalam ukuran efek bisa jadi bahwa aspek lingkungan sosial
dalam
sekolah pendidikan dan sekolah reguler —lebih dari penempatan pendidikan itu sendiri—
terkait dengan motivasi berprestasi siswa SEN (Grolnick & Ryan , 1990; Ruijs, Peetsma, &
van der Veen, 2010). Telah dikemukakan bahwa penelitian harus memeriksa karakteristik
khusus dari lingkungan belajar yang mungkin menjelaskan perbedaan penempatan dalam
hasil siswa SEN (Lindsay, 2007; Zigmond, 2003). Dalam studi ini, kami membahas dua
pertanyaan penelitian:
1. Apakah siswa SEN di sekolah pendidikan luar biasa dan sekolah biasa berbeda
2. Apakah prestasi rata-rata kelas dan dukungan sosial rata-rata kelas terkait dengan
variasi dalam konsep diri akademik siswa SEN dan kenikmatan belajar di atas variasi
pendidikan luar biasa untuk melaporkan konsep diri akademik dan kesenangan belajar yang
lebih tinggi daripada siswa SEN di sekolah biasa. Sehubungan dengan pertanyaan penelitian
kedua, kami mengharapkan bahwa prestasi rata-rata kelas dan dukungan sosial rata-rata
kelas terkait dengan konsep diri akademik dan kenikmatan belajar siswa SEN di atas variasi
yang dijelaskan oleh jenis sekolah. Setelah mengontrol pencapaian individu, siswa SEN di
ruang kelas yang berprestasi lebih rendah harus melaporkan konsep diri akademik yang lebih
tinggi dan kesenangan belajar. Selain itu, siswa SEN yang merasa didukung secara sosial
oleh teman sekelasnya juga harus melaporkan konsep diri akademis yang lebih tinggi dan
kesenangan belajar. Selain itu, dukungan sosial rata-rata kelas tinggi harus meningkatkan
konsep diri akademik siswa SEN dan kesenangan belajar. Kami berhipotesis bahwa
pengaruh penempatan pendidikan tidak akan signifikan lagi ketika perbedaan antara sekolah
pendidikan luar biasa dan sekolah reguler dalam pencapaian rata-rata kelas dan dukungan
Kami menganalisis data dari studi penilaian penampang lintang perwakilan nasional
dari siswa kelas empat Jerman yang dilakukan pada musim semi 2011 (IQB National
Assessment Study 2011, Stanat, Pant, Böhme, & Richter, 2012; Stanat et al., 2014). Studi
tersebut bertujuan untuk mengukur tingkat kemahiran siswa dengan tes prestasi standar
mengumpulkan data tentang konsep diri akademik siswa dan kenikmatan belajar, persepsi
Dalam studi ini, kami memeriksa subsampel ruang kelas dengan setidaknya satu
siswa SEN. Sampel siswa yang dihasilkan terdiri dari 8692 siswa dari 451 sekolah (satu
kelas dipilih secara acak dari masing-masing sekolah, lihat Tabel 1) termasuk 1.098 siswa
dengan SEN (Usia rata-rata = 10,90 tahun, SD = 0,63; 32,7% perempuan; 24% siswa dengan
setidaknya satu orang tua lahir di luar negeri). Class size and number of SEN students
differed between special education schools and regular schools (see Table 1). In regular
schools, classrooms consisted of both students with and students without SEN. In contrast,
3
SEN categories overlap partially between Germany and the US (Gebhardt, Sälzer, Mang, Müller, &
Prenzel, 2015). SEN categories in Germany describe educational supports intended to help children with
disabilities advance in their school career and master the curriculum. Conversely, SEN categories in the US
focus on individual disabilities (Powell, 2009). In this article, we use the names of the US categories to align
with the international literature. However, there are subtle differences between the German and US categories
which should be noted. For example, a diagnosed dyslexia or dyscalculia alone is not sufficient for the
identification of SEN in Germany (Gebhardt et al., 2015).
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 18
learning disabilities show impairments in their learning and school achievement that can be
caused by a multitude of factors reaching from problems in basic psychological processes and
sensory impairment to social and emotional difficulties (KMK, 1999). Children with speech
and/or productive abilities (KMK, 1998). Students with emotional disorders experience
difficulties in their regulation of emotions and their social behavior in interactions with peers
and teachers (KMK, 2000). Students with functional or intellectual disabilities as well as non
native German language speakers who had attended German schools for less than one year
and had very low levels of language proficiency were not tested. This is in line with the
testing guidelines of other educational large-scale assessments (see Joncas & Foy, 2012;
OECD, 2014).
2.2 Measures
was measured with four items separately for German and mathematics (see Table 2 for item
wordings and descriptive statistics). The scale was adopted from the TIMSS 2007 study, an
international large-scale assessment in primary schools (Martin & Preuschoff, 2008), and had
been tested in field trials and other large-scale assessments demonstrating good measurement
properties and predictive validity (Martin, Mullis, & Foy, 2008; Mullis, Martin, Foy, &
item wording was influenced by the Self-Description Questionnaire I (SDQ I, Marsh, 1990).
excluded one item because of a low corrected part-whole correlation with the scale and a
small factor loading in confirmatory factor analyses. This item has a reverse phrasing which
may be difficult to understand for SEN students (Marsh, 1986; Nusser, Carstensen, & Artelt,
2015). The internal consistencies of both scales were reasonably high for SEN students in
special education schools (German: α = .71; mathematics: α = .79) and SEN students in
of achievement motivation in our study. The scale was developed by Pekrun (1992) and
consists of three items assessing students' global enjoyment in school lessons (see Table 2 for
item wordings and descriptive statistics). Scale construction was based on the control-value
(Wigfield & Eccles, 2000) in that it views competence appraisals (eg, academic self concept)
and value appraisals (eg, intrinsic value) as antecedents for achievement-related emotions
(Pekrun, Goetz, Frenzel, Barchfeld, & Perry, 2011). Furthermore, items of the enjoyment of
learning scale resemble items assessing intrinsic interest value (eg, “I enjoy school classes”,
Wigfield, 1994). The scale was used in previous large-scale assessments (Jerusalem,
Drössler, Kleine, Klein-Heßling, Mittag, & Röder, 2009) and is also part of an established
Questionnaire, Pekrun et al., 2011). In previous studies, the scale demonstrated good
measurement properties (Jerusalem et al., 2009; Pekrun et al., 2011) and predictive validity
(eg, positive relations to intrinsic motivation, effort, and achievement, Pekrun et al., 2011;
Pekrun, Goetz, Titz, & Perry, 2002). Students responded on a 4-point scale (1 = do not
agree; 4 = strongly agree). The scale showed a satisfactory internal consistency for SEN
students in special education schools (α = .78) and SEN students in regular schools (α = .73).
Perceived social support. Self-reported social support was assessed with four items
(see Table 2 for item wordings and descriptive statistics). The scale focuses on emotional
questionnaire (FEESS 3-4, Rauer & Schuck, 2003) assessing children's emotional and social
school experiences in primary school. Scale construction has drawn from self-determination
theory which emphasizes the important role of warm and positive relationships in the
classroom for students' motivation (Deci & Chandler, 1986). The scale has been validated in
prior studies including students with SEN (Rauer & Schuck, 2003). It is widely used in
al., 2015), and studies on students with specific learning disabilities or learning difficulties
(Fischbach, Schuchardt, Mähler, & Hasselhorn, 2010). The scale demonstrated good
measurement properties (Schwab, 2015; Spörer et al., 2015) and plausible correlational
patterns with relevant student outcomes (eg, well-being, academic self-concept, and
achievement, Rauer & Schuck, 2003) in previous studies. Students responded on a 4-point
scale (1 = do not agree; 4 = strongly agree). The scale showed an internal consistency of α =
.54 for SEN students in special education schools and α = .68 for SEN students in regular
schools.
comprehension were used to assess students' levels of proficiency (for a detailed description,
see Stanat et al., 2012). Test items are based on national educational standards for students at
the end of 4th grade (Böhme & Bremerich-Vos, 2012; Roppelt & Reiss, 2012). These
standards define skills and competencies in form of educational goals that students should
have acquired by a certain point in their school career (Stanat et al., 2012). Although German
states differ in their educational system, the educational standards provide mandatory
guidelines for curricula development and skills that should be taught in all schools (KMK,
2017). According to the educational standards for mathematics, students at the end of 4th grade
solve real-life problems from different content areas (eg, number & operations, algebra,
geometry, measurement, data analysis & probability, for details, see Roppelt & Reiss, 2012).
Educational standards for the school subject German focus on multidimensional aspects of
reading comprehension as well as skills and competencies in listening, speaking, and writing
comprehend age-appropriate narrative and expository texts (Böhme & Bremerich-Vos, 2012).
This includes the ability to extract and report central messages from a text, to make
connections between different information of a text and connect them to prior knowledge, and
to draw conclusions about the intention of a text (Böhme & Bremerich-Vos, 2012).
A large sample of items was used to cover the content of each test domain. Applying a
multiple matrix design (Gonzalez & Rutkowski, 2010), students only respond to a subset of
those items. Test items were distributed across test booklets and those test booklets were
assigned randomly to students (for details, see Weirich, Haag, & Roppelt, 2012). This design
increases the content coverage of the test construct while decreasing the test burden for
students, allowing for accurate proficiency estimations on a group level (Gonzalez &
Rutkowski, 2010).
Nusser, Artelt, & Carstensen, 2013; LeRoy, Samuel, Deluca, & Evans, 2018), two test
accommodations were used for students in special education schools. First, students in special
education schools received test booklets for which pretests showed that they were on average
easier than test booklets in regular schools. This accommodation was used to increase the fit
between item difficulties and anticipated student abilities leading to better measurement
properties of the achievement tests (Lane & Leventhal, 2015). It is important to note that the
test content was not changed for students in special education schools as items from those test
booklets were drawn from a common item pool and were therefore also administered to
Second, test booklet length and testing time was shorter in special education schools (40
minutes instead of 80 minutes per domain). Reducing overall testing time might help
students in special education schools to take the achievement tests because they might have
steeper attention declines and lower abilities to sustain concentration over time (Heydrich et
al., 2013). As test booklet length was also decreased, the average time given to answer test
items was comparable for SEN students in special education schools and regular schools.
To evaluate the structural validity and comparability of the achievement tests for SEN
students in special education schools and regular schools, missing patterns, Rasch model fit,
differential item functioning, and correlations of the test scores were analyzed. Items used in
both special education schools and regular schools had similar psychometric properties. The
results suggest that test scores for both groups are comparable and can be reported on a
common scale (for details, see Kocaj et al., 2016). Achievement tests for reading
comprehension and mathematics were scaled using a 1-parameter logistic item response
theory (IRT) model (Rasch model). The expected a-posteriori (EAP) reliabilities of the test
scores were comparable for SEN students in special education schools (reading
comprehension: EAP reliability = .75; mathematics: EAP reliability = .89) and SEN students
reliability = .92). Weighted likelihood ability estimates were used as achievement scores in
Prior to addressing our research questions, we checked the fit of the measurement model for
each motivational outcome separately by conducting confirmatory factor analyses (CFAs).
First, for SEN students in each school setting (special education school vs. regular school), a
two-level measurement model with equal factor loadings on the student level (L1) and on the
classroom level (L2) was specified. Second, we tested for measurement invariance
for each motivational outcome between SEN students in special education schools and
regular schools. Comparable measurement models and measurement invariance are necessary
requirements for valid comparisons of the latent factor means across both student groups
1998-2015) to predict SEN students' academic self-concept and enjoyment of learning. Both
dependent variables were modeled as latent variables (Lüdtke et al. 2008). School
support were entered in the model as manifest predictors on the individual level; school type
(special education school vs. regular school), class-average achievement, and class-average
social support were entered as manifest predictors on the classroom level. We included
individual achievement and individual perceived social support as predictors for three
SEN students in special education schools and regular schools, and c) to estimate the
students' motivational outcomes (Marsh et al., 2008). Furthermore, gender was included as a
covariate on the individual level because of its association with motivational outcomes
(Frenzel, Pekrun, & Goetz, 2007; Marsh, Trautwein, Lüdtke, Köller, & Baumert, 2005;
step, the aggregation of class-average achievement and class-average social support was based
on the unstandardized values of all students in the classroom. That is, in regular schools, the
achievement and perceived social support of students without SEN are also
included in the class-average variables. In a second step, only students with SEN were
analyzed in the multilevel analyses (for a similar approach, see Ruijs et al., 2010). The class
Multilevel regression analyses were specified as random intercept models using the robust
maximum likelihood (MLR) estimator. Missing data (ranging from 22.1% to 26.1% for the
motivational variables, from 23.0% to 25.2% for the social support variables, and from 2.6%
to 4.2% for the school achievement tests) were handled by applying full information
achievement and class-average social support (for a similar approach, see Trautwein et al.,
2006). First, school type (Model 1 in Tables 4 - 6) and class-average achievement (Model 2 in
learning. In Model 3 (Tables 4 - 6), school type and class-average achievement were entered
variation in motivational outcomes over and above the variation explained by school type. In
Model 4 (Tables 4 - 6), social support was entered separately as predictor of academic self-
concept or enjoyment of learning. In Model 5 (Tables 4 - 6), class-average social support was
with school type. In Model 6 (Tables 4 - 6), social support and achievement were included
simultaneously with school type as predictors in order to examine their specific contributions
Prior to the analyses, two CFAs were conducted for SEN students in special education
schools and SEN students in regular schools for each motivational outcome separately. The
measurement models showed a reasonable fit for each group (see Model 0 in Appendix A1 -
A3). Next, we tested for measurement invariance (see Model 1 - 4 in Appendix A1 - A3 for
the fit indices). We established partial invariance for each motivational outcome respectively
by allowing the intercept of one item to differ between SEN students in special education
schools and regular schools (see Model 1 - 4 in Appendix A1 - A3). For each motivational
outcome, there were three factor loadings and at least two intercepts that were constrained
equal across groups. This allows valid inferences about differences in the latent means
between groups (Byrne, Shavelson, & Muthén, 1989; Steinmetz, 2013). We refrain from
comparing the manifest factor means, however, because full scalar equivalence could not be
Support
pattern of correlations between school type, achievement, and social support (Table 3). As
expected, there were substantial negative associations between educational placement and
education schools (r = –.83, p < .001 for reading comprehension and r = –.81, p < .001 for
mathematics achievement, Table 3). The same pattern of results emerged for the individual
level (r = –.46, p < .001 for reading comprehension and r = –.51, p < .001 for mathematics,
Table 3). The correlation between class-average social support and educational placement also
favored regular schools but it was considerably lower than for achievement (r = –.31, p <
social support on the individual level (r = .08, p = .20). These correlations indicate that SEN
also perceive higher levels of social support than SEN students in special education schools.
Results of the multilevel analyses are presented in Tables 4, 5, and 6 for academic self-
address our first research question, academic self-concept and enjoyment of learning were
regressed on school type, individual achievement, and gender (Model 1, Tables 4 - 6). When
0.31, p < .001) and mathematics (b = 0.48, p < .001) were higher for SEN students in special
education schools than for their peers in regular schools. SEN students in special education
schools also reported higher enjoyment of learning than SEN students in regular schools after
taking individual differences in reading comprehension and gender differences into account
(b = 0.30, p < .001).4These results reflect the anticipated placement differences in favor of
academic self-concept and enjoyment of learning (Tables 4 - 6). The regression coefficients
of this model were in accordance with the expected BFLPE: individual achievement was
4
Analyses with achievement in mathematics as predictor yielded similar results and are presented in
Appendix A, Table A4.
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 27
related to SEN students' enjoyment of learning (Model 2, Table 6). After individual
differences in achievement and gender differences were taken into account, the regression
p < .001) —that is, the BFLPE was detected (albeit weaker, ES2 = –0.40) not only for
In Model 3, school type and characteristics of the classroom social environment were
–0.16, p = .008; mathematics: b = –0.32, p < .001), whereas the regression coefficients of
school type were no longer significant (German: b = 0.09, p = .26; mathematics: b = 0.07, p
= .34). Similar results were obtained for enjoyment of learning (Model 3, Table 6):
Controlling for individual reading comprehension and gender, SEN students reported lower
5
ES2 is a measure of effect size in multilevel structural equation modeling and comparable to Cohen's
d for continuous class-level predictors (Lüdtke et al., 2008; Marsh et al., 2009). ES2 was computed according to
Marsh et al. (2009): ES2 = (2 * B * SDx)/Var(y), where B is the unstandardized regression coefficient of the
class-level predictor (ie, class-average achievement), SDx is the standard deviation of the class-level predictor
(ie, class-average achievement) and Var(y) is the total variance of the individual-level criterion (ie, academic
self-concept) (Marsh et al., 2009).
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 28
individual level (L1), and class-average social support (L2) were entered as predictors of
SEN students' academic self-concept and enjoyment of learning. SEN students' individual
level perceived social support was positively related to their academic self-concept (German:
<.001). In contrast, class-average social support was not related to SEN students' academic
In Model 5 (Tables 4 - 6), gender and perceived social support by classmates on the individual
level (L1), and class-average social support and school type (L2) were entered as predictors of
SEN students' academic self-concept and enjoyment of learning. SEN students' individual-
level perceived social support was positively related to their academic self-concept in both
average social support was not significantly related to academic self-concept (German: b = –
0.02, p = .97; mathematics: b = 0.06, p = .17) after accounting for the positive effect of
b = 0.27, p <.001). SEN students with higher levels of perceived social support also reported
education schools reported more enjoyment of learning than their peers in regular schools (b =
0.24, p = .001). Class-average social support had no additional effect on SEN students'
class-average social support could not explain variations in SEN students' academic self
concept and enjoyment of learning over and above the variation explained by school type.
simultaneously with gender and school type as predictors of academic self-concept and
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 29
enjoyment of learning. On the individual level, higher levels of both perceived social support
–0.16, p = .004; mathematics: b = –0.34, p <.001) predicted SEN students' academic self
and class-average social support (German: b = –0.01, p = .87; mathematics: b = 0.07, p = .07)
had no additional effect on academic self-concept. With regard to enjoyment of learning, both
students with higher achievement in reading comprehension (b = 0.12, p = .001) and students
with higher perceived social support (b = 0.20, p < .001) on the individual level reported
higher enjoyment of learning (Model 6, Table 6). Similar to academic self-concept, class
school type (b = 0.08, p = .44) nor class-average social support (b = 0.08, p = .08) were
significantly associated with SEN students' enjoyment of learning (Model 6, Table 6). These
differences; class-average social support does not additionally describe differences in SEN
educational placement and motivational outcomes for students with learning disabilities and
students with speech or language impairment (see Appendix B). We also applied propensity
score matching to control for a multitude of potentially confounding variables influencing the
relationship between educational placement and SEN students' achievement motivation which
The purpose of the study was to examine how educational placement and
characteristics of the classroom social environment are related to SEN students' achievement
motivation. Our findings support previous research showing that SEN students in special
education schools report a more positive academic self-concept and higher levels of
enjoyment of learning than SEN students in regular schools (Bear et al., 2002; Chapman,
1988; Renick & Harter, 1989). After taking differences in class-average achievement into
Our results are in line with social comparison theory and research on the big-fish little-
pond effect (Marsh et al., 2000). SEN students in special education schools can compare their
achievement with that of low-achieving classmates which benefits their academic self concept
and enjoyment of learning. This contrast effect—being a big fish in a small pond— seems to
be more important than the potential negative labeling effect of being placed in a stigmatized
school type (ie, assimilation effect). The effect sizes of the negative relationship between
achievement are similar to those from samples of students without SEN (for a review, see
Also, the absence of assimilation effects in our study reflects previous findings on the
(Chmielewski et al., 2013; Trautwein et al., 2006). The interpretation of the present findings,
school type which makes it difficult to separate assimilation and contrast effects as described
by social comparison theory (see Table 3). A direct assessment of assimilation, ie, asking
SEN students about the perceived standing or prestige of their school would help to separate
contrast and assimilation effects (Trautwein, Lüdtke, Marsh, & Nagy, 2009).
emotional support from peers. The perceived achievement of classmates seems to be the
differences in achievement motivation. Our results indicate that class-average social support
by classmates does not alleviate the negative effect of classmates' perceived achievement
level on SEN students' achievement motivation (but see Allodi, 2000). On the individual
level, however, higher perceived social support by peers contributed to higher academic self
concept and enjoyment of learning. This finding is consistent with results from previous
studies (eg, Forman, 1988) and might have several reasons: First, perceived social support
and associated feelings of relatedness can help SEN students to deal with challenging
situations in school and to maintain their motivation (Furrer & Skinner, 2003). The
against stressful events in school than the class-average support structures (Furrer & Skinner,
could lead to positive self-perceptions and to active participation in school lessons (Grolnick
The present study extends previous studies and contributes to the research field in the
(Bakker et al., 2007; Bear et al., 2002; Crabtree & Meredith, 2000; Elbaum, 2002; Forman,
1988) or on class-average achievement (Dixon, Seaton, & Dixon, 2008) to predict SEN
achievement allowed us to separate the negative assimilation effect of being placed in special
education schools from the positive contrast effect of being educated together with students
with lower achievement levels in special education schools. Furthermore, we extend prior
outcome besides academic self-concept as well as class-average social support besides class
average achievement.
Second, our sample was drawn from Germany where the majority of states implement
a dichotomous special education system (see Section 1.3). Placing students with SEN to
different school types based on prior achievement or abilities is considered to be the most
students from different tracks during the school day (Chmielewski et al., 2013). Furthermore,
placing students with SEN in either special education schools or regular schools has a strong
impact on their future educational outcomes: SEN students in special education schools
Educational Reporting, 2016; Pfahl & Powell, 2011). Taken together, tracking and status
differences between school types are highly visible for students, teachers, and parents in the
German special education system. This differentiating system facilitates the examination of
contrast and assimilation effects (Trautwein et al., 2006). Therefore, the present study adds to
the debate on tracking and strengthens the external validity of the BFLPE for students with
SEN.
Third, our analyses are based on a large sample of classrooms and students with SEN from a
score matching (Rosenbaum & Rubin, 1983, 1985) with numerous covariates (eg, individual
statistically for potential selection biases (Appendix C) as SEN students cannot be randomly
assigned to special education schools or regular schools (Lindsay, 2007). In our analyses, we
natural frame of reference for SEN students (Marsh et al., 2008). This approach, together with
the large sample size and inclusion of covariates in our analyses, allows for a more accurate
if placement effects generalize to both students with learning disabilities and students with
educational placement and motivational outcomes differs between students with learning
Several methodological limitations of our study are worth mentioning. First, questionnaires
in the present educational large-scale assessment were developed for students without SEN.
In the present study, measures of academic self-concept and social support showed lower
internal consistencies for students with SEN than for the student population without SEN (see
Richter, Böhme, Bastian-Wurzel, Pant, & Stanat, 2014). The moderate internal consistency
of the social support scale as one central predictor in our study limits the interpretation of our
results. This is even more relevant for aggregated variables of individual social support. The
aggregation of unreliable measures on the individual level could lead to spurious effects of
the corresponding measures on the classroom level (Televantou et al., 2015). Additionally,
therefore be partly due to differences in the measurement quality between both groups.
achievement motivation and social support (approximately 25%). One reason for lower
response rate might be that students participated voluntary in the questionnaires. In contrast,
participation in the achievement tests was mandatory. While this is a global explanation for
higher missing rates, students with SEN also showed higher missing rates than students
without SEN. This might be due to higher levels of fatigue and distraction as well as reduced
attentional capacities (Händel, Artelt, & Weinert, 2013; Heydrich et al., 2013; Nusser,
assessments might contribute to lower response rates for students with SEN (Marsh, Tracey,
& Craven, 2006; Nusser et al., 2016). Another potential explanation is that the assessment
situation might differ from their regular school routines. For example, SEN students can
receive additional support during instruction by special needs teachers or teaching assistants
Related to that, data collection during the assessment session could have influenced students'
self-reported motivation compared to assessments during a regular school day. Self reports of
academic self-concept have shown to be influenced by item wording and the assessment
context (Dijkstra et al., 2008; Marsh et al., 2008). For example, data collection during
assessment sessions might bias students' self-evaluations towards a stronger reliance on their
school grades, but also classroom discussions, group work, and presentations during class)
might be considered more strongly by students in assessments during regular school days
anonymously (eg, teachers do not receive feedback about students' responses) and students
do not receive any rewards or grades for their participation which might decrease social
Third, our study is limited by the cross-sectional design and it is therefore not possible
to draw conclusions about the directionality of our results. We proposed that social support
Although students who feel rejected might develop lower academic self-concept and
that students with negative self-concepts and low enjoyment of learning tend to refrain from
social interactions and activities in the classroom (Pijl & Frostad, 2010). Another limitation of
the cross-sectional design is that we were only able to examine between-person relationships
and interpret differences between groups. Longitudinal studies are necessary to examine
after experiencing social support by peers. Longitudinal designs would also allow for testing
reciprocal effects between SEN students' achievement motivation and social support (Skinner
& Belmont, 1993). Finally, cross-sectional studies tend to overestimate placement differences
and the effects of classroom composition because researchers cannot adequately control for
differences in prior motivation, achievement, and the selectivity of different school types
In the present study, we focused on emotional social support by classmates. Future studies
should include measures of social support by teachers, parents, and friends and examine their
(Patrick et al., 2007; Song et al., 2015) as well as differential effects of emotional and
academic components of social support (Wentzel et al., 2010). We would expect that
social support by teachers, parents, and friends play a distinctive role for SEN students'
achievement motivation (Song et al., 2015; Wentzel et al., 2010). Furthermore, academic
support, such as help and instruction to meet educational expectations, might differ from
2010).
Despite its limitations, the findings of the present study contribute to the empirical
debate on educational placement of students with SEN. On the one hand, educating SEN
students in special education schools together with students with similar challenges seems to
be beneficial for their achievement motivation. On the other hand, the majority of the studies
reports cognitive benefits for SEN students in regular schools (Ruijs & Peetsma, 2009).
Previous analyses with the present data also revealed higher achievement levels for SEN
students in regular schools compared to special education schools (Kocaj et al., 2014). These
diverging results regarding motivational and cognitive outcomes lead to the question why
motivational benefits of special education schools are not reflected in higher achievement
levels for SEN students? Or, as Marsh and Parker (1984) put it: “Is it better to be a relatively
large fish in a small pond even if you don't learn to swim as well?”
A potential explanation for these opposite placement effects might be that social
comparisons are based on different motives that affect motivational and cognitive outcomes
al., 2008). Self-improvement motives, in contrast, might lead to upward comparisons with
2008). Higher-achieving classmates may serve as role models and may positively affect
students' persistence and effort (Gamoran, 1986; Slavin, 1996). However, the downsides of
upward comparisons are lower self-evaluations and negative affective consequences (Dijkstra
et al., 2008).
Overall, there seems to be a positive net effect of attending a regular school on SEN
students' achievement. The positive direct effect on SEN students' achievement outweighs
the negative indirect effect on SEN students' achievement motivation. These contrasting
effects are also reflected in the opposed relationship of class-average achievement with SEN
levels promote SEN students' individual achievement (Justice, Logan, Lin, & Kaderavek,
2014), social comparisons with more able classmates lead to lower levels of achievement
motivation. However, more longitudinal studies considering different aspects of the learning
environment in special education schools and regular schools are necessary to shed light on
the interplay between SEN students' motivational and cognitive development (Bakker et al.,
2007).
Conclusions about the optimal placement for SEN students are difficult to draw because
examining the learning environment in special education schools could help to promote SEN
students' achievement motivation in inclusive school settings. SEN students might benefit
from a less competitive classroom environment where teachers do not emphasize social
comparisons but instead rely on individually oriented feedback with a focus on students'
individual improvement over time (Chapman, 1988, Lüdtke, Köller, Marsh, & Trautwein,
goals, learning tasks, and teaching materials matching students' achievement levels) might
also promote SEN students' achievement motivation (Deci & Chandler, 1986; Eccles, 2004;
McLeskey &Waldron, 2002; Roy, Guay, & Valois, 2015). Roy and colleagues (2015) found
adaptations for students with SEN (Scruggs, Mastropieri, & Marshak, 2012).
A central question for school practitioners is how teachers, school administrators, and
parents can cooperate to create a learning environment that fosters students' academic
progress and achievement motivation. Studies on inclusive school development highlight the
environments (Booth & Ainscow, 2011; Downes, Nairz-Wirth, & Rusinaitė, 2017). This
inclusive school culture includes shared beliefs to value diversity in school, to enhance
students' participation, and to acknowledge students' individual strengths and needs (Downes
et al., 2017; Dyson, Howes, & Roberts, 2002). Additionally, schools might implement
inclusive policies that revolve around reducing structural barriers to learning and
participation and around promoting cooperative classroom structures (eg, increase school
staff collaboration, Booth & Ainscow, 2011; Dyson et al., 2002). Finally, schools might
change practices towards a more holistic approach to students' development and emphasize
the role of positive relationships and social support for students' motivation and school
success (Dyson et al., 2002; Downes et al., 2017; Forman, 1988). This might be especially
relevant for students with SEN because high levels of social support may help them to
compensate for academic difficulties and to preserve a positive self-view (Allodi, 2000;
climate where contributions of every student are valued and opportunities for cooperative
References
Aguinis, H., Gottfredson, RK, & Culpepper, SA (2013). Best-practice recommendations for
Bakker, JTA, & Bosman, AMT (2003). Self-image and peer acceptance of Dutch students
in regular and special education. Learning Disability Quarterly, 26, 5–14. doi:
10.2307/1593680
Bakker, JTA, Denessen, E., Bosman, AMT, Krijger, E.-M., & Bouts, L. (2007).
Sociometric status and self-image of children with specific and general learning
Barth, JM, Dunlap, ST, Dane, H., Lochman, JE, & Wells, KC (2004). Classroom
Baumert, J., Blum, W., Brunner, M., Dubberke, T., Jordan, A., Klusmann, U., … Tsai, Y.-M.
Baumert, J., Stanat, P., & Watermann, R. (2006). Schulstruktur und die Entstehung
Bauminger, N., Edelsztein, HS, & Morash, J. (2005). Social information processing and
emotional understanding in children with LD. Journal of Learning Disabilities, 38, 45–
Bear, GG, Minke, KM, & Manning, MA (2002). Self-concept of students with learning
Böhme, K., & Bremerich-Vos, A. (2012). Die im Ländervergleich 2011 in den Fächern
Deutsch und Mathematik untersuchten Kompetenzen: Beschreibung der im Fach
Schülerinnen und Schülern am Ende der vierten Jahrgangsstufe in den Fächern Deutsch
Waxmann.
Booth, T., & Ainscow, M. (2011). Index for inclusion: Developing learning and participation
Byrne, BM, Shavelson, RJ, & Muthén, B. (1989). Testing for the equivalence of factor
Chmielewski, AK, Dumont, H., & Trautwein, U. (2013). Tracking effects depend on
10.3102/0002831213489843
Corcoran, K., Crusius, J., & Mussweiler, T. (2011). Social comparison: Motives, standards,
Crabtree, JW, & Meredith, C. (2000). Self-concept and social comparisons in learning
disabled students attending mainstream and special schools: Does integration have an
practice: Advances for the new millennium (pp. 187–193). Sydney: SELF Research
Centre, University of Western Sydney.
Deci, EL, & Chandler, CL (1986). The importance of motivation for the future of the LD
Demaray, MK, & Malecki, CK (2002). The relationship between perceived social support and
maladjustment for students at risk. Psychology in the Schools, 39, 305–316. doi:
10.1002/pits.10018
Diamond, JB, Randolph, A., & Spillane, JP (2004). Teachers' expectations and sense of
responsibility for student learning: The importance of race, class, and organizational
10.1525/aeq.2004.35.1.75
Dicke, T., Marsh, HW, Parker, PD, Pekrun, R., Guo, J., & Televantou, I. (2018). Effects of
Dijkstra, P., Kuyper, H., van der Werf, G., Buunk, AP, & van der Zee, YG (2008). Social
doi: 10.3102/0034654308321210
Dixon, RM, Seaton, M., & Dixon, RJ (2008). The big fish strikes again but in a different
place: Social comparison theory and children with special needs. International Journal
Downes, P., Nairz-Wirth, E., & Rusinaitė, V. (2017). Structural indicators for inclusive
systems in and around schools, NESET II report. Luksemburg: Kantor Publikasi Uni
Eropa.
Duncan, GJ, Magnuson, KA, & Ludwig, J. (2004). The endogeneity problem in
Dyson, A., Howes, A., & Roberts, B. (2002). A systematic review of the effectiveness of
Education.
Eccles (Parsons), J., Adler, TE, Futterman, R., Goff, SB, Kaczala, CM, Meece, JL, &
Festinger, L. (1954). Sebuah teori proses perbandingan sosial. Human Relations, 7, 117–140.
doi: 10.1177/001872675400700202
Fischbach, A., Schuchardt, K., Mähler, C., & Hasselhorn, M. (2010). Zeigen Kinder mit
Forman, EA (1988). The effects of social support and school placement on the self-concept
Fulk, BM, Brigham, FJ, & Lohman, DA (1998). Motivation and self-regulation: A
comparison of students with learning and behavior problems. Remedial and Special
Furrer, C., & Skinner, E. (2003). Sense of relatedness as a factor in children's academic
doi: 10.1037/0022-0663.95.1.148
Ganzeboom, HBG, De Graaf, PM, & Treiman, DJ (1992). A standard international socio-
economic index of occupational status. Social Science Research, 21(1), 1–56. Gebhardt,
M., Sälzer, C., Mang, J., Müller, K., & Prenzel, M. (2015). Performance of students with
Student Assessment 2012. Journal of Cognitive Education and Psychology, 14, 343–356.
doi: 10.1891/1945-8959.14.3.343
Goetz, T., Pekrun, R., Zirngibl, A., Jullien, S., Kleine, M., vom Hofe, R., & Blum, W. (2004).
Gonzalez, EJ, & Rutkowski, L. (2010). Principles of multiple matrix booklet designs and
Händel, M., Artelt, C., & Weinert, S. (2013). Assessing metacognitive knowledge:
Online, 5, 162–188.
Hansen, BB (2004). Full matching in an observational study of coaching for the SAT.
10.1198/016214504000000647
Heller, KA, & Perleth, C. (2000). KFT 4-12+R: kognitiver Fähigkeitstest für 4. bis 12.
Heydrich, J., Weinert, S., Nusser, L., Artelt, C., & Carstensen, CH (2013). Including
Challenges and approaches within the German National Educational Panel Study
Ho, DE, Imai, K., King, G., & Stuart, EA (2011). MatchIt: Nonparametric preprocessing for
parametric causal inference. Journal of Statistical Software, 42(8), 1–28. Hocutt, AM (1996).
Effectiveness of special education: Is placement the critical factor? The Future of Children,
Hornstra, L., Denessen, E., Bakker, J., van den Bergh, L., & Voeten, M. (2010). Teacher
attitudes toward dyslexia: Effects on teacher expectations and the academic achievement
10.1177/0022219409355479
Jerusalem, M., Drössler, S., Kleine, D., Klein-Heßling, J., Mittag, W., & Röder, B. (2009).
Joncas, M., & Foy, P. (2012). Sample design in TIMSS and PIRLS. In MO Martin & IVS
Mullis (Eds.), Methods and procedures in TIMSS and PIRLS 2011 (pp. 1–21). Chestnut Hill,
Justice, LM, Logan, JAR, Lin, T.-J., & Kaderavek, JN (2014). Peer effects in early
KMK = Sekretariat der Ständigen Konferenz der Kultusminister der Länder in der
http://www.kmk.org/fileadmin/pdf/PresseUndAktuelles/2000/sprache.pdf
KMK = Sekretariat der Ständigen Konferenz der Kultusminister der Länder in der
http://www.kmk.org/fileadmin/pdf/PresseUndAktuelles/2000/sopale.pdf
KMK = Sekretariat der Ständigen Konferenz der Kultusminister der Länder in der
http://www.kmk.org/fileadmin/pdf/PresseUndAktuelles/2000/emotsozentw.pdf
KMK = Sekretariat der Ständigen Konferenz der Kultusminister der Länder in der
bis 2014 [Special educational needs support in schools from 2005 to 2014]. Retrieved
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 47
from https://www.kmk.org
/fileadmin/Dateien/pdf/Statistik/Dokumentationen/Dok_210_SoPae_2014.pdf KMK =
Sekretariat der Ständigen Konferenz der Kultusminister der Länder in der Bundesrepublik
Deutschland. (2017). The education system in the Federal Republic of Germany 2014/2015.
A description of the responsibilities, structures and developments in education policy for the
https://www.kmk.org/fileadmin/Dateien/pdf/Eurydice/Bildungswesen-engl
pdfs/dossier_en_ebook.pdf
Kocaj, A., Haag, N., Weirich, S., Kuhl, P., Pant, HA, & Stanat, P. (2016). Aspekte der
Testgüte bei der Erfassung schulischer Kompetenzen von Schülerinnen und Schülern
Schulische Inklusion. 62. Beiheft der Zeitschrift für Pädagogik (p. 212–234).
Kocaj, A., Kuhl, P., Kroth, AJ, Pant, HA, & Stanat, P. (2014). Wo lernen Kinder mit
zwischen Regel- und Förderschulen in der Primarstufe [Where do children with special
educational needs learn better? A comparison between regular primary schools and
special schools]. Kölner Zeitschrift für Soziologie und Sozialpsychologie, 66, 165–191.
doi: 10.1007/s11577-014-0253-x
Köller, O., Trautwein, U., Lüdtke, O., & Baumert, J. (2006). Zum Zusammenspiel von
schulischer Leistung, Selbstkonzept und Interesse in der gymnasialen Oberstufe [On the
0652.20.1.27
learners and students with disabilities. Review of Research in Education, 39, 165–214.
doi: 10.3102/0091732X14556073
LeRoy, BW, Samuel, P., Deluca, M., & Evans, P. (2018). Students with special educational
needs within PISA. Assessment in Education: Principles, Policy & Practice. doi:
10.1080/0969594X.2017.1421523
10.1348/000709906X156881
Lüdtke, O., Köller, O., Marsh, HW, & Trautwein, U. (2005). Teacher frame of reference and
10.1016/j.cedpsych.2004.10.002
Lüdtke, O., Marsh, HW, Robitzsch, A., Trautwein, U., Asparouhov, T., & Muthén, B.
(2008). The multilevel latent covariate model: a new, more reliable approach to group
10.1037/a0012869
Margalit, M., & Efrati, M. (1996). Loneliness, coherence and companionship among children
10.1080/0144341960160106
10.3102/00028312023001129
glory effects in Hong Kong schools. Journal of Personality and Social Psychology, 78,
Marsh, HW, Lüdtke, O., Robitzsch, A., Trautwein, U., Asparouhov, T., Muthén, B., &
10.1080/00273170903333665
relatively large fish in a small pond even if you don't learn to swim as well? Journal of
Marsh, HW, Seaton, M., Trautwein, U., Lüdtke, O., Hau, KT, O'Mara, AJ, & Craven, RG
theory, methodology, and future research. Educational Psychology Review, 20, 319–
Marsh, HW, Tracey, DK, & Craven, RG (2006). Multidimensional self-concept structure
Marsh, HW, Trautwein, U., Lüdtke, O., Köller, O., & Baumert, J. (2005). Academic self
concept, interest, grades, and standardized test scores: Reciprocal effects models of
8624.2005.00853.x
Marsh, HW, & Yeung, AS (1998). Longitudinal structural equation models of academic
10.3102/00028312035004705
Martin, MO, Mullis, IVS, & Foy, P. (Eds.). (2008). TIMSS 2007 international
Science Study at the fourth and eighth grades. Chestnut Hill, MA: TIMSS & PIRLS
Martin, MO, & Preuschoff, C. (2008). Creating the TIMSS 2007 background indices. In JF
Olson, MO Martin, & IVS Mullis (Eds.), TIMSS 2007 Technical Report (pp. 281–338).
Chestnut Hill, MA: TIMSS & PIRLS International Study Center, Boston College.
McLeskey, J., & Waldron, NL (2002). Inclusion and school change: Teacher perceptions
Möller, J., Streblow, L., & Pohlmann, B. (2009). Achievement and self-concept of students
10.1007/s11218-008-9065-z
Mullis, IVS, Martin, MO, Foy, P., & Arora, A. (Eds.). (2012). TIMSS 2011 international
results in mathematics. Chestnut Hill, MA: TIMSS & PIRLS International Study Center,
Boston College.
Muthén, LK, & Muthén, BO (1998–2015). Mplus User's Guide. Seventh Edition. Los
Nagengast, B., Marsh, HW, Scalas, LF, Xu, MK, Hau, K.-T., & Trautwein, U. (2011). Who
Nusser, L., Carstensen, CH, & Artelt, C. (2015). Befragung von Schülerinnen und Schülern
Nusser, L., Heydrich, J., Carstensen, CH, Artelt, C., & Weinert, S. (2016). Validity of
survey data of students with special educational needs— results from the National
Educational Panel Study. In H.-P. Blossfeld, J. von Maurice, M. Bayer, & J. Skopek
Educational Panel Study (pp. 251–266). Wiesbaden: Springer VS. doi: 10.1007/978-3-
658-11994-2_15
Nusser, L., & Wolter, I. (2016). There's plenty more fish in the sea: Das akademische
Lernen in integrativen und segregierten Schulsettings [There's plenty more fish in the
sea. Academic self-concept of students with special educational needs in the area of
learning in inclusive and segregated school settings]. Empirische Pädagogik, 30, 130–
143.
Patrick, H., Ryan, AM, & Kaplan, A. (2007). Early adolescents' perceptions of the
Peetsma, T., Vergeer, M., Roeleveld, J., & Karsten, S. (2001). Inclusion in education:
Pekrun, R. (1992). The impact of emotions on learning and achievement: Towards a theory of
10.1111/j.1464-0597.1992.tb00712.x
Pekrun, R. (2000). A social-cognitive, control-value theory of achievement emotions. In J.
Pekrun, R., Goetz, T., Frenzel, AC, Barchfeld, P., & Perry, RP (2011). Measuring
/10.1016/j.cedpsych.2010.10.002
Pekrun, R., Goetz, T., Titz, W., & Perry, RP (2002). Academic emotions in students' self
Pfahl, L., & Powell, JJW (2011). Legitimating school segregation. The special education
profession and the discourse of learning disability in Germany. Disability & Society, 26,
Pijl, SJ, & Frostad, P. (2010). Peer acceptance and self‐concept of students with disabilities
in regular education. European Journal of Special Needs Education, 25, 93–105. doi:
10.1080/08856250903450947
divergence in the United States and Germany. Comparative Education Review, 53, 161–
R Core Team. (2015). R: A language and environment for statistical computing. Vienna,
project.org/
Rauer, W., & Schuck, KD (2003). Fragebogen zur Erfassung emotionaler und sozialer
[Questionnaire assessing emotional and social experiences of third and fourth graders].
Göttingen: Beltz Test.
Renick, MJ, & Harter, S. (1989). Impact of social comparisons on the developing self
Reyes, MR, Brackett, MA, Rivers, SE, White, M., & Salovey, P. (2012). Classroom
Richter, D., Böhme, K., Bastian-Wurzel, J., Pant, HA, & Stanat, P. (2014). IQB
Qualitätsentwicklung im Bildungswesen.
Roppelt, A., & Reiss, K. (2012). Die im Ländervergleich 2011 in den Fächern Deutsch und
P. Stanat, HA Pant, K. Böhme, & D. Richter (Eds.), Kompetenzen von Schülerinnen und
Schülern am Ende der vierten Jahrgangsstufe in den Fächern Deutsch und Mathematik:
Rose, CA, Monda-Amaya, LE, & Espelage, DL (2011). Bullying perpetration and
Rosenbaum, PR, & Rubin, DB (1983). The central role of the propensity score in
10.1093/biomet/70.1.41
Rosenbaum, PR, & Rubin, DB (1985). Constructing a control group using multivariate
matched sampling methods that incorporate the propensity score. The American
Roy, A., Guay, F., & Valois, P. (2015). The big-fish–little-pond effect on academic self
Ruijs, NM, & Peetsma, TTD (2009). Effects of inclusion on students with and without
10.1016/j.edurev.2009.02.002
Ruijs, N., Peetsma, T., & van der Veen, I. (2010). The presence of several students with
special educational needs in inclusive education and the functioning of students with
10.1080/00131910903469551
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 55
Russell, SL (2012). Individual- and classroom-level social support and classroom behavior
http://drum.lib.umd.edu/handle/1903/13204
Salend, SJ, & Duhaney, LMG (1999). The impact of inclusion on students with and without
disabilities and their educators. Remedial and Special Education, 20, 114–126.
doi:10.1177/074193259902000209
Sälzer, C., Gebhardt, M., Müller, K., & Pauly, E. (2015). Der Prozess der Feststellung
Pant, & M. Prenzel (Eds.), Inklusion von Schülerinnen und Schülern mit
Sansour, T., & Bernhard, D. (2018). Special needs education and inclusion in Germany and
Savage, R. (2005). Friendship and bullying patterns in children attending a language base in a
Schurtz, IM, Pfost, M., Nagengast, B., & Artelt, C. (2014). Impact of social and
10.1016/j.learninstruc.2014.08.001
Schwab, S. (2015). Social dimensions of inclusion in education of 4th and 7th grade pupils in
Scruggs, TE, Mastropieri, MA, & Marshak, L. (2012). Peer-mediated instruction in inclusive
secondary social studies learning: direct and indirect learning effects. Learning
Shavelson, RJ, Hubner, JJ, & Stanton, GC (1976). Self-concept: Validation of construct
10.3102/00346543046003407
Skinner, EA, & Belmont, MJ (1993). Motivation in the classroom: Reciprocal effects of
teacher behavior and student engagement across the school year. Journal of Educational
10.1006/ceps.1996.0004
Song, J., Bong, M., Lee, K., & Kim, S. (2015). Longitudinal investigation into the role of
Spörer, N., Maaz, K., Vock, M., Schründer-Lenzen, A., Luka, T., Bosse, S., … Jäntsch, C.
(2015). Lernen in der inklusiven Grundschule: Zusammenhänge zwischen fachlichen
inclusive primary schools: Relations between academic performance, social climate, and
Stanat, P., Pant, HA, Böhme, K., & Richter, D. (Eds.). (2012). Kompetenzen von
Schülerinnen und Schülern am Ende der vierten Jahrgangsstufe in den Fächern Deutsch
German and mathematics at the end of the 4th grade: Results of the IQB National
[dataset] Stanat, P., Pant, HA, Böhme, K., Richter, D., Weirich, S., Haag, N., Roppelt, A.,
http://doi.org/10.5159/IQB_LV_2011_v2
Stuart, EA (2010). Matching methods for causal inference: A review and a look forward.
Stuart, EA, & Green, KM (2008). Using full matching to estimate causal effects in
1649.44.2.395
Televantou, I., Marsh, HW, Kyriakides, L., Nagengast, B., Fletcher, J., & Malmberg, L.-E.
Trautwein, U., Lüdtke, O., Marsh, HW, Köller, O., & Baumert, J. (2006). Tracking, grading,
and student motivation: Using group composition and status to predict self concept and
doi: 10.1037/0022-0663.98.4.788
Trautwein, U., Lüdtke, O., Marsh, HW, & Nagy, G. (2009). Within-school social
comparison: How students perceive the standing of their class predicts academic self
United Nations General Assembly (UN). Convention on the Rights of Persons with
http://www.refworld.org/docid/4680cd212.html
Urdan, T., & Schoenfelder, E. (2006). Classroom effects on student motivation: Goal
Van de Schoot, R., Lugtig, P., & Hox, J. (2012). A checklist for testing measurement
10.1080/17405629.2012.686740
Vaughn, S., Elbaum, BE, & Schumm, JS (1996). The effects of inclusion on the social
598–608.
Vaughn, S., Haager, D., Hogan, A., & Kouzekanani, K. (1992). Self-concept and peer
Wang, M.-T., & Eccles, JS (2013). School context, achievement motivation, and academic
Weirich, S., Haag, N., & Roppelt, A. (2012). Testdesign und Auswertung des
Richter (Eds.), Kompetenzen von Schülerinnen und Schülern am Ende der vierten
Wentzel, KR (1998). Social relationships and motivation in middle school: The role of
parents, teachers, and peers. Journal of Educational Psychology, 90, 202–209. doi:
10.1037/0022-0663.90.2.202
Wentzel, KR, Battle, A., Russell, SL, & Looney, LB (2010). Social supports from
Wheeler, L., & Suls, J. (2005). Social comparison and self-evaluations of competence. In AJ
Elliot & CS Dweck (Eds.), Handbook of competence and motivation (pp. 566–578).
Wiener, J., & Tardif, CY (2004). Social and emotional functioning of children with
M., & Schleyer, EJ (1999). The big-fish–little-pond effect for academic self concept, test
anxiety, and school grades in gifted children. Contemporary Educational Psychology, 24,
Zigmond, N. (2003). Where should students with disabilities receive special education
services? Is one place better than another? The Journal of Special Education, 37, 193–
Table 1
Sample Sizes and Descriptive Statistics by Educational Placement
Educational placement
Special education schools
Student level
Regular schools
N SEN students overall 420 678 n SEN students with learning disabilities 261 289 n SEN students with speech or language
impairment 140171 n SEN students with emotional disorders 19 218
Average number of SEN students per classroom (SD) 9.84 (4.05) 1.90 (1.36) Range number of SEN
students per classroom 2 – 26 1 – 9 N Students overall 420 8272 Classroom level
N Classroomsa49 402 Average class size (SD) 9.84 (4.05) 20.58 (4.25) Range class size 2 – 26 6 – 33 Note.
SEN = special educational needs. aone classroom per school was chosen randomly to participate in the
study.
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 61
Table 2
Item Wordings and Descriptive Statistics for Academic Self-Concept, Enjoyment of Learning, and
Perceived Social Support
Educational placement
Special education schools
Regular schools
Item wordingsa M SD M SD Academic self-concept in German
I am satisfied with my achievement in German 3.14 1.05 2.76 1.01 I understand most of the things we
have to learn in German 3.13 1.00 2.92 0.92 I am usually good at solving exercises in German lessons
3.13 1.03 2.89 0.94 Academic self-concept in mathematics
I am satisfied with my achievement in mathematics 3.18 1.09 2.86 1.09 I understand most of the things
we have to learn in
mathematics 3.16 1.03 2.95 1.00 I am usually good at solving exercises in mathematics lessons 3.24
1.01 2.95 1.00 Enjoyment of learning
I enjoy acquiring new knowledge in school lessons 3.26 1.04 3.05 0.97 I enjoy some school topics so
much that I am looking forward
to them in advance 3.29 1.04 3.13 1.00 I am often excited about school lessons 3.02 1.11 2.67 1.03
Perceived social support
My classmates are nice to me 3.26 0.99 3.06 0.96 My classmates cheer me up when I am sad 2.98 1.16
2.81 1.07 I have few friends in my class [inverted item] 2.02 1.24 2.10 1.15 Others start arguments with
me [inverted item] 1.87 1.15 1.77 1.02
Note. Responses were given by students on a 4-point scale (1 = do not agree; 4 = strongly agree).
a
Items translated from German by the authors.
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 62
Table 3
Correlations Between Educational Placement, Achievement, and Social Support
123
Student level
1 Educational placementa
(1 = special education school)
2 Reading achievement (WLE)–.46***
3 Mathematics achievement (WLE)–.51*** .66***
4 Social support .08 .05 .04
b
Classroom level
1 Educational placementa
(1 = special education school)
2 Reading achievement (WLE)–.83***
3 Mathematics achievement (WLE)–.81*** .91***
4 Social support–.31*** .32*** .34***
Note. WLE = weighted likelihood ability estimates (Warm, 1989).
a
Point-biserial correlations are reported for educational placement. b Correlations
on the classroom level are based on unstandardized aggregations of individual
characteristics.
*p < .05. **p < .01. ***p < .001
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 63
Table 4
Predicting SEN Students' Academic Self-concept in German
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
B SE B SE B SE B SE B SE B SE
Student level (L1)
Reading achievement 0.11*** 0.03 0.16*** 0.03 0.15*** 0.04 0.15*** 0.03 Perceived social support 0.15***
0.03 0.14*** 0.03 0.13*** 0.03 Gender (1 = female) 0.12* 0.05 0.11* 0.05 0.11* 0.05 0.09 0.05 0.09 0.05 0.09
0.05 Classroom level (L2)
School type (1 = special education
school) 0.31*** 0.07 0.09 0.08 0.21*** 0.06 0.07 0.08 Class-average reading achievement –0.20 *** 0.05 –
0.16** 0.06 –0.16** 0.06 Class-average social support –0.06 0.03 –0.02 0.03 –0.01 0.03 Model fit
Chi-Square test of model fit (df) 35.94*** (9) 38.45*** (9) 39.09*** (11) 39.28*** (9) 41.97*** (11) 49.48***
(17) Bayesian Information Criteria (BIC) 9668.56 10604.50 9665.25 9016.24 9010.61 12022.62 Akaike
Information Criteria (AIC) 9578.72 10504.47 9570.41 8930.03 8919.62 11908.03 Comparative Fit Index (CFI)
0.95 0.95 0.95 0.95 0.95 0.95 Root Mean Square Error (RMSEA) 0.05 0.06 0.05 0.06 0.06 0.04 Described
Variance
R² on student level (L1) 4.40% 7.90% 7.50% 7.20% 6.10% 11.50% R² on classroom level (L2) 26.80% 46.80%
46.60% 6.20% 21.10% 46.60% Note. SEN = special educational needs. Intraclass correlation (ICC) = .09.
*p < .05. **p < .01. ***p < .001.
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 64
Table 5
Predicting SEN Students' Academic Self-concept in Mathematics
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
B SE B SE B SE B SE B SE B SE
Student level (L1)
Mathematics achievement 0.30*** 0.04 0.41*** 0.04 0.41*** 0.05 0.40*** 0.04 Perceived social support
0.13*** 0.04 0.12** 0.04 0.10** 0.04 Gender (1 = female) –0.13* 0.06 –0.12 0.06 –0.11 0.06 –0.26*** 0.06 –
0.26*** 0.06 –0.15* 0.06 Classroom level (L2)
School type (1 = special education
school) 0.48*** 0.08 0.07 0.08 0.27*** 0.07 0.06 0.08 Class-average mathematics achievement –0.35 *** 0.05 –
0.32*** 0.06 –0.34*** 0.06 Class-average social support 0.01 0.04 0.06 0.04 0.07 0.04 Model fit
Chi-Square test of model fit (df) 46.79*** (10) 46.34*** (10) 47.09*** (12) 24.27** (10) 28.47** (12) 62.45***
(18) Bayesian Information Criteria (BIC) 9595.29 10510.38 9568.8 9056.32 9050.50 11941.62 Akaike
Information Criteria (AIC) 9510.61 10415.39 9484.13 8975.00 8964.40 11832.16 Comparative Fit Index (CFI)
0.95 0.95 0.96 0.98 0.98 0.95 Root Mean Square Error (RMSEA) 0.06 0.06 0.05 0.04 0.04 0.05 Described
Variance
R² on student level (L1) 17.10% 28.10% 27.60% 6.20% 5.50% 29.10% R² on classroom level (L2) 24.60%
55.40% 55.40% 0.20% 12.10% 58.30% Note. SEN = special educational needs. Intraclass correlation (ICC) = .
14.
*p < .05. **p < .01. ***p < .001.
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 65
Table 6
Predicting SEN Students' Enjoyment of Learning With Reading Achievement
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
B SE B SE B SE B SE B SE B SE
Student level (L1)
Reading achievement 0.09* 0.04 0.13** 0.04 0.13** 0.04 0.12** 0.04 Perceived social support 0.23*** 0.03
0.21*** 0.03 0.20*** 0.03 Gender (1 = female) 0.20** 0.07 0.19** 0.07 0.19** 0.07 0.14* 0.07 0.14* 0.07
0.14* 0.07 Classroom level (L2)
School type (1 = special education
school) 0.30*** 0.08 0.09 0.11 0.24** 0.07 0.08 0.10 Class-average reading achievement –0.18 *** 0.05 –0.14*
0.07 –0.16* 0.07 Class-average social support 0.02 0.04 0.06 0.04 0.08 0.04 Model fit
Chi-Square test of model fit (df) 58.40*** (9) 55.13*** (9) 54.68*** (11) 29.47*** (9) 33.83*** (11) 64.55***
(17) Bayesian Information Criteria (BIC) 9694.95 10635.45 9696.43 8997.11 8992.37 12015.39 Akaike
Information Criteria (AIC) 9605.13 10535.42 9601.62 8911.07 8901.55 11900.80 Comparative Fit Index (CFI)
0.92 0.92 0.93 0.97 0.97 0.93 Root Mean Square Error (RMSEA) 0.07 0.07 0.06 0.05 0.05 0.05 Described
Variance
R² on student level (L1) 3.90% 5.70% 5.50% 12.30% 10.90% 13.40% R² on classroom level (L2) 12.30% 20.50%
20.00% 0.50% 12.70% 24.70% Note. SEN = special educational needs. Intraclass correlation (ICC) = .15.
*p < .05. **p < .01. ***p < .001.
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 66
Appendix A
Testing for Measurement Invariance and Predicting Enjoyment of Learning With Mathematics
Achievement
In supplementary analyses, we first checked the fit of the measurement model for each motivational
outcome separately and second, we tested for measurement invariance between students with special
educational needs (SEN) in special education schools and regular schools. Comparable measurement
models and measurement invariance are necessary requirements for valid comparisons of the latent
factor means across both student groups. First, for SEN students in each school setting (special
education school vs. regular school), a two-level measurement model with equal factor loadings on the
student level (L1) and on the classroom level (L2) was specified (see Model 0 in Tables A1 - A3).
Second, we ran different models constraining (1) all factor loadings to be equal for SEN students in
special education schools and regular schools (metric invariance, see Model 1 in Tables A1 - A3), (2)
all intercepts to be equal across both groups (intercept-only invariance, see Model 2 in Tables A1 - A3),
(3) all intercepts and all factor loadings to be equal across both groups (scalar invariance, see Model 3
in Tables A1 - A3), and (4) allow one intercept to vary between groups while constraining the other
intercepts and all factor loadings to be equal (partial invariance, see Model 4 in Tables A1 - A3). The
measurement models showed a reasonable fit for SEN students in special education schools and regular
schools (see Model 0 in Tables A1 - A3). Furthermore, we established partial invariance for each
motivational outcome allowing valid inferences about differences in the latent means between groups.
Table A4 shows that analyses with mathematical achievement predicting students' enjoyment of
Table A1
Testing Measurement Invariance for Academic Self-Concept German
χ² (df) p CFI TLI RMSEA BIC AIC Δχ² (df) p (Δχ²)
Model 0a: Configural invariance (regular school) 0.90 (2) .64 1.00 1.00 .00 4084.38 4028.54 - - Model
0a: Configural invariance (special education school) 9.23 (2) .01 .96 .89 .11 2573.14 2524.19 - - Model
1: Metric invariance (equal factor loadings) 3.49 (3) .32 1.00 1.00 .02 6610.31 6538.94 - - Model 2:
Intercept-only invariance (equal intercepts) 26.36 (3) .00 .95 .90 .14 6633.18 6561.81 22.87 (0) .00
Model 3: Scalar invariance (equal factor loadings & intercepts) 30.71 (6) .00 .95 .95 .10 6617.26
6560.16 4.35 (3) .23 Model 4b: Partial invariance 30.61 (8) .00 .95 .96 .08 6603.63 6556.05 4.24 (5) .52
Note. aModel 0 refers to a two-level measurement model with equal factor loadings on student level
(L1) and classroom level (L2). bModel 4 allows differences in the intercept of item 1 between special
education and regular schools and restricts residual variances to be equal across both groups. The fit
of Model 4 does not differ significantly (p = .52) from Model 2.
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 68
Table A2
Testing Measurement Invariance for Academic Self-Concept Mathematics
χ² (df) p CFI TLI RMSEA BIC AIC Δχ² (df) p (Δχ²)
Model 0a: Configural invariance (regular school) 2.24 (2) .33 1.00 1.00 .02 4225.63 4169.75 - - Model
0a: Configural invariance (special education school) 1.88 (2) .39 1.00 1.00 .00 2497.57 2448.58 - -
Model 1: Metric invariance (equal factor loadings) 0.48 (3) .92 1.00 1.01 .00 6686.29 6614.87 - - Model
2: Intercept-only invariance (equal intercepts) 19.75 (3) .00 0.98 0.96 .11 6705.56 6634.14 19.27 (0) .00
Model 3: Scalar invariance (equal factor loadings & intercepts) 20.38 (6) .00 0.98 0.98 .07 6685.91
6628.77 0.63 (3) .89 Model 4b: Partial invariance 15.91 (5) .01 0.99 0.98 .07 6688.20 6626.30 3.84 (2) .
15 Note. aModel 0 refers to a two-level measurement model with equal factor loadings on student level
(L1) and classroom level (L2). bModel 4 allows differences in the intercept of item 1 between special
education and regular schools. The fit of Model 4 does not differ significantly (p = .15) from Model 2.
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 69
Table A3
Testing Measurement Invariance for Enjoyment of Learning
χ² (df) p CFI TLI RMSEA BIC AIC Δχ² (df) p (Δχ²)
Model 0a: Configural invariance (regular school) 3.81 (2) .15 .99 .98 .04 4168.72 4112.91 - - Model 0a:
Configural invariance (special education school) 0.70 (2) .71 1.00 1.01 .00 2526.17 2477.15 - - Model
1: Metric invariance (equal factor loadings) 4.02 (3) .26 1.00 1.00 .03 6661.89 6590.50 - - Model 2:
Intercept-only invariance (equal intercepts) 19.43 (3) .00 .97 .94 .11 6677.30 6605.91 15.41 (0) .00
Model 3: Scalar invariance (equal factor loadings & intercepts) 24.24 (6) .00 .97 .97 .08 6661.83
6604.72 4.81 (3) .19 Model 4b: Partial invariance 13.97 (5) .02 .98 .98 .07 6658.32 6596.45 5.46 (2) .07
Note. aModel 0 refers to a two-level measurement model with equal factor loadings on student level
(L1) and classroom level (L2). bModel 4 allows differences in the intercept of item 3 between special
education and regular schools. The fit of Model 4 does not differ significantly (p = .07) from Model 2.
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 70
Table A4
Predicting SEN Students' Enjoyment of Learning With Mathematics Achievement
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
B SE B SE B SE B SE B SE B SE
Student level (L1)
Mathematics achievement 0.09* 0.04 0.14** 0.05 0.13* 0.05 0.12** 0.05 Perceived social support 0.23*** 0.03
0.21*** 0.03 0.21*** 0.03 Gender (1 = female) 0.22** 0.07 0.22** 0.07 0.22** 0.07 0.14* 0.07 0.14* 0.07
0.17* 0.07 Classroom level (L2)
School type (1 = special education
school) 0.30*** 0.08 0.13 0.10 0.24** 0.07 0.09 0.09 Class-average mathematics achievement –0.19 *** 0.05 –
0.13* 0.06 –0.16** 0.06 Class-average social support 0.02 0.04 0.06 0.04 0.08 0.04 Model fit
Chi-Square test of model fit (df) 88.66*** (9) 82.45*** (9) 84.01*** (11) 29.47*** (9) 33.83*** (11) 90.57***
(17) Bayesian Information Criteria (BIC) 9645.42 10582.22 9629.75 8997.11 8992.37 11953.59 Akaike
Information Criteria (AIC) 9555.74 10482.24 9535.11 8911.07 8901.55 11839.13 Comparative Fit Index (CFI)
0.87 0.88 0.89 0.97 0.97 0.90 Root Mean Square Error (RMSEA) 0.09 0.09 0.08 0.05 0.05 0.06 Described
Variance
R² on student level (L1) 4.30% 6.80% 6.30% 12.30% 10.90% 14.10% R² on classroom level (L2) 12.20% 21.10%
20.40% 0.50% 12.70% 27.00% Note. SEN = special educational needs. Intraclass correlation (ICC) = .15.
*p < .05. **p < .01. ***p < .001.
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 71
Appendix B
Testing for Differential Effects of Type of Special Educational Needs on Students' Achievement
Motivation
In supplementary analyses, we tested if the relation between educational placement and motivational
outcomes differs between students with learning disabilities and students with speech or language
impairment. The small amount of students with emotional disorders in special education schools
learning for this group. To test for differential effects of classroom composition on the relationship
between type of SEN and academic self-concept and enjoyment of learning, respectively, random slope
models were run in a first step. In the case of classrooms differing in the magnitude of the relation
between type of SEN and both motivational outcomes (ie, significant slope variation), we planned to
composition (Aguinis, Gottfredson, & Culpepper, 2013). After accounting for differences in individual
achievement, no significant main effects of type of SEN were found (see Model 1 and Model 3 in Table
B1 and Table B2). Random slope models were then estimated to test if the relation between type of
SEN and motivational outcomes varied between classrooms. The slope variance was not statistically
different from zero for both academic self-concept and enjoyment of learning (see Model 2 and Model 4
in Table B1 and Table B2). Therefore, no additional cross-level interaction effects between class-
Table B1
Testing for Differential Effects of Type of SEN on Academic Self-Concept
Academic self-concept German Academic self-concept mathematics
Model 1 Model 2 Model 3 Model 4
B SE B SE B SE B SE
Student level (L1)
Reading achievement 0.16*** 0.04 0.16*** 0.04
Mathematics achievement 0.39*** 0.05 0.39*** 0.05 Type of SEN1: Learning disabilities (1= yes) –
0.01 0.07 –0.01 0.07 –0.05 0.09 –0.06 0.09 Type of SEN1: Speech or language impairment (1= yes) –
0.09 0.07 –0.09 0.07 0.11 0.08 0.12 0.09 Gender (1 = female) 0.12* 0.05 0.12* 0.05 –0.12 0.06 –0.12*
0.06 Classroom level (L2)
School type (1 = special education school) 0.13 0.09 0.13 0.09 0.01 0.09 –0.02 0.09 Class-average
reading achievement –0.14* 0.06 –0.14* 0.07
Class-average mathematics achievement –0.35*** 0.07 –0.36*** 0.07 Variance Components
Slope (L2) variance: Learning disabilities 0.04 0.27 0.12 0.22 Slope (L2) variance: Speech or language
impairment 0.02 0.18 0.02 0.17 Intercept-slope (L2) covariance: Learning disabilities –0.01 0.13 –0.001
0.12 Intercept-slope (L2) covariance: Speech or language
impairment –0.01 0.10 –0.02 0.10 Model fit
Bayesian Information Criteria (BIC) 9677.06 9702.82 9582.05 9595.96 Akaike Information Criteria
(AIC) 9572.25 9578.04 9482.45 9476.44 Note. SEN = special educational needs. Intraclass correlation
(ICC) = .09 for self-concept German and ICC = .14 for self-concept mathematics.
*p < .05. **p < .01. ***p < .001.
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 73
Table B2
Testing for Differential Effects of Type of SEN on Enjoyment of Learning
Enjoyment of learning Enjoyment of learning
(Reading achievement) (Mathematics achievement)
Model 1 Model 2 Model 3 Model 4
B SE B SE B SE B SE
Student level (L1)
Reading achievement 0.14*** 0.04 0.13** 0.04
Mathematics achievement 0.15** 0.05 0.15** 0.05 Type of SEN1: Learning disabilities (1= yes) 0.10
0.08 0.10 0.08 0.13 0.08 0.13 0.09 Type of SEN1: Speech or language impairment (1= yes) 0.11 0.08
0.10 0.08 0.11 0.08 0.10 0.09 Gender (1 = female) 0.18** 0.06 0.18** 0.07 0.21** 0.07 0.21** 0.07
Classroom level (L2)
School type (1 = special education school) 0.06 0.12 0.07 0.11 0.10 0.11 0.12 0.12 Class-average
reading achievement –0.15* 0.07 –0.16* 0.07
Class-average mathematics achievement –0.14* 0.07 –0.13* 0.07 Variance Components
Slope (L2) variance: Learning disabilities 0.05 0.19 0.04 0.22 Slope (L2) variance: Speech or language
impairment 0.10 0.21 0.11 0.21 Intercept-slope (L2) covariance: Learning disabilities –0.002 0.10 –
0.003 0.12 Intercept-slope (L2) covariance: Speech or language impairment –0.08 0.12 –0.09 0.12
Model fit
Bayesian Information Criteria (BIC) 9708.44 9728.01 9641.11 9661.29 Akaike Information Criteria
(AIC) 9603.64 9603.25 9536.51 9536.76 Note. SEN = special educational needs. Intraclass correlation
(ICC) = .15.
*p < .05. **p < .01. ***p < .001.
EDUCATIONAL PLACEMENT AND ACHIEVEMENT MOTIVATION 74
Appendix C
Propensity Score Matching Analyses to Control for Differences in Background Variables Between
In the present study, we compared the academic self-concept and enjoyment of learning of students with
SEN in special education schools and regular schools. However, students with SEN are not randomly
assigned to different school types. They are selected to special education schools or regular schools
based on their prior achievement and other background variables (Peetsma, Vergeer, Roeleveld, &
Karsten, 2001; Zigmond, 2003) as well as based on variables on the federal state level as the German
federal states differ in their implementation of inclusive schooling. Therefore, pre-existing group
differences might influence the association between school type and SEN students' achievement
motivation (selection bias; Duncan, Magnuson, & Ludwig, 2004). In supplementary analyses, we
applied propensity score matching (PSM; Rosenbaum & Rubin, 1983, 1985) to control for differences
between SEN students in special education schools and regular schools in confounding background
variables. PSM combines relevant background variables into one single measure that indicates the
probability of receiving a treatment (in the present study: attending a regular school). Then, SEN
students in special education schools and regular schools are matched based on their propensity score.
Subsequently, only students with similar propensity scores are compared in their achievement
motivation. Differences in SEN students' achievement motivation could therefore be attributed more
We applied a full matching procedure (Hansen, 2004; Stuart, 2010), in which the sample was divided
into subgroups consisting of at least one student from a regular school (treatment group) and at least one
subgroups was estimated in a way that minimizes the differences in the propensity scores in each
subgroup (Stuart, 2010). SEN students from the treatment group (regular schools) received a weight of
1 whereas SEN students from the control group (special education schools) received a weight
proportional to the number of SEN students from the treatment group (Stuart & Green, 2008). The
selection of relevant background variables for our matching model (see Figure C1) was based on a
previous study (Kocaj et al., 2014). Our matching model includes 16 variables: students' verbal and
numeric cognitive abilities (derived from an established test inventory for grade 4 to 12, KFT 4-12+ R;
Heller & Perleth, 2000), various indicators of students' socio-economic and cultural background
(number of books at home; parents' socio-economic status as operationalized by the international socio-
economic index of occupational status, Ganzeboom, De Graaf, & Treiman, 1992; parents' educational
parents' educational aspirations (parents' desired and anticipated degree for their child), parents'
educational goals (see Baumert et al., 2008), parents born in Germany, language at home, students'
gender, age, school entrance age, years spent in kindergarten, diagnosed learning disability, and the
German federal state where the student attends school (not depicted in Figure C1).
Full matching was implemented in R (R Development Core Team, 2015) using the package MatchIt
(Ho, Imai, King, & Stuart, 2011). After the full matching, both groups of SEN students were more
similar with regard to the propensity score and the background variables. The full matching
substantially reduced group differences in the propensity score and in the background variables (Figure
C1). Furthermore, the distributions of the propensity scores were more similar for both groups and the
matching resulted in a large reduction in standardized bias (Figure C2 & Table C1). Subsequently, we
replicated the central analyses of our study by running multilevel regression analyses in Mplus 7.3
with the matched sample using the weights from the full matching procedure. The results were similar
to those from the unmatched sample analyses (Table C2 - C5). However, the negative effect of class-
average achievement on SEN students' enjoyment of learning was not significant after the matching
(Model 3 & Model 6, Table C4: b = –0.15, p = .07; Model 3, Table C5: b = –0.17, p = .06). In the
analyses without matching, the effect was quite similar in magnitude but significant (Model 3, Table 6:
b = –0.14, p = .03; Model 6, Table 6: b = –0.16, p = .02; Table A4, Model 3: b = –0.13, p = .04). One
explanation for this diverging result pattern might be an increase in the standard error because of the
Table C1
Distribution of Propensity Scores and Standardized Bias Before and After Full Matching