Indoor air problems and the perceived social climate in schools : A multilevel structural equation analysis

1 1 2 3 4 This is he pos p in e sion o he a icle, which has been published in Science o he To al En i onmen 2018, 624, 1504–1512. The inal publica ion is a ailable a h ps://doi.o g/10.1016/j.sci o en .2017.12.126 5 6 7 Running head: IAQ and social clima e in schools 8 9 10 Indoo ai p oblems and he pe cei ed social clima e in schools: a mul ile el s uc u al equa ion 11 analysis 12 13 Ee ika Finell* 14 Asko Tol anen 15 Ulla Ha e inen-Shaughnessy 16 Seppo Laaksonen 17 Saka i Ka onen 18 Reijo Sund 19 Pauliina Luopa 20 Juha Pekkanen 21 Timo S åhl 22 23 *Reques s o ep in s should be add essed o Ee ika Finell, Facul y o Social Sciences, Linna, 33014 24 Uni e si y o Tampe e, Tampe e, Finland. E-mail: [email p o ec ed] 25 26 27 2 O he a ilia ions: 1 2 Asko Tol anen: Me hodology Cen e o Human Sciences, P.O. Box 35, 40014 Uni e si y o 3 Jy äskylä, Finland 4 Ulla Ha e inen-Shaughnessy: Depa men o Heal h P o ec ion, Na ional Ins i u e o Heal h and 5 Wel a e, P.O. Box 95, 70701 Kuopio, Finland 6 Seppo Laaksonen: Depa men o Social Resea ch, P.O. Box 68, 00014 Uni e si y o Helsinki, 7 Finland 8 Saka i Ka onen: Depa men o Heal h and Social Ca e Sys ems, Na ional Ins i u e o Heal h and 9 Wel a e, P.O. Box 30, 00271 Helsinki, Finland 10 Reijo Sund: 1) Cen e o Resea ch Me hods, Depa men o Social Resea ch, P.O. Box 18, 00014 11 Uni e si y o Helsinki, Finland, 2) Ins i u e o Clinical Medicine, Uni e si y o Eas e n Finland, 12 Kuopio, Finland 13 Pauliina Luopa: Depa men o Wel a e, Na ional Ins i u e o Heal h and Wel a e, P.O. Box 30, 14 00271 Helsinki, Finland 15 Juha Pekkanen: 1) Depa men o Public Heal h, P.O. Box 20, 00014 Uni e si y o Helsinki, Finland, 16 2) Depa men o Heal h P o ec ion, Na ional Ins i u e o Heal h and Wel a e, Finland 17 Timo S åhl: Depa men o Wel a e, Na ional Ins i u e o Heal h and Wel a e, Bioka u 10, 18 33520 Tampe e, Finland 19 20 21 3 Abs ac 1 Indoo ai p oblems in schools appea o in luence lea ning ou comes and absence a es. 2 Howe e , p e ious esea ch has no in es iga ed whe he indoo ai p oblems in luence he social 3 clima e o schools. The e o e, we s udied whe he indoo ai p oblems obse ed in schools 4 associa e wi h s uden s’ pe cep ions o he eache -s uden ela ionship and class spi i . The 5 na ionwide sample o Finnish schools (N = 194 schools/27 153 s uden s) was analyzed using 6 mul ile el s uc u al equa ion modeling. Da a on he schools’ social clima e collec ed om 7 s uden s we e me ged wi h independen ly collec ed da a on obse ed indoo ai p oblems om 8 school p incipals. We ound ha he eache -s uden ela ionship was epo ed o be wo se in 9 schools wi h obse ed indoo ai p oblems compa ed o hose wi hou obse ed indoo ai 10 p oblems. Fu he mo e, he epo ed class spi i was wo se in schools wi h obse ed indoo ai 11 p oblems, bu only among s uden s wi h a high g ade poin a e age. Ou indings indica e ha 12 indoo ai p oblems may a ec he s uden -pe cei ed social clima e. 13 Keywo ds (6): indoo ai quali y, social clima e, eache -s uden ela ionship, psychosocial e ec , 14 lowe seconda y school, class spi i 15 16 17 18 19 20 21 22 23 4 1. In oduc ion 1 P oblems in schools’ indoo ai quali y (IAQ) a e widesp ead. Fo example, some 24% o schools in 2 Finland, 20% in he Ne he lands, and 40% Spain we e es ima ed o ha e mois u e p oblems, such 3 as dampness o mold. 1 Fu he mo e, many s udies ha e ound ha a la ge p opo ion o schools 4 a e inadequa ely en ila ed. 2–4 Such p oblems can inc ease he isk o s uden s’ heal h, 5,6 5 absence a es, 7,8 and dec eased academic pe o mance (e.g., s anda dized es sco es). 3,7,9 6 Despi e hese indings, li le is known abou whe he and how poo IAQ associa es wi h he 7 s uden s’ pe cep ion o social clima e in schools. The social clima e e lec s, o ins ance, he goals, 8 no ms, in e pe sonal ela ionships and eaching p ac ices, o ganiza ional pa e ns, and school 9 acili ies p esen in a speci ic school. 10,11 Poo s uden -pe cei ed social clima e associa es wi h 10 s uden s’ dec eased well-being and academic pe o mance, 10–13 lowe sel -es eem 14 and 11 p oblems in hei school adjus men 15,16, o example. Gi en ha he pe cei ed social clima e 12 plays an impo an ole in s uden s’ psychosocial well-being and lea ning, i is essen ial o s udy 13 how schools’ IAQ p oblems ela e o i . The need o s udy he e ec s o IAQ on he social clima e 14 in schools is acknowledged in he “indoo ai esea ch” li e a u e as well. 17 15 16 1.1. S uden -pe cei ed social clima e and indoo ai p oblems in schools 17 Al hough he e is no p e ious esea ch assessing whe he and how schools’ IAQ p oblems ela e o 18 he s uden -pe cei ed social clima e, some indi ec e idence sugges s ha he e may be an 19 associa ion be ween hem. P e ious esea ch has associa ed schools’ IAQ p oblems wi h inc eased 20 a es o eache s’ sick lea e and dec eased well-being, 18, 19 and i may induce in e pe sonal 21 con lic s as well as anxie y and ea .20 Fu he mo e, i has been shown ha eache s, who pe cei e 22 hei school’s physical en i onmen unsa is ac o y, epo mo e nega i e a i udes and lowe 23 5 mo al owa d hei wo k and no ice mo e p oblems in he school’s social clima e han eache s 1 who pe cei e he en i onmen as sa is ac o y. 21–23 Finally, also p incipals’ pe cep ions o hei 2 school’s physical en i onmen has been associa ed wi h hei unde s anding o he school’s abili y 3 o deli e ins uc ion.24 4 Howe e , i is unlikely ha IAQ p oblems in luence only school pe sonnel. Fi s , gi en ha 5 school’s o ganiza ional cul u e and eache s’ well-being and a i udes ela e o he s uden - 6 pe cei ed social clima e, 25–28 i is plausible ha also s uden s in schools wi h IAQ p oblems epo 7 lowe social clima e han s uden s in schools wi hou such p oblems. Fu he mo e, p e ious 8 esea ch indica es ha also s uden s’ pe cep ions o hei school’s physical en i onmen associa e 9 wi h how hey pe cei e he school’s social clima e. 11,29 Finally, s uden s’ absence a es ha e been 10 ound o be highe in schools wi h poo IAQ han in schools wi hou such p oblems, 8 which may 11 also e lec issues in he social clima e, since he social clima e associa es wi h school 12 absen eeism. 30 Based on he e idence, i appea s ha schools’ IAQ p oblems may ela e o ac o s 13 which cons i u e he s uden -pe cei ed social clima e, such as pe cei ed social suppo , discipline, 14 and pee ela ionships, in many ways. 15 16 1.2. The p esen s udy 17 In his s udy, we ocus on wo highly ele an ac o s o he s uden -pe cei ed social clima e: 18 pe cei ed quali y o eache -s uden ela ionship and pe cei ed class spi i (i.e., he class oom’s 19 emo ional clima e). 10,31 These ac o s a e e y in luen ial o s uden s’ academic pe o mance, 20 school engagemen , heal h, and heal h- ela ed beha io s, o example. 32–34 The e o e, i is 21 impo an o know whe he and how IAQ p oblems in schools ela e o hese ac o s. Ou i s 22 esea ch ques ion asks whe he he pe cei ed eache -s uden ela ionship and class spi i among 23 6 lowe seconda y school s uden s di e be ween schools wi h obse ed IAQ p oblems and schools 1 wi hou such p oblems. Based on he li e a u e ci ed abo e, 18,20,22,26 we hypo hesize ha s uden s 2 in schools wi h IAQ p oblems epo wo se pe cei ed quali y o he eache -s uden ela ionship 3 and class spi i han s uden s in schools wi hou such p oblems (H1). 4 Ou second esea ch ques ion asks whe he he associa ions be ween s uden -le el 5 cha ac e is ics (e.g., gende ) and he s uden -pe cei ed social clima e a e di e en in schools wi h 6 and wi hou IAQ p oblems. The need o s udy his issue is based on p e ious esea ch showing 7 ha s uden s’ pe cep ions o hei school’s social clima e is associa ed wi h many s uden -le el 8 cha ac e is ics, such as gende , socioeconomic s a us, e hnici y, social suppo , and academic 9 pe o mance; 13,35,36 di e en s uden s pe cei e he social clima e o hei school di e en ly. Gi en 10 ha no only s uden cha ac e is ics bu also con ex ual ac o s can ha e e ec s on he s uden - 11 pe cei ed social clima e, 37 we also es ed whe he he associa ion be ween school-le el 12 cha ac e is ics (e.g., school size) and he s uden -pe cei ed social clima e a e di e en in hese 13 wo school con ex s. Analyzing hese in e ac ions was conside ed impo an since is p o ides 14 in o ma ion on whe he he e a e ce ain ypes o s uden s o schools ha a e especially 15 suscep ible o epo poo eache -s uden ela ionship o poo class spi i when acing IAQ 16 p oblems. Gi en ha he e a e no p e ious s udies conside ing his issue, no hypo hesis is posed. 17 18 19 7 2. Ma e ial and me hods 1 2.1. Da a and pa icipan s 2 The da a we e ob ained om wo sou ces: a) he School Heal h P omo ion S udy (SHP) 2013, 3 ocusing on comp ehensi e school s uden s in g ades eigh and nine (14–16 yea s old) and b) he 4 Benchma king Sys em o Heal h P omo ion Capaci y Building (BSHPCB) da a collec ions om 5 comp ehensi e schools in 2013 ( o mo e de ails, see Finell e al., 2017). 38 The ques ionnai e o 6 BSHPCB is usually illed in by he school’s p incipals. Two a iables om BSHPCB we e used in he 7 analysis. The i s a iable measu ed i and when he iennial inspec ion equi ed by Heal h Ca e 8 Ac 1326/2010 had been ca ied ou in he school. This egula o y inspec ion is done in co- 9 ope a ion wi h he ep esen a i es o he school (e.g., a p incipal), he heal h au ho i ies, and 10 au ho i ies esponsible o cons uc ion and main enance o he school building. I includes a 11 numbe o ac o s (no jus IAQ ela ed) which can in luence he well-being o he school 12 communi y. The inspec ion o he indoo en i onmen includes e iew o exis ing documen s, such 13 as esul s om possible ques ionnai es on IAQ and heal h, IAQ measu emen s, and o he building 14 ela ed epo s as well as a building walk h ough ocusing on a eas whe e p oblems ha e been 15 ound o suspec ed (see Finell e al., 2017; Hie anen-Pel ola and Ko pilah i, 2015). 38, 39 16 The second a iable measu ed whe he o no he e we e biological exposu es obse ed in 17 he school du ing his egula o y inspec ion (see 2.2.2.). BSHPCB ques ionnai e also asks abou 18 physical and chemical exposu es. Howe e , we ocus only on he i em measu ing biological 19 exposu es he e because hese issues, especially dampness and mould (“mouldy schools”) ha e been unde 20 in ensi e public conce n since 1990’s in Finland. In he u u e, i would be in e es ing o pe o m compa a i e 21 analyses on he o he wo i ems o es whe he he associa ions wi h schools’ social clima e would be 22 di e en . Fo he analysis, we included only hose schools whe e a) he inspec ion was ca ied ou in 23 8 2012 o 2013 and b) whe e 1) biological exposu es we e iden i ied du ing he inspec ion bu he 1 p oblems had no been emedia ed o 2) no indoo ai p oblems we e iden i ied. This means ha 2 73 schools (N=10181 s uden s, 27%) we e excluded om he g oup o schools whe e he egula o y 3 iennial inspec ion was ca ied ou in 2012 and 2013. In hese schools a) he biological exposu es 4 we e no included in he inspec ion, b) p oblems we e ound and hey we e al eady co ec ed o c) 5 no in o ma ion we e p o ided (see Measu es). Six een schools wi h ewe han 10 s uden s we e 6 also excluded om he analysis ( o al o 66 s uden s, 0.2%). In addi ion, esponden s who epo ed 7 hei age as a leas wo yea s younge o h ee yea s olde han he a e age age o hei classma es 8 we e excluded om he analysis (n= 64, 0.2%). Two schools wi h 16 and 12 s uden s we e excluded, 9 because hey we e conside ed o be school-le el ou lie s (in he eache -s uden ela ionship). The 10 inal da a consis o 27153 s uden s om 194 schools. Abou 63 pe cen o he s uden s we e om 11 schools wi hou obse ed indoo ai p oblems (17127 s uden s om 127 schools). 12 13 2.2. Measu es 14 2.2.1. Ou come a iables 15 The pe cei ed quali y o he eache -s uden ela ionship was measu ed by h ee i ems: “Teache s 16 a e in e es ed in how I am doing”, “Teache s ea us, he s uden s, ai ly”, and “The opinions o 17 s uden s a e aken in o conside a ion in he de elopmen o school wo k”. The esponse scale was 18 1 = ully ag ee, 2 = ag ee, 3 = disag ee, and 4 = ully disag ee. These i ems ha e been used in many 19 p e ious s udies as indica o s o he eache -s uden ela ionship. 27,40,41 We modeled eache - 20 s uden ela ionship as a la en ac o wi h hese h ee i ems. The s anda dized loadings o he 21 con i ma o y ac o analysis a e p esen ed in Table 2. The i indica o s a e epo ed in Resul 22 sec ion (see also Suppo ing in o ma ion). The C onbach alpha was 0.63. 23 9 The class spi i was measu ed by h ee i ems: “The s uden s in my class ge along well”, “The 1 class oom discipline in my class is good” and “The mood in ou class is such ha I da e o exp ess 2 my opinion eely”. The esponse scale was he same as abo e. These i ems ha e been used 3 p e iously by Ka onen and colleagues 40 as an indica o o class spi i . As abo e, we modeled class 4 spi i as a la en ac o . The s anda dized loadings a e p esen ed in Table 2. The i indica o s a e 5 epo ed in Resul sec ion (see also Suppo ing in o ma ion). The C onbach alpha was 0.67. The 6 da a sou ce o he ou come a iables was he SHP. 7 8 2.2.2. P edic o 9 The BSHPCB su ey i em “We e he ollowing issues e alua ed in he mos ecen inspec ion: 10 Exposu e o biological agen s (indoo ai , mold, e c.)” had he ollowing esponse op ions: no da a 11 a ailable; no included in he inspec ion; inspec ed, no de iciencies de ec ed; inspec ed, 12 de iciencies de ec ed bu no ye co ec ed; inspec ed, de iciencies de ec ed and co ec ed. In his 13 s udy, we ocus only on he ollowing op ions: 0 = inspec ed, no de iciencies de ec ed and 1 = 14 inspec ed, de iciencies de ec ed bu no ye co ec ed. The esul ing a iable is e e ed o as 15 “obse ed IAQ p oblems”. 16 17 2.2.3. Backg ound a iables 18 Gende (0 = boy, 1 = gi l), age, pe cei ed heal h, pa en al suppo , and educa ion o he mo he 19 we e included only a he s uden -le el (see S a is ical analyses and Figu es 1 and 2 below). 20 Pe cei ed heal h was measu ed by he i em: “How is you heal h in gene al?” The esponse scale 21 was 1 = e y good, 2 = ai ly good, 3 = mode a e, and 4 = ai ly o e y poo The scale was 22 ca ego ized in o wo ca ego ies (0 = e y good, 1 = no e y good), because only a small 23 p opo ion epo ed mode a e o bad heal h. The simila kind o ca ego iza ion has been used in 24 16 3.1. The associa ion be ween obse ed IAQ p oblems and he s uden -pe cei ed schools’ social 1 clima e 2 Fi s , we es ed he i o each measu emen model (see Figu es 1 and 2) by ca ying ou a wo- 3 le el con i ma o y ac o analysis o each a iable sepa a ely. In o de o do ha we ixed he 4 loadings equally a he s uden and school-le els. 47 The model i s we e good bo h o he 5 eache -s uden ela ionship (RMSEA = 0.01, CFI = 1.00, SRMR: s uden -le el = 0.00, SRMR: school- 6 le el = 0.08) and class spi i (RMSEA = 0.01, CFI = 1.00, SRMR: s uden -le el = 0.00, SRMR: school- 7 le el = 0.05). The s anda dized ac o loadings a e p esen ed in Table 2 (see u he in o ma ion on 8 measu emen models and he i indica o s in Suppo ing in o ma ion). 9 Table 2. S anda dized loadings o he eache -s uden ela ionship and class spi i la en 10 a iables (s uden -le el: N = 27018–27025/ School-le el: N = 194). 11 Teache -s uden ela ionship Class spi i S uden -le el S uden -le el λ λ The opinions o s uden s a e aken in o conside a ion in he de elopmen o school wo k. 0.65* The s uden s in my class ge along well. 0.73* Teache s a e in e es ed in how I am doing. 0.56* The class oom discipline in my class is good. 0.54* Teache s ea us, he s uden s, ai ly. 0.60* The mood in ou class is such ha I da e o exp ess my opinion eely. 0.63* School-le el School-le el The opinions o s uden s a e aken in o conside a ion in he de elopmen o school wo k. 0.91* The s uden s in my class ge along well. 0.87* Teache s a e in e es ed in how I am doing. 0.67* The class oom discipline in my class is good. 0.57* Teache s ea us, he s uden s, ai ly. 0.71* The mood in ou class is such ha I da e o exp ess my opinion eely.a 1.00* * p < 0.001 12 a The esidual a iance is se o ze o a he school-le el. 13 14 17 The e was a s a is ically signi ican a iabili y bo h in he la en eache -s uden ela ionship 1 - a iable wi hin (σ2W = 0.204, p < 0.001) and be ween schools (σ2B = 0.010, p < 0.001) and in he 2 la en school spi i - a iable wi hin (σ2W = 0.276, p < 0.001) and be ween schools (σ2B = 0.012, p < 3 0.001). The ICC o he bo h la en a iables we e 0.05 and 0.04 espec i ely, meaning ha abou 4 4–5% o he a iabili y occu ed be ween schools. Al hough he ICCs we e ela i ely small, hey 5 had s ong design e ec s (DEFF;61 eache -s uden ela ionship: DEFF = 7.46; school spi i : DEFF = 6 6.76). 7 Then we es ima ed wo andom in e cep models (i.e., Model 1a and Model 1b) in o de o 8 es whe he he pe cei ed eache -s uden ela ionship and class spi i di e ed be ween schools 9 wi h and wi hou obse ed IAQ p oblems (see Table 3). The model i s we e good bo h o he 10 eache -s uden ela ionship (Model 1a: RMSEA = 0.02, CFI = 0.97, SRMR: s uden -le el = 0.01, 11 SRMR: school-le el = 0.07, see Table 2) and o school spi i (Model 1b: RMSEA = 0.02, CFI = 0.99, 12 SRMR: s uden -le el = 0.01, SRMR: school-le el = 0.06, see Table 2). A e all backg ound a iables 13 we e inse ed in o Model 1a, obse ed IAQ p oblems associa ed signi ican ly wi h he eache - 14 s uden ela ionship. The e ec size was 0.5, indica ing a medium-sized e ec 57. Howe e , a e 15 all backg ound a iables we e inse ed in o Model 1b, obse ed IAQ p oblems did no associa e 16 signi ican ly wi h school spi i (see u he in o ma ion on uni a ia e associa ions in Suppo ing 17 in o ma ion) . 18 In Model 1a, he mos impo an s uden -le el backg ound a iables we e pa en al suppo , 19 pe cei ed heal h, and GPA. The mos impo an school-le el backg ound a iables we e e hnic 20 composi ion (measu ed by s uden s’ sel - epo ed e hnic backg ound) and he school’s a e age 21 GPA. In Model 1b, he mos impo an s uden -le el backg ound a iables we e pa en al suppo , 22 gende , and pe cei ed heal h. The mos impo an school-le el backg ound a iables we e school 23 18 size and he school’s a e age GPA (see Table 2). Model 1a explained 11% o he o al a iance a 1 he s uden -le el (R2 = 0.114) and 22% a he school-le el (R2 = 0.222). Model 1b explained 8% o 2 he o al a iance a he s uden -le el (R2 = 0.077) and 16% a he school-le el (R2 = 0.157), 3 espec i ely (see Table 2). 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 19 Table 3. Pa ame e es ima es o he eache -s uden ela ionship and he class oom’s social 1 clima e1 as a unc ion o s uden - and school-le el a iables: A andom in e cep model 2 (N=27153 s uden s/194 schools). 3 Model 1a Teache -s uden ela ionship Model 1b Class spi i B (SE)a β (SE)b B (SE)a β (SE)b S uden -le el Gende c -0.06 (0.01)*** -0.07 (0.01)*** 0.13 (0.01)*** 0.13 (0.01)*** Aged 0.01 (0.01) 0.02 (0.01) -0.02 (0.01) -0.02 (0.01) Age squa ede -0.06 (0.01)*** -0.06 (0.01)*** Pe cei ed heal h 0.08 (0.01)*** 0.09 (0.01)*** 0.16 (0.01) *** 0.15 (0.01)*** Pa en al suppo g 0.12 (0.00)*** 0.26 (0.01)*** 0.07 (0.01)*** 0.13 (0.01)*** Pa en al suppo squa ede 0.03 (0.00)*** 0.06 (0.01)*** 0.02 (0.01)*** 0.03 (0.01)*** Educa ion o mo he h 0.05 (0.01)*** 0.06 (0.01)*** -0.02 (0.01) -0.01 (0.01) G ade poin a e agei -0.06 (0.01)*** -0.11 (0.01)*** 0.01 (0.01)* 0.02 (0.01)* G ade poin a e age squa ede -0.02 (0.01)* -0.02 (0.01)* E hnic backg oundj -0.02 (0.03) -0.01 (0.01) 0.08 (0.03)** 0.03 (0.01)** School-le el Obse ed IAQ p oblemsk 0.05 (0.02)** 0.25 ( 0.08)** 0.02 (0.02) 0.08 (0.08) G ade poin a e aged -0.22 (0.09)*l -0.28 (0.10)** -0.24 ( 0.09) ** -0.27 (0.10)** E hnic backg ound -0.36 (0.12)** -0.19 (0.09)*n 0.03 (0.19) 0.02 (0.09) School size o 0.01 (0.00)m 0.16 (0.08)* 0.01 (0.01)* 0.24 (0.09)* σ2W 0.17 (0.01) 0.89 (0.01) 0.24 (0.01) 0.92 (0.01) σ2B 0.01 (0.00) 0.78(0.07) 0.01 (0.00) 0.84 (0.08) *p < 0.05, **p < 0.01, *** p < 0.001 4 1 A highe alue in he eache -s uden ela ionship and he class oom’s social clima e means mo e p oblems. 5 a Uns anda dized be a 6 b S anda dized be a 7 c Scale 0–1 (0 = boy) 8 d G and mean cen e ed 9 e Quad a ic e m. Only i he quad a ic e m was signi ican (p < 0.05), i was included in he inal model and epo ed. 10 The e we e no signi ican quad a ic e ms a he school-le el. 11 Scale 0–1 (0 = e y good heal h) 12 20 g Scale 1–4 (1 = good suppo , g and mean cen e ed) 1 h Scale 0–1 (1 = uni e si y educa ion) 2 i Scale 0–2 (0 = GPA less han 7.5, g and mean cen e ed) 3 j Scale 0–1 (0 = one o bo h pa en s a e bo n in Finland) 4 k Scale 0–1 (0 = no obse ed IAQ p oblems in he school) 5 l p = 0.012 6 m p = 0.064 7 n p = 0.027 8 o In hese models he scale was changed as ollows: school size/100. 9 10 3.2. The e ec o he school’s IAQ p oblems on he associa ions be ween he backg ound and 11 ou come a iables 12 We es ed di e ences on he associa ions be ween a) s uden -le el cha ac e is ics and he 13 s uden -pe cei ed social clima e and b) school-le el cha ac e is ics and he s uden -pe cei ed 14 social clima e in schools wi h and wi hou IAQ p oblems. In o de o es hese di e ences we 15 used mul ig oup wo-le el s uc u al equa ion modeling and Wald- es . The associa ion be ween 16 s uden -le el GPA and class spi i was mode a ed by obse ed IAQ p oblems (Wald- es = 6.04, p = 17 0.014). This associa ion was s onge in schools wi h IAQ p oblems (s anda dized be a = 0.05, p < 18 0.001) han in schools whe e he e we e no such p oblems (s anda dized be a = 0.00, p = 0.821; 19 see u he in o ma ion on uni a ia e associa ions in Suppo ing in o ma ion). In o de o be e 20 unde s and his esul , we an h ee sepa a e analyses. We es ed s uden s whose GPA was a) less 21 han 7.5, b) be ween 7.5 and 8.4 and c) o e 8.4. Excep o GPA, he o he backg ound a iables 22 in hese h ee analyses we e he same as in Model 1b (see Table 2). S uden s wi h he highes GPA 23 (i.e., o e 8.4) pe cei ed mo e p oblems in class spi i in schools wi h obse ed IAQ p oblems han 24 in schools wi hou such p oblems (s anda dized be a = 0.19, p = 0.032). The e ec size is 0.4, 25 indica ing a small-sized e ec . 57 The e we e no di e ences be ween schools among s uden s in 26 he lowe GPA clus e s (GPA 7.5–8.4: s anda dized be a = 0.12, p = 0.193; GPA < 7.5: s anda dized 27 be a = 0.01, p = 0.948). 28 21 1 Figu e 3. The mean di e ence o class spi i be ween schools wi h and wi hou obse ed IAQ 2 p oblems di ided by he s anda d de ia ion o class spi i . 3 0 = GPA < 7.5, p = 0.948; 1 = GPA 7.5–8.4, p = 0.193; 2= GPA > 8.4, p = 0.032 (N = 193–194 4 schools). 5 6 Fu he mo e, he linea e ms o he pa en al suppo and eache -s uden associa ions we e 7 signi ican ly di e en (Wald- es = 4.75, p = 0.029) be ween he wo school con ex s (no obse ed 8 IAQ p oblems: s anda dized be a = 0.27, p < 0.001; obse ed IAQ p oblems: s anda dized be a = 9 0.23, p < 0.001; see u he in o ma ion on uni a ia e associa ions in Suppo ing in o ma ion). The 10 Wald- es be ween quad a ic e ms was no signi ican (Wald- es = 0.50, p = 0.479). This small 11 e ec indica es ha he di e ence in he pe cei ed quali y o he eache -s uden ela ionship 12 be ween he wo school con ex s was la ge among s uden s wi h high pa en al suppo han 13 s uden s wi h low pa en al suppo . In o al, s uden s wi h low pa en al suppo , howe e , 14 e alua ed hei eache -s uden ela ionship mo e nega i ely han s uden s wi h high pa en al 15 suppo . 16 17 18 G ade poin a e age Class spi i 22 4. Discussion 1 Ou main inding was ha on a e age, s uden s in schools wi h obse ed IAQ p oblems epo ed 2 wo se eache -s uden ela ionship han s uden s in schools wi hou such p oblems. Fu he mo e, 3 we ound ha s uden s wi h a high GPA epo ed wo se class spi i in schools wi h IAQ p oblems 4 han in schools wi hou such p oblems. 5 P e ious esea ch has epo ed associa ions be ween poo IAQ in schools and bo h heal h 6 and lea ning ou comes (e.g., espi a o y symp oms, illness absence, and academic pe o mance). 7 3,7,9,62 Ou indings con ibu e o his esea ch by showing ha IAQ p oblems in schools may ha e 8 an e ec on he s uden -pe cei ed social clima e. To ou knowledge, his is he i s ime when 9 hese e ec s ha e been assessed. Gi en ha social clima e is shown o in luence a ious ac o s, 10 such as academic pe o mance, absence a es, and s uden s’ psychosocial well-being, 10,30,31 ou 11 indings a e impo an . Fu he mo e, we ound ha a school’s obse ed IAQ p oblems signi ican ly 12 modi ied he associa ions be ween s uden -le el cha ac e is ics and he s uden -pe cei ed social 13 clima e. Al hough hese associa ions we e weak, he indings suppo ed he p e ious esea ch by 14 Finell and colleagues38 by concluding ha schools wi h and wi hou IAQ p oblems may ep esen 15 wo di e en social con ex s. They also sugges ha he social and psychological consequences o 16 IAQ p oblems in schools esul om complex p ocesses whe e bo h he s uden s’ and school 17 pe sonnel’s cha ac e is ics play a ole. 18 An impo an ask o u u e esea ch is o examine why IAQ p oblems in schools may 19 in luence he s uden -pe cei ed social clima e. Ou esul s and he li e a u e p o ide h ee 20 po en ial in e wined pa hs. The i s pa h ela es o he impac o poo IAQ on school pe sonnel, 21 o example by dec easing hei well-being, inc easing sick lea e, and causing possible p oblems in 22 managemen and decision-making p ocedu es. 19,20,26,28,63 The second pa h ela es o he impac o 23 23 poo IAQ on s uden s. The s uden s’ pe cei ed heal h is associa ed wi h he s uden -pe cei ed 1 social clima e, 10 as was also ound in ou s udy. The hi d pa h ela es o he poo condi ion o he 2 school acili ies as well as an icipa ed emedia ion o he school building. 22,23 In Finland, school 3 gene al managemen and acili y managemen a e independen om each o he . In o he wo ds, 4 school p incipals and s a do no ha e a di ec ole o o e sigh o he acili y managemen . They 5 can epo any issues o conce ns bu he alloca ion o unds o school acili y managemen and 6 emedia ion decisions a e made by he municipali y echnical boa d. Howe e , emedia ion 7 p ocesses o en esul in dis u bances (e.g., mo ing o empo al buildings and/o noise), which can 8 be s ess ul bo h o school pe sonnel and s uden s, and may he e o e in luence he s uden - 9 pe cei ed social clima e. 10 F om a p ac ical poin o iew, an impo an ques ion is how i is possible o p e en he 11 po en ial social consequences caused by IAQ p oblems in schools. The main ask is na u ally o 12 in es iga e and p ope ly emedia e indoo ai p oblems in he school building. Howe e , hese 13 p ocesses a e o en slow. In case he emedia ion is delayed, he p e en ion o heal h p oblems 14 among eache s and s uden s can in ol e sou ce con ol (e.g., mo e equen cleaning), enhanced 15 en ila ion, and exposu e p e en ion (e.g., no using p oblema ic a eas). Fu he mo e, 16 managemen and decision-making p ocedu es should be de eloped in o de o a oid con lic s 17 inside he school o ganiza ion and expe iences o injus ice.64 Open communica ion and sha ing 18 up- o-da e ( egula , ac ual) in o ma ion wi h school pe sonnel, s uden s, and pa en s should help 19 o main ain us be ween pa ies. 20 The s eng hs o he s udy include he la ge and ep esen a i e sample. The use o mul ile el 21 s uc u al equa ion models has many ad an ages, including be e s a is ical powe compa ed o 22 models using agg ega ed school-le el da a and a oiding spu ious indings om using s uden -le el 23 24 da a wi hou aking he hie a chical da a s uc u e in o accoun . 48 The limi a ions o he s udy 1 include he c oss-sec ional design and he use o ques ionnai e da a. Fu he mo e, we did no 2 ha e physical measu es. This means ha we ha e o ely on p inciples epo on he indings o 3 iennial inspec ion on biological exposu es and ha we a e canno be e dis inguish schools 4 based on he ac o s ha ha e caused hei IAQ p oblems. I is likely ha he e exis s a ia ion in 5 he se e i y o schools’ IAQ p oblems and hei social consequences. We also do no know he 6 deg ee eache s and s uden s we e awa e o hei school’s IAQ p oblems. In he u u e, hese 7 issues should be u he s udied using longi udinal da a and physical measu es. Fu u e esea ch 8 should also de ec ac o s ha can media e he associa ion be ween school’s indoo en i onmen 9 and he pe cei ed social clima e. Also possible building- ela ed con ounde s (e.g., poo sani a y 10 condi ions o unpleasan school en i onmen ) should be conside ed. 11 The p esen s udy is pa o a la ge p ojec aiming o cla i y associa ions be ween 12 psychosocial ac o s and IAQ p oblems among comp ehensi e school s uden s in Finland. In a 13 p e ious s udy, we ound associa ions be ween bo h he physical and psychosocial en i onmen 14 and subjec i e IAQ in schools. 38 In he p esen s udy, psychosocial ac o s a e he ou come 15 a iables. Ou s udy unde lines he ac ha he associa ions be ween he pe cep ion o IAQ 16 p oblems and psychosocial ac o s a e highly complex and a e likely o in luence each ano he in 17 a ious ways. 18 19 20 21 22 25 5. Conclusions 1 Ou s udy ound ha p oblems in a school’s IAQ associa ed wi h he s uden -pe cei ed social 2 clima e. Fu he mo e, we ound ha he school’s obse ed IAQ p oblems signi ican ly modi ied 3 he associa ions be ween s uden -le el cha ac e is ics and he s uden -pe cei ed social clima e. 4 Ou s udy poin s o a need o s udy schools’ IAQ p oblems om a mul idisciplina y pe spec i e in 5 o de o be e unde s and how IAQ p oblems in luence s uden s’ psychosocial well-being. 6 Fu he mo e, om p ac ical poin o iew, when e alua ing he u gency o emedia ion, also IAQ 7 p oblems’ possible social consequences should be aken in o accoun . Fas and e ec i e 8 emedia ion p ocess no only p e en ad e se heal h e ec s bu can also p e en p oblems in 9 school’s social clima e. 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 32 59. Rubin DB. In e ence and missing da a. Biome ika. 1976;63:581–592. 1 60. Ende s C, Bandalos D. The ela i e pe o mance o ull in o ma ion maximum likelihood 2 es ima ion o missing da a in s uc u al equa ion models. S uc . Equ. Model. Mul idiscip. J. 3 2001;8:430–457 . 4 61. Ukoumunne OC, Gulli o d, MC, Chinn S, e al. Me hods o e alua ing a ea-wide and 5 o ganisa ion-based in e en ions in heal h and heal h ca e: a sys ema ic e iew. Heal h 6 Technol. Assess. 1999;3. 7 62. Mendell MJ, Elisee a EA, Da ies MM, Lobscheid A. Do class oom en ila ion a es in Cali o nia 8 elemen a y schools in luence s anda dized es sco es? Resul s om a p ospec i e s udy. 9 Indoo Ai . 2016;26:546–557. 10 63. Mille RT, Mu nane RJ, Wille JB. Do eache absences impac s uden achie emen ? 11 Longi udinal e idence om one u ban school dis ic . Educ. E al. Policy Anal. 2008; 30: 181– 12 200. 13 64. Finell E, Seppälä T. Indoo ai p oblems and expe iences o injus ice in he wo kplace: A 14 quan i a i e and a quali a i e s udy. Indoo Ai . 2017; doi:10.1111/ina.12409. 15 16 17 18 19 20 21 22 23 24 25 33 1 2 Figu e 1 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 = measu emen model 33 34 Figu e 1. Hypo hesized s uc u al equa ion model: The eache -s uden ela ionship as a la en 35 ou come a iable. 36 Teache -s uden ela ionship Teache s a e in e es ed in how I am doing Gende Teache s ea us, he s uden s, ai ly The opinions o s uden s a e aken in o conside a ion in he de elopmen o school wo k Age Pe cei ed heal h Pa en al suppo GPA E hnic backg ound Educa ion o mo he Teache -s uden ela ionship S uden -le el School-le el Teache s a e in e es ed in how I am doing Teache s ea us, he s uden s, ai ly The opinions o s uden s a e aken in o conside a ion in he de elopmen o school wo k GPA E hnic backg ound School size Obse ed IAQ p oblems 34 1 Figu e 2 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 = measu emen model 32 33 Figu e 2. Hypo hesized s uc u al equa ion model: Class spi i as a la en ou come a iable. 34 Class-spi i S uden -le el School-le el The s uden s in my class ge along well. The class oom discipline in my class is good. The mood in ou class is such ha I da e o exp ess my opinion eely. GPA E hnic backg ound School size Obse ed IAQ p oblems Gende Age Pe cei ed heal h Class-spi i Pa en al suppo The s uden s in my class ge along well. The class oom discipline in my class is good. The mood in ou class is such ha I da e o exp ess my opinion eely. GPA E hnic backg ound Educa ion o mo he