The mediating role of constructs representing reasoned-action and automatic processes on the past behavior-future behavior relationship

This is a sel -a chi ed e sion o an o iginal a icle. This e sion may di e om he o iginal in pagina ion and ypog aphic de ails. Au ho (s): Ti le: Yea : Ve sion: Copy igh : Righ s: Righ s u l: Please ci e he o iginal e sion: CC BY-NC-ND 4.0 h ps://c ea i ecommons.o g/licenses/by-nc-nd/4.0/ The media ing ole o cons uc s ep esen ing easoned-ac ion and au oma ic p ocesses on he pas beha io - u u e beha io ela ionship © 2020 Else ie Accep ed e sion (Final d a ) B own, D. J.; Hagge , M. S.; Hamil on, K. B own, D. J., Hagge , M. S., & Hamil on, K. (2020). The media ing ole o cons uc s ep esen ing easoned-ac ion and au oma ic p ocesses on he pas beha io - u u e beha io ela ionship. Social Science and Medicine, 258, A icle 113085. h ps://doi.o g/10.1016/j.socscimed.2020.113085 2020 Jou nal P e-p oo The Media ing Role o Reasoned-Ac ion and Au oma ic P ocesses om Pas - o- Fu u e Beha io Daniel J. B own, Ma in S. Hagge , Ky a Hamil on PII: S0277-9536(20)30304-X DOI: h ps://doi.o g/10.1016/j.socscimed.2020.113085 Re e ence: SSM 113085 To appea in: Social Science & Medicine Re ised Da e: 18 Feb ua y 2020 Accep ed Da e: 22 May 2020 Please ci e his a icle as: B own, D.J., Hagge , M.S., Hamil on, K., The Media ing Role o Reasoned- Ac ion and Au oma ic P ocesses om Pas - o-Fu u e Beha io , Social Science & Medicine, h ps:// doi.o g/10.1016/j.socscimed.2020.113085. This is a PDF ile o an a icle ha has unde gone enhancemen s a e accep ance, such as he addi ion o a co e page and me ada a, and o ma ing o eadabili y, bu i is no ye he de ini i e e sion o eco d. This e sion will unde go addi ional copyedi ing, ypese ing and e iew be o e i is published in i s inal o m, bu we a e p o iding his e sion o gi e ea ly isibili y o he a icle. Please no e ha , du ing he p oduc ion p ocess, e o s may be disco e ed which could a ec he con en , and all legal disclaime s ha apply o he jou nal pe ain. © 2020 Published by Else ie L d. The Media ing Role o Reasoned-Ac ion and Au oma ic P ocesses om Pas - o- Fu u e Beha io Daniel J. B own G i i h Uni e si y Ma in S. Hagge Uni e si y o Cali o nia, Me ced and Uni e si y o Jy äskylä Ky a Hamil on G i i h Uni e si y Au ho No e Daniel J. B own, School o Applied Psychology, Menzies Heal h Ins i u e Queensland, G i i h Uni e si y, B isbane, Queensland, Aus alia; Ma in S. Hagge , Psychological Sciences, Uni e si y o Cali o nia, Me ced, USA and Facul y o Spo and Heal h Sciences, Uni e si y o Jy äskylä, Jy äskylä, Finland; Ky a Hamil on, School o Applied Psychology, Menzies Heal h Ins i u e Queensland, G i i h Uni e si y, B isbane, Queensland, Aus alia. Funding: This esea ch did no ecei e any speci ic g an om unding agencies in he public, comme cial, o no - o -p o i sec o s. This p ojec was suppo ed by he Aus alian Go e nmen Resea ch T aining P og am. Ma in S. Hagge ’s con ibu ion was suppo ed by a Finland Dis inguished P o esso (FiDiP o) ellowship om Tekes, he Finnish unding agency o inno a ion. *Co espondence conce ning his a icle should be add essed o Daniel B own, Heal h and Psychology Inno ai ons (HaPI) Resea ch Lab, School o Applied Psychology, G i i h Uni e si y, 176 Messines Ridge Road, M G a a , QLD 4122. Email: [email p o ec ed] Figu e 1. Rela ions among p oposed model cons uc s including s anda dized pa ame e es ima es. Es ima es on he uppe , cen e , and lowe lines a e es ima es o he binge d inking, lossing, and sun sa e y beha io s, espec i ely. E ec s o age, gende , and educa ion on each model cons uc ha e been omi ed o cla i y. In en ion T2 Au oma ici y Beha io T1 Au oma ici y Pe cei ed Beha io al Con ol A i ude Subjec i e No ms F equency .344 *** .426*** .719*** .105 .263*** .135 .111 .094 -.164* Recency Rou ine Pas Beha io .601 *** .530*** .364*** .391 *** .249*** -.050 .203 ** .081 -.015 .917 .967 .919 .328 *** .345*** .363*** .670 *** .899*** .710*** .420 *** .424*** .466*** .454 *** .301*** .369*** .104 .147* .089 .158 * .205** -.087 .366 *** .526*** .428*** .204 ** .514*** .383*** .902 .941 .933 .884 .897 .938 Table 1 Model Fi and Quali y Indices o S uc u al Equa ion Models o Binge D inking, Flossing, and Sun Sa e y Index Binge d inking Flossing Sun sa e y GoF .575 .678 .592 AR 2 .382 *** .518 *** .382 *** APC .164 ** .187 ** .201 ** AVIF 1.272 1.601 1.342 No e. * p < .05 ** p < .01 *** p < .001 1 In oduc ion To de elop e ec i e in e en ions in o de o modi y people’s beha io one needs o i s isola e he mechanisms ha guide he beha io and hen es he ex en o which he mechanisms magni y o diminish beha io al engagemen . P e ious esea ch has o en u ned o heo ies o social cogni ion o guide in es iga ions aimed a iden i ying he de e minan s o heal h beha io s and, impo an ly, he p ocesses by which hese de e minan s ela e o each o he and he beha io . A close examina ion o he majo heo ies ha ha e been applied o he unde s anding o heal h beha io assumes ha beha io is de e mined by a easoned, in en ional p ocess in which an indi idual in es s e o in o de o pu sue an ac ion. The heo y o planned beha io (TPB: Ajzen, 1991), gi en in en ion is cen al o he model, is pe haps he mos widely used social–cogni i e heo y o beha io . Acco ding o he model, in en ion is he mos p oximal p edic o o beha io , wi h in en ion de e mined by h ee social–cogni i e a iables: a i ude (o e all e alua ions o pe o ming he beha io ), subjec i e no m (social p essu e om impo an o he s o pe o m he beha io ), and pe cei ed beha io al con ol (pe cei ed amoun o con ol o e beha io al pe o mance; also mo e ecen ly heo ized as mode a ing he in en ion-beha io ela ionship; Ajzen, 1991). Me a-analy ic s udies suppo he use o he TPB in p edic ing heal h- ela ed beha io s (e.g., McEachan e al., 2011; Rich e al., 2015) Social cogni ion heo ies such as he TPB end o ocus on a ela i ely na ow se o de e minan s ha include cons uc s ha ep esen easoned, in en ional de e minan s o ac ion. Such app oaches end no o explici ly accoun o he pe asi e e ec s o pas beha io on key cons uc s o psychological heo ies and hei ela ions wi h heal h beha io s (e.g., Alba acín e al., 2001; Conne e al., 1999; Hagge e al., 2018). Howe e , esea ch has demons a ed ha including pas beha io in hese heo ies accoun s o subs an i e addi ional a iance in subsequen ly measu ed ( u u e) beha io h ough he esidual e ec o 2 pas beha io on u u e beha io . Inclusion o pas beha io has also been shown o a enua e he size o he e ec s o in en ion and o he social-cogni ion cons uc s on u u e beha io . Some a gue ha hese esidual e ec s a e likely an a i ac o assessing pas beha io and u u e beha io using he same measu e, leading o sha ed me hod a iance (Ajzen, 1991, 2002). Howe e , esea ch using di e en measu es o pas and u u e beha io has e ealed esidual e ec s (B own e al., 2018; Ve planken, 2006). O he s, he e o e, a gue ha he esidual e ec s a o ded by pas beha io may model non-conscious p ocesses including habi s and decisions based on implici cogni ion o beha io al ‘sc ip s’ (Hagge , 2020; Hagge & Cha zisa an is, 2014; S ack & Deu sch, 2004; T iandis, 1977). In seeking o unde s and he e ec s o pas beha io in social-cogni i e heo ies, Ouelle e and Wood (1998) p oposed wo pa hways by which pas beha io ela es o u u e beha io : a di ec pa hway and an indi ec pa hway. The di ec e ec is said o model an au oma ic p ocess, simila o ha p oposed in dual-p ocess heo ies (S ack & Deu sch, 2004). The second, indi ec pa hway, in which pas beha io a ec s u u e beha io ia conscious, in en ional p ocesses, such as hose cha ac e ized in social-cogni ion heo ies (Ajzen, 1991). Ouelle e and Wood (1998) hypo hesized ha well-p ac iced beha io s, occu ing in consis en con ex s likely e lec habi ual pa e ns ha a e expec ed o be au oma ically epea ed in he u u e. These au oma ic beha io s can be in en ional and goal- dependen o be pe cei ed as non- oli ional and coun e -in en ional (i.e., ou side o an indi idual’s awa eness/con ol o in opposi ion o an indi idual’s in en ion; Ga dne e al., 2015; Ji & Wood, 2007). Con e sely, he au ho s desc ibe how no el o less p ac iced beha io s a e likely go e ned by conscious, in en ional p ocesses. While pas beha io has been shown o a enua e he e ec s o social cogni i e a iables (B own e al., 2018), s udies sugges ha hei cons uc s s ill accoun o unique a iance in beha io and media e, a leas pa ially, he e ec s o pas beha io on u u e beha io (Hagge e al., 2018). 3 Al hough p e ious esea ch has demons a ed suppo o he di ec and indi ec e ec s o pas beha io on u u e beha io in social cogni ion heo ies models (Ouelle e & Wood, 1998), ew s udies ha e es ed he e ec s o o he cons uc s ha e lec non- conscious decision making, such as habi s, on beha io alongside pas beha io (Bambe g e al., 2003; an B ee e al., 2015; Ve planken, 2006). I indi ec e ec s o pas beha io on u u e beha io model au oma ic, spon aneous p ocesses o beha io al enac men as p oposed in dual p ocess heo ies (S ack & Deu sch, 2004) and, hus, cap u e he au oma ici y componen o habi hey would be expec ed o media e he e ec o pas beha io on subsequen beha io . We aimed o examine hese p oposi ions in he p esen s udy by es ing he TPB in a ange o heal h beha io s and included measu es o pas beha io and beha io al au oma ici y ac oss wo ime-poin s. An addi ional issue wi h esea ch examining he ole o pas beha io in social cogni ion heo ies is he la ge a ia ion in a bi a y ime ames used o measu e pas beha io . Fo example s udies ha e measu ed pas beha io o e one week (Mullan e al., 2015), ou weeks (Caudwell e al., 2019), and six mon hs (Luszczynska & Cieslak, 2009). Measu ing pas beha io o e a bi a y ime ames may miss pa e ns o beha io al engagemen ha could po en ially be in o ma i e on he mechanisms and p ocesses ha de e mine subsequen heal h beha io . Fo example, engagemen in sun sa e y beha io s is likely o be highly seasonal in ha hey a e pe o med equen ly in he summe mon hs and seldom in win e mon hs (Sun e al., 2014; Xiang e al., 2015). A measu e o pas beha io ha e e s o sun sa e y beha io s in close p oximi y o subsequen beha io is likely o lead o s onge pas beha io e ec s han i he ime ame o he pas beha io measu e ex ended beyond he cu en season. Fu he mo e, Ouelle e and Wood (1998) sugges ed ha a di e en pa e n o pas beha io e ec s will occu o in equen ly compa ed o egula ly pe o med beha io s. I would seem easonable o accoun o bo h long- e m (i.e., dis al) and 4 sho - e m (i.e., ecen ) enac men s o pas beha io as well as he equency wi h which he beha io is measu ed. Pe ugini and Bagozzi (2001) p oposed di e en ia ing be ween long- e m (up o 1 yea ) and mo e ecen (wi hin he las mon h) measu es o pas beha io . Simila ly, o he s ha e used i ems ela ing he ex en o which he beha io is ou inized o u he cap u e he egula i y o daily beha io al pe o mance (Ve planken & O bell, 2003). Ex ending his esea ch, he p esen s udy es ed e ec s o pas beha io using mul iple me hods ha accoun o he equency, ecency, and ou iniza ion o beha io s. Speci ically, he measu e included i ems assessing long- e m (las yea ), sho - e m (las mon h), and ou ine pa e ns o beha io , p oposing ha hese would indica e a mul i-componen measu e o pas beha io ha would cap u e he essence o indi iduals pas pe o mance o heal h beha io s. The Cu en S udy and Hypo heses Based on Ouelle e and Wood’s (1998) p oposi ions, we p opose a se o key hypo heses ela ing o he e ec s o easoned, in en ional pa hways and non-conscious, au oma ic pa hways o ac ion o h ee heal h beha io s in h ee independen samples: binge d inking in uni e si y s uden s, den al lossing in communi y-dwelling adul s, and pa en al sun sa e y beha io s o hei 2 o 5 yea -old child en. The p oposed model is p esen ed in Figu e 1. In he p esen model, he easoned, in en ional p ocesses a e ep esen ed by he e ec s o he social-cogni ion cons uc s om he TPB. Speci ically, a i ude, subjec i e no m, and pe cei ed beha io al con ol a e expec ed o signi ican ly and di ec ly p edic in en ions and in en ions and pe cei ed beha io al con ol a e expec ed o signi ican ly and di ec ly p edic he a ge beha io s. We also expec he e ec s o he TPB cons uc s on beha io would be media ed by in en ions (Ajzen, 1991). 11 suppo ing cons uc alidi y o he ac o s. Impo an ly, he second-o de ac o loadings o he equency, ecency, and ou ine indica o s o he pas beha io la en a iable we e la ge (>.884) and s a is ically signi ican (p < .001) o all beha io s. Composi e (ρ) eliabili y coe icien s, AVE, and in e co ela ions o model a iables a e p esen ed in Appendix C (supplemen al ma e ials). Reliabili y coe icien s exceeded he .700 c i e ion, and AVE alues exceeded he ecommended .500 c i e ion. Co ela ions among he la en a iables also indica ed no p oblems wi h disc iminan alidi y. Missing alues analysis using Li le’s (1988) missing comple ely a andom (MCAR) es e ealed a signi ican alue o he lossing beha io sample (χ 2 = 437.599, d = 357, p = .002), bu no o he binge d inking beha io (χ 2 = 431.810, d = 416, p = .286) o pa en sun sa e y beha io s (χ 2 = 296.450, d = 385, p = 1.000) samples. Model E ec s S anda dized pa ame e es ima es o he hypo hesized ela ions among ac o s a e p esen ed in Figu e 1 and as a able in Appendix D (supplemen al ma e ials). O e all, he model accoun ed o 68.8%, 65.3%, and 61.6% o he a iance in in en ion o binge d ink, loss, and adop sa e sun beha io s, espec i ely, and 39.4%, 78.8%, and 27.9% o he a iance in he binge d inking, lossing, and sun-sa e beha io s, espec i ely. Resul s e ealed s a is ically signi ican e ec s o a i udes on in en ions o engage in binge d inking and lossing, bu no o sun sa e y beha io s. The e was a s a is ically signi ican e ec o subjec i e no ms on in en ions o lossing, bu no o he o he beha io s. Pe cei ed beha io con ol signi ican ly p edic ed in en ion o binge d inking and lossing bu no o sun sa e y beha io s. In en ion s a is ically p edic ed lossing, bu no e ec was ound o binge d inking and sun sa e y beha io s. The e was a s a is ically signi ican e ec o pe cei ed beha io al con ol on beha io o sun sa e y beha io s, bu no o he o he beha io s. Au oma ici y a T1 p edic ed au oma ici y 6-weeks la e (T2), and au oma ici y a 12 T2 p edic ed beha io , o each o he h ee beha io s. S a is ically signi ican e ec s we e ound om pas beha io on a i ude, subjec i e no ms, pe cei ed beha io al con ol, in en ions, and au oma ici y a T1 and T2 o all beha io s. The e was a s a is ically signi ican e ec o pas beha io on beha io o binge d inking, bu no o lossing o sun sa e y beha io s. Focusing on he indi ec e ec s, we ound a s a is ically signi ican indi ec e ec o a i udes on beha io media ed by in en ion o lossing beha io , bu no o he indi ec e ec s o he model a iables on beha io media ed by in en ion. We also ound indi ec e ec s o T1 au oma ici y on beha io media ed by T2 au oma ici y o all beha io s. We ound no indi ec e ec s o pas beha io on beha io h ough a i udes, subjec i e no ms, pe cei ed beha io al con ol, and in en ion in each o he models. The e was, howe e , a s a is ically signi ican indi ec e ec o pas beha io on beha io ia T1 au oma ici y, and ia T2 au oma ici y o all beha io s. Mul i-g oup analyses iden i ied a ew di e ences in e ec s ac oss he h ee samples, al hough he di e ences e lec ed he ela i e size o he e ec s ac oss samples, a he han di e ences e lec ing e ec s ha we e di e en om ze o and hose ha we e indis inguishable om ze o. E ec s o a i udes and pe cei ed beha io al con ol on in en ions, and in en ions on beha io , we e signi ican ly smalle in he sun sa e y sample ela i e o he binge d inking and lossing samples. E ec s o pe cei ed beha io al con ol on in en ion and beha io we e signi ican ly la ge in he sun sa e y sample ela i e o he binge d inking and lossing sample. The e ec o pas beha io on a i ude was signi ican ly la ge in he binge d inking sample han he sun sa e y sample. Simila ly, he e ec o pas beha io on subjec i e no ms was signi ican ly g ea e in he binge d inking sample han he lossing sample. The e ec s o pas beha io on T1 and T2 au oma ici y we e signi ican ly g ea e in he lossing sample compa ed o he binge d inking sample. The e ec o pas beha io on 13 in en ion was signi ican ly smalle in he binge d inking and lossing samples ela i e o he sun sa e y sample. Rela i e o he sun sa e y sample, he e ec o pas beha io on beha io was la ge in he binge d inking sample Fo comple ion, we compa ed model e ec s wi hou impu a ion o missing alues, using lis wise dele ion o da a wi h missing cases ins ead. The model es ima ed wi h lis wise- dele ed da a did no esul in subs an i e di e ences in he pa e n o e ec s ac oss he samples. Full esul s o hese analyses a e p esen ed in Appendix D (supplemen al ma e ials). Discussion Social cogni ion heo ies end o ocus on a na ow ange o de e minan s o heal h beha io s. As a consequence, hey may no accoun o e ec s o o he a iables ha could be po en ially in o ma i e when i comes o p edic ing heal h beha io s. One po en ial a iable whose e ec s wi hin social cogni ion models may p o ide impo an in o ma ion on he de e minan s o beha io is pas beha io . The e is al eady a subs an i e body o esea ch examining pas beha io e ec s wi hin social cogni ion heo ies. S udies including equency o pas beha io wi hin hese heo ies ha e consis en ly shown ha pas beha io a enua es e ec s o in en ion and o he social cogni ion cons uc s on subsequen ly measu ed ( u u e) beha io . Howe e , he e is some deba e o e wha such e ec s ep esen . Some esea che s ha e a gued ha pas beha io may model non-conscious and au oma ic p ocesses, such as habi (Ouelle e & Wood, 1998). O he s ha e a gued ha pas beha io e ec s should be media ed by cons uc s om social cogni ion heo ies i he heo y is o be conside ed su icien as an accoun o u he beha io (Ajzen, 1991, 2002). D awing on Ouelle e and Wood’s (1998) p oposi ions, we es ed a se o key hypo heses ela ed o easoned ac ion and au oma ic p ocesses in a social cogni ion model including pas beha io in h ee heal h beha io s, wi h h ee, independen samples: binge d inking in uni e si y s uden s, den al lossing in communi y-dwelling adul s, and sun sa e y beha io s by pa en s o hei 2 o 5 14 yea -old child en. Fu he mo e, we adop ed a comp ehensi e measu e o pas beha io , which encompassed equency, ecency, and ou ine pa e ns o p e ious beha io . Resul s e ealed a consis en pa e n o e ec s o he p oposed model in he binge d inking and den al lossing beha io samples consis en wi h he TPB. These indings sugges ha indi iduals a e mo e likely o engage in hese beha io s when hey ha e posi i e a i udes owa d pe o ming he beha io in u u e and belie e doing so is wi hin hei con ol. Subjec i e no ms did no p edic any o he beha io s, sugges ing a lesse ole o he pe cei ed in luence o signi ican o he s o hese beha io s. Impo an ly, beha io was also p edic ed by au oma ici y in all h ee samples, sugges ing, ha he e y leas , all h ee beha io s we e somewha de e mined by habi s o ou ine. These indings sugges ha cons uc s ep esen ing bo h easoned and au oma ic p ocesses de e mine beha io simul aneously. A possible in e p e a ion o his pa e n o e ec s, p o ided by Hagge e al. (2016), is ha hese beha io s a e de e mined by cons uc s ep esen ing one o he o he o he p ocesses o g oups o pa icipan s wi hin he sample, each wi h su icien s eng h so as o be p esen ed as s a is ically signi ican o e all. The key challenge o u u e esea ch is o iden i y he mode a o a iables ha de e mine when each pa e n o e ec pe ades. In con as wi h p e ious s udies (Hamil on, Ki kpa ick, Reba , & Hagge , 2017), none o he TPB a iables p edic ed pa en s’ in en ions, and in en ions did no p edic beha io o pa en al sun sa e y beha io s. One possible eason o his pa e n o e ec s may be ha hese beha io s a e highly ou inized and habi ual, pa icula ly in an Aus alian con ex whe e exposu e o he sun is bo h likely and egula . A means o es his hypo hesis would be o examine e ec s o he cons uc s ep esen ing he easoned p ocess p edic beha io when e ec s o cons uc s ep esen ing he au oma ic p ocess a e emo ed. We he e o e e-es ima ed he model emo ing e ec s o au oma ici y in he model o his beha io . As p edic ed, esul s e ealed in en ions p edic ed pa en al sun sa e y beha io (β = 15 .227, p < .01). This a enua ion e ec has been obse ed consis en ly in p e ious s udies (see Hagge e al., 2016, 2018), and sugges s au oma ici y is he pe ading de e minan o beha io in his con ex . Resea ch has sugges ed ha pas beha io - u u e beha io ela ions e ec i ely ep esen habi s (Hagge , 2019; Hagge e al., 2016, 2018; Ouelle e & Wood, 1998). We easoned ha i his was he case, esidual e ec s o pas beha io on u u e beha io should be media ed by au oma ici y, o he ex en ha indi idual’s e lec ions on au oma ici y su icien ly cap u e a key componen o habi . Consis en wi h p e ious esea ch ( an B ee e al., 2015), ou esul s e ealed consis en indi ec e ec s o pas beha io on u u e beha io h ough au oma ici y a bo h ime poin s. This esul subs an ia es one o he p oposi ions se by Ouelle e and Wood (1998), ha e ec s o pas beha io model habi ual o au oma ic ac ions. Howe e , con a y o p e ious esea ch (Hagge e al., 2016, 2018) and he p oposals by Ouelle e and Wood, e ec s o pas beha io we e no ound o be media ed by he social cogni ion cons uc s in he cu en model. Ajzen (2002) sugges ed ha he TPB cons uc s should accoun o he e ec s o pas beha io i i is o p o ide a su icien accoun o beha io . He sugges ed ha indi ec e ec s may e lec ha ing made simila decisions in he pas o he e ec o pas expe ience in in o ming belie s ega ding u u e pe o mance o he beha io . Howe e , i seems ha o he cu en se o beha io s, belie s ega ding u u e pa icipa ion jn beha io a e no based on pas expe ience. One possible in e p e a ion is ha a mino i y o indi iduals in hese samples pe haps ha e low le els o p e ious expe ience and, hus, hei belie s owa d pe o ming he beha io s in u u e a e no based on hei pas expe ience. Howe e , o he majo i y, hese beha io s a e likely de e mined by habi s. Fo example, i is well documen ed ha uni e si y s uden s equen ly engage in haza dous binge d inking (Da o en e al., 2015), e en compa ed o hei non-s uden pee s (Kyp i e al., 16 2005). Simila ly, sun sa e y p ac ices is a ele an beha io o mos pa en s in an Aus alian con ex (Hamil on e al., 2016; Hamil on, Ki kpa ick, Reba , Whi e, e al., 2017). I is he e o e likely ha he beha io s a e likely hose ha a e la gely habi ually de e mined. Unlike o he p oposed models ha include delibe a i e and au oma ic pa hways (Caudwell e al., 2019; Hamil on, Ki kpa ick, Reba , & Hagge , 2017; an B ee e al., 2015), a no el aspec o he cu en esea ch is he measu emen o au oma ici y o e wo ime poin s, his may ha e con ibu ed o he ull media ion o pas beha io e ec s by au oma ici y o he den al lossing and pa en al sun sa e y beha io samples. A esidual e ec o pas beha io was s ill obse ed in he binge-d inking sample, which sugges ha au oma ici y may no ully accoun o e ec s o au oma ic, non-conscious beha io s. Fo example, i could sugges ha indi iduals sel - epo o au oma ici y o binge d inking may no be en i ely p ecise because hey do no ake in o accoun o ‘in- he-momen ’ decision making. O he cons uc s ha e lec au oma ic e alua ions o binge d inking beha io such as implici a i udes owa d alcohol, impulsi i y, and implici alcohol iden i y may be u he impo an media o s o pas beha io e ec s (Caudwell & Hagge , 2014; Houben & Wie s, 2009). An inno a i e con ibu ion o he cu en s udy is he use o a second-o de la en a iable o pas beha io ha included long- e m equency o pe o mance, ecen pe o mance (up o 1 mon h), and ou ine pe o mance o he a ge beha io . Typically, pas beha io is measu ed using an a bi a y ime ame, such as 1 week. Such measu es may miss p e ious pa e ns o beha io ha may be impo an in he de e mina ion o u u e beha io . Bagozzi and Wa saw (1990) in hei heo y o ying, and la e Pe ugini and Bagozzi (2001) in hei model o goal-di ec ed beha io s, a gued ha pas beha io should be sepa a ed in o long- e m and ecen componen s. They a gued ha while he wo componen s may be ela ed, hey a e concep ually di e en and he e o e add impo an independen in o ma ion 17 in he p edic ion o beha io . Fo example, an indi idual may ha e only ecen ly aken up an ac i i y (e.g., lossing a e ad ice om hei den is ), o could ha e egula ly engaged in an ac i i y o e a long pe iod, bu has no been able o ecen ly (e.g., an indi idual who usually binge d inks each weekend bu has ecen ly educed spending o sa e o hei uni e si y ex - books). Pe ugini and Bagozzi (2001) p oposed ha ecency o beha io al engagemen may in luence u u e beha io by ancho ing biases ha may ca y implici in o ma ion abou in en ions o a deg ee highe han by wha is consciously a ailable. Fu he mo e, consis en wi h Ouelle e and Wood’s (1998) p emise, hey sugges ha long- e m equency o beha io al engagemen likely maps o habi ual occu ences o he beha io . Despi e hese p emises, we ound ha hese concep ually dis inc componen s o pas beha io we e highly co ela ed and se ed as indica o s o a second-o de pas beha io la en a iable. These indings sugges ha all h ee componen s a e cap u ed by an o e all pas beha io cons uc . Measu es o he di e en pas beha io componen s end o con e ge in la ge samples, and sepa a ion o he di e en componen s do no o e addi ional in o ma ion in e ms o he p edic ion o u u e beha io . In ac , sepa a ion o he di e en componen s would likely con ound analyses due o mul i-collinea i y due o he high in e co ela ions. Ne e heless, jus i ica ion o sepa a ing pas beha io in o sepa a e componen s may be jus i ied when s udying pas beha io e ec s o in equen ly pe o med beha io s, like blood dona ion, o new beha io s ha ha e only jus been ini ia ed. The pe asi e e ec s o habi on he heal h beha io s in he cu en s udy ha e impo an implica ions o p ac i ione s and clinicians. Findings sugges ha once an indi idual has adop ed a heal h-beha io , such as applying sun-sa e p ac ices o hei child en o lossing, hey should ocus hei e o s on building beha io al au oma ici y o main ain he heal h-beha io . S a egies ha p omo e habi s such as inc easing beha io al pe o mance in he p esence o a cue could he e o e be adop ed. Fu he mo e, p e ious esea ch has 18 demons a ed ha habi o ma ion ollows an asymp o ic cu e; au oma ici y g ows as es in he ini ial weeks o habi o ma ion, and hen pla eaus (Lally e al., 2010). This pa e n may mean ha while i can ake se e al mon hs o a habi o ully o m (Lally e al., 2010), he mo e in ensi e wo k o build he habi (e.g., ocusing on building a esponse o a s able and applicable cue) can be pe o med in he ini ial ew weeks o an in e en ion be o e su icien sel - egula o y ac ions (e.g., de elopmen o sel -moni o ing cha s) could enable pe sis en epe i ion o he habi ual ac ion. In addi ion, in e en ions may need o ocus on building he awa eness o when an unheal hy habi is being cued. The ocus o in e en ions may be o b eak o swap a habi ual esponse, o example, i an indi idual egula ly begins binge- d inking a e wo k on a F iday, hey may change hei esponse o he cue (i.e., ‘ inishing wo k on a F iday) wi h some hing mo e heal hy (e.g., ea ing a a es au an ha does no se e alcohol o by engaging in physical exe cise). This wo k equi es e o ul, conscious e lec ion and decisions, and he e o e u u e in e en ions may also need o use echniques ha ocus on he social–cogni i e aspec o he cu en model (i.e., ocus on building pe cei ed beha io al con ol and posi i e a i udes o change). S eng hs, Limi a ions, and Resea ch Di ec ions The cu en s udy has a numbe o s eng hs, including a comp ehensi e es o a model ha included cons uc s ep esen ing easoned and au oma ic p ocesses in h ee dis inc beha io s: den al lossing, binge d inking, and pa en s’ sun sa e y beha io s o hei child en. Impo an ly, his is he i s s udy, o he au ho s’ knowledge, o use a measu e o h ee dis inc componen s o pas -beha io : equency, ecency, and ou ine. The esea ch is also inno a i e as i explo es he ex en o which e ec s o he pas beha io ac o on u u e beha io we e accoun ed o by measu es o beha io al au oma ici y, a key componen o habi . Also unique was he measu emen o au oma ici y a wo poin s in ime, which akes he empo al s abili y o his cons uc in o accoun . 19 Howe e , his s udy is no wi hou limi a ions, which we ou line nex along wi h di ec ions o u u e esea ch. The cu en s udy used a p ospec i e co ela ional design, so he di ec ion o ela ions can only be in e ed om he p oposed ela ionships ou lined in he ele an heo ies. C oss-lagged panel and expe imen al designs a e needed o con i m he di ec ion o causali y (Chan e al., 2020; Liska, 1984). The cu en s udy p ima ily ec ui ed ela i ely homogenous samples o pa icipan s ha was low on di e si y, which places limi s on he gene alizabili y o he esul s (Hen ich e al., 2010). Ano he limi a ion is he eliance on sel - epo measu es o beha io . Gi en one o he beha io s could be socially undesi able (i.e., binge d inking), and ano he could be seen as an e alua ion o posi i e pa en ing p ac ices (i.e., he use o sun sa e y beha io s on you child), social desi abili y e ec s ha e unduly in la ed epo s o hese beha io s. Fu u e s udies should ocus on collec ing beha io al da a ha does no ely on sel - epo s which may ob ia e hese biases (Bulle & Bo land, 1999). In addi ion, he pa en al sun sa e y beha io was de ined as a collec ion o beha io s, which may ha e had unin ended consequences in he way some pa icipan s e lec ed on he au oma ici y o he beha io . Fo example, i is plausible ha di e en sun sa e y beha io s may a y in hei deg ee o au oma ici y, which may ha e made i p oblema ic o pa icipan s o espond o he measu e. Howe e , gi en each o hese beha io s a e likely cued in he same way (e.g., going ou side du ing he day) and ha he collec ion o beha io s ha e been ad oca ed o occu oge he in a long-s anding public heal h campaign in Aus alia (Mon ague e al., 2001), i is likely ha each has a simila le el o au oma ici y. Las , au oma ic p ocesses in he cu en model we e ep esen ed by sel - epo ed au oma ici y alone. Fu u e esea ch should conside including o he measu es ha ap in o hese p ocesses, such as coun e -in en ional habi s (Ga dne e al., 2015) and implici belie s (Hagge & Cha zisa an is, 2014; S ack & Deu sch, 2004). These addi ional cons uc s may play an impo an ole in accoun ing o e ec s o pas beha io (Hagge e al., 2017). 20 Conclusions The cu en s udy es ed a social cogni ion model ha encompassed cons uc s ep esen ing easoned ac ion and au oma ic p ocesses o p edic h ee heal h beha io s in h ee sepa a e samples: binge d inking in uni e si y s uden s, den al lossing in communi y- dwelling adul s, and pa en al sun sa e y beha io s o hei 2 o 5 yea -old child en. Resul s indica ed ha cons uc s ep esen ing he easoned ac ion and au oma ic p ocesses signi ican ly p edic ed lossing, whe eas binge d inking and sun sa e y beha io s we e gene ally p edic ed by cons uc s ep esen ing au oma ic beha io s. The cu en in es iga ion also ound suppo o he media ion o he pas beha io - u u e beha io ela ionship by au oma ici y. The cu en s udy ills a knowledge gap in he cu en li e a u e on he mul iple p ocesses ha guide beha io and p o ide u he e idence ha cons uc s ha ep esen au oma ic p ocesses play a key ole. Fu u e esea ch should ocus on explo ing he ole o o he cons uc s ha ep esen au oma ic p ocesses such as coun e -in en ional habi s and implici belie s. 27 Ve planken, B. (2006). Beyond equency: Habi as men al cons uc . B i ish Jou nal o Social Psychology, 45(3), 639–656. h ps://doi.o g/10.1348/014466605X49122 Ve planken, B., & O bell, S. (2003). Re lec ions on pas beha io : A sel ‐ epo index o habi s eng h. Jou nal o Applied Social Psychology, 33(6), 1313–1330. h ps://doi.o g/10.1111/j.1559-1816.2003. b01951.x Xiang, F., Ha ison, S., Nowak, M., Kimlin, M., Van de Mei, I., Neale, R. E., Sinclai , C., & Lucas, R. M. (2015). Weekend pe sonal ul a iole adia ion exposu e in ou ci ies in Aus alia: In luence o empe a u e, humidi y and ambien ul a iole adia ion. Jou nal o Pho ochemis y and Pho obiology B: Biology, 143, 74–81. h ps://doi.o g/10.1016/j.jpho obiol.2014.12.029 S udy Highligh s • Cons uc s ep esen ing easoned ac ion and au oma ic p ocesses p edic ed heal h beha io s • Cons uc s ep esen ing au oma ic p ocesses media ed he pas - u u e beha io ela ionship • The s udy used a no el mul i-componen measu e o pas beha io C edi au ho s a emen Daniel J. B own: concep ualiza ion, me hodology, da a colle ion, o mal analysis, w i ing – o iginal d a and edi s. Ma in S. Hagge : concep ualiza ion, me hodology, w i ing – edi s, supe ision. Ky a Hamil on: concep ualiza ion, me hodology, w i ing – edi s, supe ision.