Chpter 2 Socil Indictors Reserch nd Helth-Relted Qulity o Lie Reserch Alex C. Michlos Introduction Reserch relted to qulity-o-lie ought to beneit rom eorts o reserchers trined in diverse disciplines, ddressing more or less similr problems, with more or less similr ims. One o the erly, gret philosophers o science o the twentieth century, Rudol Crnp, sid tht wht surprised nd distressed him most bout his work in philosophy o physics ws tht while he thought his reserch bridging two disciplines would be welcomed by both sides, in ct it ws unwelcomed (Schilpp 1963). Both sides thought his work ws neither ish nor owl, nd it ws only when the ield o philosophy o science reched something like disciplinry sttus, with its own journls, recognized uthorities, clssic texts, prdigms, stndrd courses nd so on, tht his work received the ttention nd pprobtion it deserved. I do not think socil indictors reserch nd helth-relted qulity o lie reserch constitute distinct disciplines, but there re reltively independent reserch groups devoted to the rnge o issues in ech ield. In this pper I try to build something like modest ootbridge between the two ields in order to mke it esier or reserchers in both ields to drw on the body o knowledge ccumulted in ech ield. I hope I hve more success thn Crnp, but whether or not I do, I m hppy This essy ws written or presenttion t n invited plenry session o the nnul meeting o the Interntionl Society or Qulity o Lie Reserch held in Amsterdm, 7 10 November 2001. The originl pln ws to give specil ttention to reserch on hppiness, but the ocus becme broder s the story unolded. Michlos, A.C.: 2004. Socil indictors reserch nd helth-relted qulity o lie reserch. Socil Indictors Reserch, 65, pp. 27 72. 2004 Kluwer Acdemic Publishers. Printed in the Netherlnds. A.C. Michlos (&) University o Northern British Columbi, Prince George, BC, Cnd e-mil: michlos@brndonu.c Springer Interntionl Publishing AG 2017 A.C. Michlos, Connecting the Qulity o Lie Theory to Helth, Well-Being nd Eduction, DOI 10.1007/978-3-319-51161-0_2 25
26 2 Socil Indictors Reserch nd Helth-Relted to report tht my own understnding o the reserch undertken on ech side o the bridge hs been immensely enriched by this investigtion. So, I m grteul to the orgnizers o this conerence or the invittion to write this pper. The structure o the essy is s ollows. In the next section Socil Indictors o the Qulity o Lie: Bsic Deinitions I briely review some bsic deinitions concerning socil indictors o the qulity o lie. Ater tht, there is section Historicl Origins nd Motives reviewing some o the historicl origins nd motives behind socil indictors reserch. I ssume, with some evidence, tht mesures o hppiness provide t lest one importnt wy to opertionlize the vriety o ides reerred to by the phrse qulity o lie. In section Explining Hppiness with Lie s Whole rom Selected Domin Stisction Items I review results o eleven surveys undertken to explin hppiness on the bsis o the personl stisction respondents got rom dozen speciic domins o their lives, e.g., stisction with their jobs, their living prtners, their inncil security nd their helth. Following tht, in section Explining Hppiness with Lie s Whole rom Helth nd Other Vribles I review results o three surveys undertken to explin hppiness on the bsis o respondents helth mesured by such things s SF-36, chronic illnesses nd Body Mss Index scores. Then there re two sections in which I review studies designed to explin stisction with one s helth on the bsis o mesured helth sttus (section Explining Stisction with One s Own Helth ) nd on the bsis o Multiple Discrepncies Theory (MDT, section Multiple Discrepncies Theory (MDT) ). While the mteril in the sixth section is bsed on single study, tht in the seventh section is bsed on 58 studies o undergrdute student smples tken rom 83 universities in 39 countries, involving over 18000 students. In section Helth-Relted Qulity o Lie I review nerly two dozen studies in the trdition o helth-relted qulity o lie in order to illustrte the serious conounding nd question-begging tht one encounters with explined nd explntory vribles in this trdition. Finlly, there is short overview nd concluding section Conclusion. Socil Indictors o the Qulity o Lie: Bsic Deinitions Since the middle o the 1960s there hve been people working ll over the world in wht mny o us used to cll the socil indictors reserch movement. The term socil indictor is used to denote socil sttistic tht is supposed to hve some signiicnce or the qulity o lie, nd the term socil report designtes n orgnized collection o socil indictors. From linguistic point o view, socil indictors usully consist o term denoting subject clss nd term denoting wht I cll n indictor property. For exmple, the second term o the phrse innt mortlity denotes the indictor property mortlity nd the irst term denotes prticulr clss o things, nmely, innts tht my possess tht property. By replcing the subject term innt by one yer old, two yer old, etc., we cn routinely generte (socil) mortlity indictors or s mny ge groups s we like.
Socil Indictors o the Qulity o Lie: Bsic Deinitions 27 Similrly, by replcing the subject term by mle, Indin, etc., we cn routinely generte mortlity indictors or s mny kinds o groups s we like. Socil indictor phrses re like vrible nmes in logic nd mthemtics, nd socil indictors re like the vribles themselves. Furthermore, just s we spek o the vlues o vribles in logic nd mth, we my spek o the indictor-vlues o socil indictors. For exmple, humn body weight is (socil) helth indictor which my hve n indictor-vlue o 120 lb. Finlly, nd most importntly, strictly speking it is the indictor-vlue o socil indictor which indictes whtever it is tht prticulr indictor is supposed to indicte. For exmple, i the indictor-vlue o one s body weight is 600 lb then this indictes tht one is probbly obese (or very, very tll), ets more crbohydrtes, strches nd ts thn most people, shops t specil clothing stores, suers tigue nd shortness o breth rom little exertion, cnnot sit comortbly in thetre set, nd my hve diiculty inding suitble mte. Usully, the reltionship between n indictor-vlue nd wht it indictes is one-mny reltionship, s my exmple indictes. Socil indictors tht reer to personl eelings, ttitudes, preerences, opinions, judgments or belies o some sort re clled subjective indictors, e.g., stisction with one s helth, ttitudes towrd exercise, belies bout the beneits o some therpy. Socil indictors tht reer to things tht re reltively esily observble nd mesurble re clled objective indictors, e.g., the height nd weight o people, numbers o people dmitted to hospitl, numbers o people dying per yer. Positive indictors re such tht most people will ssume tht i their indictor-vlues increse, some cet o the qulity o lie is improving, e.g., elderly citizens incomes nd minority-group eductionl ttinment. Negtive indictors re such tht most people will ssume tht i their indictor-vlues increse, some cet o the qulity o lie is deteriorting, e.g., innt mortlity rtes nd murder rtes. (Notice tht n indictor is here regrded s positive or negtive not in virtue o whether or not its vlues in ct increse or decrese, but only in virtue o whether or not most people would like its vlues to increse or decrese, i.e., not in virtue o the ct but in virtue o the desirbility o n increse or decrese in its vlues.) Uncler indictors re such tht either () most people will not be willing or ble to sy whether bigger indictor-vlues indicte better or worse stte o irs, e.g., welre pyments, or (b) there is serious disgreement bout whether bigger indictor-vlues indicte better or worse stte o irs, e.g., divorce rtes. In the cse o welre pyments, we do not know wht to sy becuse s the vlues increse there my be n increse o people in need o such ssistnce, which is bd; while, t the sme time, there is n increse in the mount o ssistnce given, which is good. In the cse o divorce rtes, on the other hnd, mny people know exctly wht they wnt to sy, nd they hppen to disgree with wht some other people wnt to sy. Input indictors indicte some sort o inputs into process or product, e.g., dily exercise time. Output indictors indicte some sort o output o process or product, e.g., Body Mss Index. Unlike the previous clssiictions o indictors, wht counts s n input or output indictor depends on the purposes o the clssiiction. For exmple, rom the point o view o therpist or coch, person s dily exercise time could be regrded s the output resulting rom dvice, dmonitions
28 2 Socil Indictors Reserch nd Helth-Relted Tble 2.1 Types o socil indictors or helth domin Positive (P) Negtive (N) Input (I) Objective (O) IOP (e.g., dily exercise time) ION (e.g., cigrettes smoked per dy) Subjective (S) ISP (e.g., positive ttitude towrd exercise) ISN (e.g., positive ttitude towrd smoking) Output (U) Objective (O) UOP (e.g., norml BMI score) UON (e.g., stined ingers nd teeth) Subjective (S) USP (e.g., stisction with weight) USN (e.g., disstisction with stines) nd threts given to ptient or student. Similrly, person s BMI could be regrded s n input indictor or therpist or coch trying to decide wht sort o exercise to prescribe. All o these kinds o indictors re summrized in Tble 2.1, with exmples o ech. (Michlos 1988 shows how these distinctions could be used to systemticlly expnd UNESCO s nnul interntionl reports.) When people use the phrse qulity o lie, they sometimes intend to contrst it with quntities or numbers o something. There re, then, two quite dierent sorts o things tht one might wnt to reer to when using the phrse qulity o lie. In the irst plce, one might wnt to reer to sorts, types or kinds o things rther thn to mere numbers o things. For exmple, one might wnt to know not merely how mny people were dmitted to hospitl lst yer nd or how long, but lso wht sorts o people they were, mle or emle, young or old, rich or poor, nd so on. When the term qulity in the phrse qulity o lie is used in this sense, one my sy tht it nd the phrse in which it occurs is intended to be primrily descriptive. In the second plce, however, one might wnt to reer to the vlue or worth o things by using the term qulity in the phrse qulity o lie. For exmple, one requently hers o people mking trde-o between, sy, high slry on the one hnd nd better working conditions on the other. Presumbly the exchnge here involves monetry nd some other vlue. Tht is, one exchnges the vlue o certin mount o money or the vlue o certin set o working conditions. When the term qulity in the phrse qulity o lie is used in this sense, one my sy tht it nd the phrse in which it occurs is intended to be primrily evlutive. (Dierent theories bout vlues re reviewed in Michlos (1980, 1981), nd ggregtion problems re reviewed in Michlos (1970, b, c, d, 1997, 2001).) Both senses o the phrse qulity o lie re importnt. It is importnt to be ble to describe humn existence in irly relible nd vlid shion, nd it is importnt to be ble to evlute humn existence in the sme wy. In philosophic jrgon, one would sy tht complete ccount o the qulity o lie requires thorough exmintion o its nture or being s well s n exmintion o its vlue or good, i.e., n ontologicl ccount nswering the question Wht is its nture? nd n xiologicl ccount nswering the question Wht is its vlue?. I suppose it is
Socil Indictors o the Qulity o Lie: Bsic Deinitions 29 trivilly true (tutologicl) tht the improvement o one s reltively objective nd subjective conditions implies n improvement in the qulity o one s lie in n evlutive sense. From consequentilist morl point o view, our im should be to improve or increse the vlue or worth o the lives o ll people (Michlos 1995). Thus, the generl line o my pproch is in the nturlistic subjectivist trdition o the Americn prgmtists, especilly Jmes (1909), Perry (1926, 1954), Dewey (1939) nd Lewis (1946). Since the trdition hs been round or long time, there is no need to review its strengths nd weknesses. From this point o view, the qulity o person s or community s lie, in n evlutive sense, is directly proportionte to the net levels o stisction generted by tht lie or everyone ected by it. As Jmes put it in n essy irst published in 1891, Tht ct must be the best ct which mkes or the best whole, in the sense o wkening the lest sum o disstisctions. In the csuistic scle, thereore, those idels must be written highest which previl t the lest cost, or by whose reliztion the lest possible number o other idels re destroyed. The course o history is nothing but the story o men s struggles rom genertion to genertion to ind the more nd more inclusive order (Jmes 1977, p. 623). When reserchers mesure person s overll qulity o lie in terms o his or her reported hppiness or stisction, they re most likely ssuming tht people re the best judges o the vlue or worth o their own lives nd tht s their reported levels o hppiness or stisction increse or decrese, the qulity o their lives (its vlue or worth) is incresing or decresing. I ll other things re equl, it is resonble nd morlly right to give personl reports some privileged sttus in ssessing the qulity o person s lie (e.g., Nord et l. 2001). Nevertheless, one hrdly needs scientiic reserch to relize tht ll other things re oten not equl, lthough there is plenty o reserch to prove it (Michlos 1990, 1991). Thus, giving people the irst word to ssess the overll qulity o their own lives does not imply tht ech person s word is lwys the only, best or inl word, or tht such irst words re in principle incorrigible (Lewis 1946; Brybrooke 1967). It is quite possible tht some externl observers might ssess the vlue or worth o some person s lie with dierent results bsed on dierent nd even more resonble nd morlly cceptble methods, ll things considered. Essentilly, tht is why loyl gency to ptient or ny sort o client cnnot ber the weight o serving s supreme morl principle (Michlos 1979; Brunk 1987; Po1987). In his brod pproch to qulity o lie mesurement, Hornquist (1989) explicitly, systemticlly nd I think quite properly llows or such circumstnces. Apriori one cnnot sy or ll people nd or ll time wht is the single best method or ssessing the overll qulity o everyone s lie, ll things considered. As I explined in Michlos (1978, 1992, 1995), one cn not even give rigorous ccount o wht it would men to consider ll things. Thus, the best one cn do is try to ssemble resonble proile o the objective nd subjective etures o people s lives, nd to evlute the proile in resonble nd morlly cceptble wy. In short, in the end, we muddle our wy through to something we think is worthy o being clled good or bd qulity o lie (Michlos 2001).
30 2 Socil Indictors Reserch nd Helth-Relted Historicl Origins nd Motives Since good histories o the recent origins o the socil indictors movement hve been published by Michlos (1980, 1999), Lnd (1983), McRe (1985), Ferris (1990), Noll nd Zp (1994), Vogel (1994) nd Sirgy (in press), there is no need or n extensive review here. By the middle 1950s nd 1960s there ws lurry o ctivity in ll industrilized countries nd in interntionl orgniztions. It ws period in which governments were committed to ctivism. Government sponsored socil progrms blossomed cross the globe, bringing with them the need to monitor results in order to evlute the progrms. It ws reminiscent o the 1930s when everyone struggled to pull the world out o recession nd plnning ssocitions sprung up t ll levels o government (Krl 1974; President s Reserch Committee on Socil Trends 1933; Legue or Socil Reconstruction Reserch Committee 1935). The longer historicl trdition o socil indictors reserchers my be trced bck irly directly to Jeremy Benthm s ides bout eliciic clculus tht would llow decision mkers to clculte the net plesure or pin connected to every ction or everyone ected by tht ction, with public policy choices mde to get the gretest net plesure or lest net pin or the gretest number o people (Benthm 1789). For economists, utility theory is the pprent heir o Benthm s clculus (Mitchell 1918). Utility theory is ormlly elegnt nd hs been n enormously ruitul source o reserch progrms in individul nd group decision mking relted to commercil mrkets, socil nd politicl reltions, brgining, conlict resolution, gming nd scrce resource lloction in prcticlly ll res, including helth (Nussbum nd Sen 1993; Torrnce 1986). It is the hert o micro economics. Unortuntely, utility theory (s mny other theories) is much better on pper nd in clssroom exercises thn it is in prctice, especilly in public policy mking. Wht is worse, utility theory begins with reveled preerences which re the mere tips o socilly, psychologiclly nd prgmticlly constructed icebergs o more or less coherent systems o knowledge, opinions, ttitudes, desires nd needs (Michlos 1967, 1978, 1985; Schwrz nd Strck 1999) While economists nd other utility theorists see the iceberg tips nd sk How cn we use them?, psychologists nd sociologists see them nd sk Wht is their source? nd Is it resonble to use them?. More oten thn not, socil indictors reserchers hve been on the side o psychologists nd sociologists. At lest since the mid 1930s, sttistics rom Ntionl Income nd Product Accounts hve been routinely used or public policy mking in totl disregrd o ny microeconomic utility nlysis. Thus, mny o us (socil indictors reserchers) thought tht wht ws required or public policy mking ws set o socil nd lter environmentl indictors tht could be combined with the Ntionl Accounts to crete System o Socil, Economic nd Environmentl Accounts. In most industrilized countries tody, government unded reserch on this project is continuing. I believe one o the motivting orces tht most socil indictors reserchers hve in common is this vision o grnd ccounting system with
Historicl Origins nd Motives 31 plce or everything importnt nd everything importnt in its plce (Juster nd Lnd 1981; Fox 1985). A Tsk Force o the Interntionl Society or Qulity o Lie Studies recently published long pper in which some dequcy criteri or such grnd system were presented, long with some nlyses o some contemporry constructions (Hgerty et l. 2001). Needless to sy, we re r rom ny generlly ccepted system. Some reserchers in the helth-relted qulity o lie trdition seem to envision comprehensive systems tht re similr to those envisioned by socil indictors reserchers, e.g., Hörnquist (1989), Renwick nd Brown (1996). Given most socil indictors reserchers shred vision o grnd ccounting scheme, it will come s no surprise tht most o us lso shre love or crunching numbers. While most o us grnt the importnce o qulittive specultive nd empiricl work, we re primrily interested in ttching numbers to things nd to the vlue o things, nd using quntittive inormtion to solve public policy problems. It will lso come s no surprise tht most socil indictors reserchers re mles nd tht, silly s it is, mny o us seem to hve the ttitude tht our numbers cn bet their numbers. Our numbers re the numbers produced by sociologists, psychologists nd environmentlists, while their numbers re the ones produced by economists. Admitting some oversimpliiction, most socil indictors reserchers think tht it is mistke or people interested in the qulity o lie to limit their ocus o ttention to commercil mrket ctivities nd to evlute everything countble by the single mesure o money. We would judge it eqully mistken to limit one s ttention to humn helth. Put positively, we think the qulity o people s lives cn be mesured nd compred directly using indictors o such things s lie expectncy ree o disbility, eductionl ttinment, crime rtes, litercy rtes, resource depletion rtes, ir pollution rtes, stress t work, stisction with mily reltions, sel-esteem nd so on. Most likely, we suppose tht i grnd ccounting scheme is ever designed, it will involve some sort o n input-output mtrix o proile o socil, economic nd environmentl indictors tht collectively cpture the qulity o humn existence in descriptive nd evlutive senses. Although in much o this pper it is ssumed tht, rom the point o view o person s personl experience or eelings, the qulity o person s lie my be mesured by reported hppiness, there is in ct much more to lie nd its vried qulities thn hppiness. While hppiness my be one o the best globl indictors o the qulity o lie, it is certinly not the only one (Veenhoven nd Ouweneel 1995; Diener nd Lucs 1999, 2000). Severl reserchers working in the re o helth-relted qulity o lie regrd helth s n spect or domin o the broder ield o overll qulity o lie. For exmple, members o the Qulity o Lie Reserch Unit t the Centre or Helth Promotion (CHP) o the University o Toronto regrd the qulity o peoples lives s multidimensionl, touching on their being, belonging nd becoming (Renwick nd Brown 1996) in wys tht re reminiscent o Allrdt s (1975) hving, loving nd being. Fletcher et l. (1987, p. 557) were cler tht The concept o qulity o lie encompsses mny components: culturl nd politicl vlues s well s more tngible mesures such s living nd eductionl stndrds nd reedom rom disese. But they limit their explortions to those dimensions o qulity o lie
32 2 Socil Indictors Reserch nd Helth-Relted which re ected by disese nd its tretment, nd which hve the potentil to be chnged by the therpeutic sitution. Njmn nd Levine (1981) took similr line in their review o the impct o medicl technologies on the qulity o ptients lives. Explining Hppiness with Lie s Whole rom Selected Domin Stisction Items I suppose the ides designted by the term hppiness or their vrious synonyms in dierent lnguges re the oldest cndidtes s globl indictors o n overll good qulity o lie. It is necessry to use the plurl ides becuse dierent people hve hd very dierent ides bout the nture o hppiness or o n overll good qulity o lie (Ttrkiewicz 1976). Virtully ll reserch hs shown tht hppiness nd lie stisction shre some common mening, e.g., see Ttrkiewicz (1976), Veenhoven (1984), Argyle (1987), Michlos (1991), Michlos nd Zumbo (1999, 2001). When people tlk bout stisction or hppiness with their whole lives, they re typiclly reerring to reltively lsting, justiied, good eeling nd ttitude bout their lives. So, theory o lie stisction could be interpreted s theory o hppiness, nd in generl such theories would be theories o subjective well-being. However, becuse mesures o hppiness nd lie stisction do not hve identicl connottions or denottions, it is worthwhile to nlyze their components nd correltes seprtely. Tble 2.2 provides some comprtive igures rom eleven surveys tken over 20 yers resulting rom regressing men hppiness scores on men scores or stisction with vrious domins o lie. The simple liner model used to obtin these igures ws one o the irst models pplied by socil indictors reserchers nd it is still one o the most requently used models (Michlos 1991). It is bsed on the ide tht people s overll or globl levels o hppiness re the result o some sort o combintion o the stisction tht they get rom dierent spects o their lives. The model llows one to precisely mesure the impct o stisction with, or exmples, people s mily reltions, helth, housing nd jobs on their perceived overll hppiness with lie. Although there re plenty o indexes o hppiness nd subjective well-being generlly, in most o my work I use vrints o the single item, 7-point mesures tht were thoroughly investigted by Andrews nd Withey (1976), nd especilly by Andrews (1984). In the ltter pper, Andrews showed tht typicl survey item [e.g., 7-point mesures], when dministered by respected survey orgniztion to generl popultion smple, cn be expected to yield 50 83% vlid vrince, 0 7% method eects vrince, nd 14 48% residul vrince over two-thirds o the vrition in mesurement qulity could be explined by 13 survey design chrcteristics (p. 409). Although McDowell nd Prught (1982, pp. 949, 958) recommended the Brdburn Aect Blnce Scle (Brdburn nd Cplovitz 1965) s generl mesure o well-being or hppiness in popultion
Explining Hppiness with Lie s Whole 33 Tble 2.2 Comprison o hppiness regressions rom 11 surveys Percent o vrince explined in hppiness Predictors 1979 1981 1982 1984 1985/86 1997 1997 1999n 1999s 2000 2000 A B C D E F G H I J K 45 32 36 39 28 38 27 45 47 35 46 Stisction with: Stndrdized regression coeicients Helth 0.12 0.12 0.18 0.17 0.09 b b 0.10 Finncil security 0.09 0.06 0.21 b 0.13 0.14 0.19 0.09 b Fmily reltions 0.38 0.03 0.09 0.14 0.06 0.14 b b b Job 0.03 c 0.09 0.18 b b b 0.14 0.20 Friendships 0.23 0.23 0.01 0.21 0.10 0.09 b b 0.21 Housing 0.01 0.01 0.10 b 0.07 b b b b 0.19 0.06 0.14 b b 0.16 b b b b b b b b b c b b c c b b Are lived in 0.01 0.05 Recretion ctivity 0.03 0.04 0.05 b 0.07 b b b b 0.14 b Religion c 0.01 0.03 b b b b b b 0.15 b Sel-esteem 0.07 0.09 0.14 0.26 0.13 0.27 0.25 0.38 0.29 0.23 0.32 Trnsporttion 0.05 0.05 b b b b c c c c c Gov. services c 0.08 0.03 c c 0.09 b c c b c Living prtner c 0.30 c c 0.18 0.22 0.22 0.13 0.17 b 0.24 Eduction 0.3 c c b 0.10 c c c c b b Fed. Gov. O. c c c c c c c 0.10 b b b Personl sety c c c c c c c 0.08 0.16 b b P < 0.01 A Clericl st (N = 312), B Rurl seniors (N = 273), C Cochrne residents (N = 328), D Guelph students (N = 296), E World students (N = 5036), F Prince George residents (N = 466). G Jsper residents (N = 210), H B.C. northerners (N = 713), I B.C. southerners (N = 348). J Quesnel (N = 370), K Prince George residents Nov. 2000 b Signiicnce level too low to enter eqution c Not in questionnire. Sources o the surveys re identiied in the Reerences using the letters heding ech column; e.g., A = Michlos (1980)
34 2 Socil Indictors Reserch nd Helth-Relted surveys, their exmintion o tht scle nd single item hppiness mesure reveled tht The ltter perormed remrkbly well on mny o [their] nlyses nd probbly provides the best single pproch to mesuring well-being. Unortuntely, other reserchers hve gone long with the recommendtion insted o with the most likely best single pproch, e.g., Dibetes Control nd Complictions Tril (1988). Serious problems with the Brdburn scle were reported in Sndvik et l. (1993). Good cses or sel-reported, single-item, globl indictors o helth sttus were mde by Idler nd Benymini (1997), nd Verbrugge et l. (1999). On verge, or the eleven smples represented in Tble 2.5, we were ble to explin 38% o the vrince in reported hppiness rom some subset o the 16 predictor vribles listed in the exhibit. Our best success cme rom the smple o southern British Columbins in 1999 (column I), t 47%. A dozen o the 16 domin vribles were drwn rom Andrews s (1974) extensive reserch. When ll vribles re stndrdized to hve mens o zero nd stndrd devitions o one, the stndrdized regression coeicients (Bets) mesure the percent o movement in the dependent vrible when predictor vrible moves one ull unit nd every other predictor in the set is held constnt. For exmple, under column A one inds tht the Bet vlue or stisction with one s mily reltions is b = 0.38, which mens tht or every ull stndrd devition increse in stisction with one s mily reltions, one s hppiness increses bout 38% o stndrd devition. Figurtively speking, this mens tht or every ull step increse in how good one eels bout one s mily reltions, one s overll hppiness goes up bout 38% o step. Inspection o the other igures in tht column revels tht stisction in no other domin hd s gret n impct on overll hppiness or the 312 members o the smpled clericl st. For our purposes, it is most importnt to notice tht stisction with one s own helth ws never the strongest predictor o hppiness in ny smple. In ive o the eleven smples, stisction with one s own helth iled to enter the inl explntory regression eqution or lck o sttisticl signiicnce. In nother three smples, it iled to pper s one o the top three predictors, it rn third in two o the three smples in which it did pper in the top three nd in the remining smple it rn only second. Clerly, the old dge bout hving everything when you hve your helth is lse or the people in these eleven smples. Reding cross the row to the right rom 0.38 in column A, one sees tht stisction with one s mily reltions ws not the strongest predictor o hppiness or ny other smple. Stisction with one s own sel-esteem ws the strongest predictor in seven smples, nd by tht ct lone good cndidte or the most importnt domin in the set o 16. More oten thn not, the better one eels bout one s own sel, the hppier one will be. The igures just reviewed clerly illustrte the ct tht dierent groups o people with dierent lie circumstnces, resources nd constrints use dierent mixtures o ingredients to determine their hppiness. The explntory model used here llowed the groups represented in Tble 2.2 to drw stisction rom roughly the sme set o domins. Anlyticlly speking, some groups used more nd some used ewer o these domins to build their hppiness. The Jsper group used only 3 o the 12
Explining Hppiness with Lie s Whole 35 possibilities open to them while the rurl seniors used ll 12 o 12 open to them. There is no evidence tht people consciously selected more or ewer domins to crt their hppiness, but dierent people did in ct use dierent building blocks. Supplementing these indings, in severl provoctive studies Schwrz nd collegues hve demonstrted tht the order in which domin nd globl items re presented in questionnires, s well s the number o competing domins presented cn hve proound eects on the mesured reltionships mong ll the vribles; e.g., Schwrz nd Strck (1991, 1999), Schwrz nd Bless (1992). Concluding ine review rticle, Schwrz nd Strck (1999, p. 79) wrote, reports o well-being re subject to number o trnsient inluences. Like other socil judgments, they re best considered constructions in response to prticulr questions posed t prticulr time. Although the inormtion tht respondents drw on relects the relity in which they live, which spects o this relity they consider nd how they use these spects in orming judgment is prooundly inluenced by etures o the reserch instruments. Explining Hppiness with Lie s Whole rom Helth nd Other Vribles An stute observer would sy tht while stisction with one s helth my be wek predictor o overll hppiness, helth itsel should be strong predictor. Tbles 2.3, 2.4 nd 2.5 show the results o our exmintion o this suggestion using three dierent smples nd dierent sets o potentil predictors. In ech cse, sel-reported helth ws mesured by the ner-gold stndrd eight dimensions o the Medicl Outcomes Study Short Form (SF-36) (Wre et l. 1993), lthough in the third cse dditionl helth mesures were lso used. I will discuss ech cse in turn. Tble 2.3 summrizes results o survey o 875 dults ged 55 to 95 living in the Northern Interior Helth Region o British Columbi in September 1999 (Michlos et l. 2001). Our interests lie in the irst nd lst numericl columns. The irst numericl column shows tht 2 o the 8 SF-36 helth dimensions explined 34% o the vrition in hppiness scores. The dimension o Mentl Helth (b = 0.49) ws over three times s inluentil s tht o Socil Functioning (b = 0.14), nd the other 6 dimensions o SF-36 were not sttisticlly signiicntly ssocited with hppiness. As one moves cross the columns rom let to right, results o regressing hppiness scores on other things re listed. Other things include, or exmples, 4-item index o socil support, worries bout crime, stisction with wht one is ccomplishing in lie now, stisction with mily reltions, nd so on. The lst column gives the results o regressing hppiness scores on ll the signiicnt predictors rom the previous columns. It shows, irst, tht in the context o tht set o potentil predictors, only the Mentl Helth dimension o SF-36 remined signiicnt nd its inluence ws reduced
36 2 Socil Indictors Reserch nd Helth-Relted Tble 2.3 Regressions o hppiness on helth, worries, socil support nd domin stisction % o vrince explined Helth dimensions Crime worries, soc. supp. b Other worries, ge identity c Domin stisction d All pred. e 34 21 21 40 44 N = 623 N = 807 N = 411 N = 584 N = 431 Predictors Bets Bets Bets Bets Bets Mentl helth 0.49 0.24 Socil 0.14 g unctioning Socil support 0.34 0.1 index Fir tretment 0.16 g index Physicl ge 0.17 g identity Diiculties with 0.22 g housing costs Personl worries 0.18 g (Fctor 1) Living re 0.15 0.09 (Fctor 5) Retil store 0.12 g ccess (Fctor 3) Idel ge 0.13 g preerence Friendships 0.11 0.13 Present ge 0.17 0.15 Sel-esteem 0.19 0.12 Accomplishing 0.17 0.14 in lie now Locl 0.09 g government oicils Fmily reltions 0.09 g Opportunities to socilize 0.09 g 8 dimensions rom SF-36 b Index o crime-relted worries, personl ir tretment index, socil support index c Five ctors rom preerences nd problems items, 5 ge-identity items, diiculties nd % o income spent on housing d 24 domin stisction scores e +b+c+d Vrible not in eqution g Signiicnce level too low or vrible to enter eqution
Explining Hppiness with Lie s Whole rom Helth nd Other Vribles 37 Tble 2.4 Regressions o hppiness on helth, leisure beneits nd ctivities, domin stisction nd socil support % o vrince explined Helth dimensions Leisure beneits b Domin stisction c Dom. st. + Helth d All pred. e 22 16 42 42 42 N = 390 N = 418 N = 323 N = 306 N = 296 Predictors Bet Bet Bet Bel Bet Mentl helth 0.47 0.12 g Psych. beneits 0.4 g g Socil support 0.15 Sel-esteem 0.3 0.28 0.32 Job 0.19 0.18 0.19 Fmily 0.12 0.12 g reltions Living prtner 0.17 0.19 0.22 Present ge 0.11 g g 8 dimensions rom SF-36 b 6 indexes o perceived leisure beneits + 7 indexes o Ctlogue beneits + 4 indexes o ctive nd sedentry recretion c 15 domin stisction scores d +c e + b + c + socil support index Not in eqution g Signiicnce level too low to enter eqution Tble 2.5 Regression o hppiness scores on helth nd domin stisction scores Explntory vribles Bet vlues Bet vlues Bet vlues Generl helth 0.1 Mentl helth 0.24 0.14 CES-D 0.32 0.26 Stress index 0.16 0.1 Role physicl 0.06 Vitlity 0.04 Role emotionl 0.07 Chronic cond. 0.04 Fmily rel. st. 0.17 0.1 Living prtner st. 0.16 Finncil sec. st. 0.15 Sel-esteem st. 0.33 0.17 Prov. Gov. st. 0.13 Fed. Gov. st 0.07 Present ge st. 0.14 Vrince explined (%) 45 40 53 Vrible not in eqution
38 2 Socil Indictors Reserch nd Helth-Relted considerbly (b = 0.24). Second, lthough Mentl Helth ws the most inluentil o the seven signiicnt predictors, it hd much less inluence thn the other six predictors combined. Third, the ddition o Mentl Helth to the seven signiicnt domin stisction predictors rom the ourth column incresed our totl explntory power by only our percentge points, i.e., rom 40 to 44% explined vrince in hppiness scores. From this point o view, sel-rted helth hd very little indeed to oer towrd respondents hppiness. Fourth, it shows clerly tht the mesured importnce o helth to hppiness vries with the totl number nd kinds o potentil predictors employed. I one were to judge tht importnce merely on the bsis o the inormtion in the irst column, one s judgment would be severely distorted. This is n excellent exmple o the sort o blnced view tht socil indictors reserch cn provide o the importnce o helth to one s overll qulity o lie (here opertionlized merely s hppiness). Tble 2.4 summrizes the results o survey o rndom smple o 440 dults ged 18 yers nd older living in Prince George in November 1999 (Michlos nd Zumbo 2003). The primry ocus o the survey ws to exmine reltionships between sel-rted helth mesured by the 8 dimensions o SF-36 nd leisure ctivities mesured in vriety o wys. As in the previous prgrph, our interests lie minly in the irst nd lst numericl columns o the exhibit. The irst numericl column o Tble 2.4 shows tht only one o the 8 SF-36 helth dimensions explined 22% o the vrition in hppiness scores. Mentl Helth did ll the work, with b = 0.47. The lst numericl column shows tht o the 8 potentil predictors in the regression eqution, only our hd ny sttisticlly signiicnt inluence nd together these our explined 42% o the vrince in hppiness scores. Wht is more importnt or our purposes, however, is the ct tht while the set o predictors vried in the lst three columns, the percent o vrince explined ws lwys only 42%. A set o ive domin stisction vribles lone, or our domin stisction vribles plus Mentl Helth, or three domin stisction vribles plus 4-item index o socil support provided eqully eicient explntions o respondents reported hppiness. Tble 2.5 summrizes the results o survey o rndom smple o 723 dults ged 18 yers nd older living in Prince George in November 1998 (Michlos et l. 2000). The min im o the survey ws to gther some bseline dt relted to helth nd helth cre, nd to ssess its impct on the qulity o people s lives mesured in dierent wys. Besides SF-36, our questionnire included the 20 item Center or Epidemiologic Studies Depression (CES-D) scle (Rdlo 1977), 12 items mesuring generl lie stress, 11 items mesuring stress t work, 5 items on smoking nd drinking behviour, Body Mss Index scores, list o chronic helth conditions nd 10 items on helth cre services in our re. The irst column in the exhibit shows the results o regressing hppiness scores on ll o our helth-relted potentil predictors. 45% o the vrition in these hppiness scores ws explined. From SF-36, only two dimensions hd sttisticlly signiicnt inluence, nmely, Generl Helth (b = 0.10) nd Mentl Helth (b = 0.24). As one would hve expected, CES-D weighed in negtively (b = 0.32), s did the generl stress index (b = 0.16). The second column shows tht set o ive domin
Explining Hppiness with Lie s Whole rom Helth nd Other Vribles 39 stisction scores ccounted or 40% o the vrition in hppiness scores, nd the third column shows tht the two sets together ccounted or 53% o the vrition in hppiness scores, with CES-D scores hving the gretest inluence (b = 0.26). Altogether, the seven helth-relted predictors ccounted or 60% o the totl vrince explined nd the our domin stisction predictors ccounted or the remining 40%. This gin shows us tht depending on the number nd kinds o potentil predictors employed, sel-rted helth ppers to be more or less importnt to hppiness. In ll these explntions, good mentl helth mkes substntil contribution to hppiness. While tht is not surprising or troubling in itsel, it hs n impliction tht is both. I will not dwell on the problem, but the ollowing remrks rom n intriguing review essy by Tylor nd Brown (1988, pp. 204 205) re worth repeting here. The mentlly helthy person ppers to hve the envible cpcity to distort relity in direction tht enhnces sel-esteem, mintins belies in personl eiccy, nd promotes n optimistic view o the uture. These three illusions, s we hve clled them, pper to oster trditionl criteri o mentl helth, including the bility to cre bout the sel nd others, the bility to be hppy or contented, nd the bility to engge in productive nd cretive work. The individul who responds to negtive, mbiguous, or unsupportive eedbck with positive sense o sel, belie in personl eiccy, nd n optimistic sense o the uture will, we mintin, be hppier, more cring, nd more productive thn the individul who perceives the sme inormtion ccurtely nd integrtes it into his or her view o the sel, the world, nd the uture. In this sense, the cpcity to develop nd mintin positive illusions my be thought o s vluble humn resource to be nurtured nd promoted, rther thn n error-prone processing system to be corrected. In ny cse, these illusions help mke ech individul s world wrmer nd more ctive nd beneicent plce in which to live. Explining Stisction with One s Own Helth Hving discovered tht one s sel-rted helth nd one s reported stisction with it ech hve some independent inluence on one s hppiness, resercher nturlly wonders how much inluence selrted helth hs on one s reported stisction with one s helth. Tble 2.6, rom Michlos et l. (2000), nswers this question. Using the sme set o helth-relted potentil predictors described two prgrphs Tble 2.6 Regression o helth stisction scores on helth scores Explntory vribles Bet vlues Generl helth 0.51 Vitlity 0.17 Body mss index 0.06 Helth worries 0.14 Stress index 0.09 Explined vrince 56%
40 2 Socil Indictors Reserch nd Helth-Relted bove, we were ble to explin 56% o the vrince in respondents stisction with their own helth. The Generl Helth dimension rom SF-36 clerly dominted the set o signiicnt predictors, with Bet vlue o b = 0.51. The Vitlity dimension hd reltively smll impct, b = 0.17, nd scores or Body Mss, Helth Worries nd Stress were ppropritely negtive t b = 0.06, b = 0.14 nd b = 0.09, respectively. Using the Prtt mesure (Thoms et l. 1998), we ound tht the Generl Helth dimension ccounted or 65% o ll the vrince explined by our vribles. Multiple Discrepncies Theory (MDT) In Michlos (1985), I proposed nd tested theory o subjective well-being clled Multiple Discrepncies Theory (MDT). Opertionlly, subjective well-being ws mesured by single-item, seven-point scles o reported hppiness nd stisction. MDT ws supposed to explin why people re hppy or stisied with their whole lives s well s with their jobs, living prtners, innces, housing, riendships nd so on. It ws lso supposed to explin people s reveled preerences or ctul choices in terms o their levels o reported stisction, which would mke it deeper theory (Bunge 1961) s well s n empiricl lterntive to utility theory. Most importntly or present purposes, it ws supposed to explin stisction with one s own helth. In ct it perormed this unction very well, s I will now show. The bsic postultes o MDT re s ollows. H1: Reported net stisction (hppiness or subjective well-being) is positive liner unction o perceived discrepncies between wht one hs nd wnts, relevnt others hve, the best one hs hd in the pst, expected to hve 3 yers go, expects to hve ter 5 yers, deserves nd needs. H2: All perceived discrepncies, except tht between wht one hs nd wnts, re positive liner unctions o objectively mesurble discrepncies, which lso hve direct eects on stisction nd ctions. H3: The perceived discrepncy between wht one hs nd wnts is mediting vrible between ll other perceived discrepncies nd reported net stisction, i.e., besides the direct eects posited in H1, six discrepncies hve indirect eects. H4: The pursuit nd mintennce o net stisction motivtes humn ction in direct proportion to the perceived expected levels o net stisction. H5: All discrepncies, stisction nd ctions re directly nd indirectly ected by ge, sex, eduction, ethnicity, income, sel-esteem nd socil support. H6: Objectively mesurble discrepncies re liner unctions o humn ction nd conditioners. Figures 2.1 nd 2.2 illustrte the reltions speciied in the six postultes. More detiled explntions o the postultes nd their sources my be ound in Michlos (1985, 1991). It is perhps worth mentioning here tht lthough MDT is deeper
Multiple Discrepncies Theory (MDT) 41 Fig. 2.1 Multiple discrepncies theory Fig. 2.2 Perceptul core o MDT thn utility theory, it suers rom some circulrity roughly similr to tht suered by utility theory. Ater ll, judgments bout the best one hs hd in the pst, wht one deserves nd needs (H1) re essentilly evlutions. So, lthough my view o the nture o vlue is in the nturlistic subjectivist trdition, insor s MDT is supposed to provide n explntion o stisction, my view is both less nturlistic nd more circulr thn the clssic nturlists or I would hve liked.
42 2 Socil Indictors Reserch nd Helth-Relted Applying MDT to convenience smples o undergrdutes rom the University o Guelph, I ws ble to explin 46% o the vrition in men s (N = 296) nd 53% o the vrition in women s (N = 341) stisction with their own helth (Michlos 1985). For men nd women, the most inluentil predictors o ll those postulted were the gps between the levels o helth wnted nd the levels ctully obtined. Encourged by this success, I went ter bigger ish. With the help o 68 collegues round the world, in the 1984 86 period, dt were collected rom over 18000 undergrdutes in 39 countries, using 19 dierent lnguges. The complete results o the investigtion were given in Michlos (1991, b, 1993, b). Tble 2.7 gives the results o using MDT to explin stisction with one s own helth. In the column under S (or stisction with one s own helth ) we ind totl smple o N = 8076 students, including 4,116 mles nd 3960 emles. MDT explined 51% o the vrince in reported helth stisction scores or the whole group, 52% or mles nd 51% or emles. Regrding discrepncies or the whole group, the sel/wnts gp (SW = 0.46) hd the gretest impct on helth stisction, ollowed by sel/others (SO = 0.37). Note tht the impct o the sel/wnts vrible on stisction is mesured by its bet vlue, which equls its direct eect, while the impct o every other gp vrible is mesured by the product o the vrious bets involved in its direct nd indirect eects on stisction. From the column under SW one inds tht the MDT set o predictors explins 42% o the vrince in scores indicting the perceived gp between the sort o helth students hve nd the sort they wnt. In the TESW column, one inds tht it is socil comprison scores (SO = 0.34) tht dominte the lot o predictors. The next most inluentil predictor is sel/needs (SN = 0.19). Considering perceived discrepncies reltive to helth stisction, the rnk ordering o the three most inluentil predictors or emles is exctly the sme s the ordering or the group s whole. For emles, we hve sel/wnts (SW = 0.45), sel/others (SO = 0.35) nd sel/needs (SN = 0.15). For mles, we hve sel/wnts (SW = 0.46) nd sel/others (SO = 0.39), ollowed by sel/deserved (SD = 0.13). Regrding the reltive impcts o sel/needs versus sel/deserved or mles nd emles in the domin o helth, considertions o equity re more inluentil thn considertions o need or mles, while considertions o need re more inluentil thn considertions o equity or emles. Considering the reltive impcts o my predictors on the gps between wht one hs nd wnts regrding one s own helth, the rnk orderings o the two most inluentil vribles or mles nd emles re the sme. The sel/others gp hs the gretest impct or mles (SO = 0.35) nd emles (SO = 0.33), ollowed by sel/needs or mles (SN = 0.17) nd emles (SN = 0.21). Since dierent smples were used to mesure the reltionships mong scores or helth sttus, helth stisction nd MDT predictors, we hve no wy o knowing the reltive strength o the reltionships in generl. Lest nyone concludes too rpidly rom the numbers we hve just seen tht sel-rted helth sttus predictors re lwys superior to MDT predictors o reported helth stisction, the ollowing
Multiple Discrepncies Theory (MDT) 43 Tble 2.7 Stisction with one s helth regressed on MDT predictors (deciml points omitted) S* SW SO SD SN SP SF SB TES TESW N 8076 8107 0 8562 8762 8705 8624 0 R 2 51 42 0 1 1 1 1 0 Predictor Sex 3 5 0 3 0 4 3 0 0 6 Age 3 0 0 4 6 3 0 0 4 2 WS 5 0 0 4 5 4 3 0 6 2 LED 2 0 0 0 0 0 3 0 2 0 ETH 2 3 0 5 4 7 7 0 1 5 SO 21 34 37 34 SD 6 12 12 12 SN 4 19 13 19 SP 6 12 12 12 SB 6 9 10 9 SW 46 46 Mles N 4116 4132 0 4375 4494 4456 4415 4482 R 2 52 42 0 1 1 1 I 1 Age 2 0 0 3 6 3 0 4 4 2 WS 5 0 0 5 5 4 5 4 6 2 LED 0 0 0 0 0 0 5 0 0 0 ETH 0 5 0 7 7 9 9 5 5 9 SO 23 35 39 35 SD 7 13 13 13 SN 0 17 8 17 SP 4 13 10 13 SB 9 7 12 7 SW 46 46 Femles N 3960 3975 0 0 0 0 0 0 R 2 51 42 0 0 0 0 0 0 WS 6 0 0 0 0 0 0 0 6 0 ETH 3 0 0 0 0 0 0 0 3 0 SO 20 33 35 33 SD 6 10 11 10 SN 5 21 15 21 SP 8 11 13 11 SB 3 11 8 11 SW 45 45 *Abbrevitions re s ollows. S Stisction with one s helth, WS Work sttus, LED Level o eduction, ETH Country o birth, SO Sel/other gp, SD Sel/deserved gp, SN Sel/needs gp, SP Sel/progress gp, SB Sel/best gp, SW Sel/wnts gp, SF Sel/uture gp, TES Totl eects (=sum o direct nd indirect eects) on stisction, TESW Totl eects on sel/wnts vrible
44 2 Socil Indictors Reserch nd Helth-Relted remrks re worth considering. For the 38 countries in my globl survey, on verge MDT explined 48% o the vrince in helth stisction scores or the whole group nd mles, nd 49% or emles. For the whole group, percents o vrince explined in scores indicting stisction with one s helth rnged rom high o 73% in Finlnd to low o 24% in Jordn. There ws one other country with igure in the 70s, Kore = 71%, nd there were 4 countries with igures in the 60s, Belgium = 69%, Austri = 64%, nd Germny nd Switzerlnd = 60%. Altogether, there were 10 (26%) countries in which t lest 55% o the vrince in helth stisction scores or the whole group ws explined. For mles, percents o vrince explined in helth stisction scores rnged rom high o 72% or Kore to low o 21% or Greece. There were 4 countries with igures in the 60 s, nmely, Finlnd = 69%, Austri nd Portugl = 64%, nd Switzerlnd = 61%. Altogether, there were 8 (24%) countries with t lest 55% o the vrince explined in helth stisction scores. For emles, percents o vrince explined in helth stisction scores rnged rom high o 76% or Finlnd to low o 25% or Chile. There were 5 countries with igures in the 60 s, nmely, Austri = 65%, Sweden = 64%, Germny nd the Netherlnds = 62%, nd Switzerlnd = 60%. Altogether, there were 11 (38%) countries with t lest 55% o the vrince explined in helth stisction scores (Michlos 1993b, pp. 77 80). Clerly, in certin circumstnces, the potentil predictors ssembled in MDT cn provide considerble power to explin people s reported stisction with their own helth, occsionlly perhps even more power thn resonbly brod set o mesures o sel-reported helth. O course the only wy to mesure the reltive explntory power o ctul helth sttus predictors versus MDT predictors is by undertking surveys in which ll relevnt predictors re included. I hve never done tht, but someone probbly will do it some dy. It would lso be worthwhile or someone to employ MDT in n exmintion o the impct o response-shit bis relted to pre-, post- nd then-test scores (Sprngers nd Schwrtz 1999; Sprngers 1996). Given the rich rry o judgment stndrds in MDT, wide vriety o response-shits might occur cross the period o time rom ptient s initil dignosis, through nd beyond therpy. Regrding overll hppiness with lie, MDT explined 42% o the vrince in hppiness scores or the whole student group, 41% or mles nd 44% or emles. The most inluentil predictors or ll three groups were the gps or sel/best, sel/others nd sel/wnts, in tht order (Michlos 1991, pp. 115 125). Helth-Relted Qulity o Lie Brodly speking, there re two types o mesures o helth-relted qulity o lie, nmely, those tht re speciic to diseses, conditions, unctions nd/or popultions nd those tht re not (Guytt et l. 1989; Ptrick nd Deyo 1989). Disese-speciic qulity o lie mesures re designed to mesure the qulity o disesed lie (QDL) nd s such they re not likely to be conused with generl mesures o the