2
CLINICIANS' G UIDE TO RESEARCH METHODS AND STATISTI CS Editor: Robert j. Harmon, M .D. A Tale of 'fWD Paradigms J ErFREY A. GUNER, I'II.D., GEORGE A. MORGAN, I'II.D ., AN Il RORERT J. HARMON, M.D. There is confusion in the social and health sciences about the ditlcrcn cc between quarui nuivc and qualitative approaches to research. Much of the confusion , as ind icated in the last article in this series, comes from equat ing these paradigms with the way data are collected . Investigators often associate "sub- jective " methods of data collection, such as open-ended inter- views, with the qualitative approach and "objective" method s of data collection with the quantitative approach. The type of data collection is nor adequate to distinguish between the 2 paradigms. Paradigms V ;;/J,l/ iJ ,I The te rm, coined hy Thomas Kuhn (\9 70 ). rdl:rs to tl« hclii:# mcmbrrs f{ '1 scirn- ti/ic community. It is not strictly a methodology but more of a that guides how the research is to he condu cted .A par ;ldigm det ermines the typl' of questions that will he asked and how they will he answered. There currently are 2 major paradigms within the so cial and health sciences. One paradigm . of tell referred to as th e quantitative paradigm, has been the dom inant one and is usually associated with rhc so-called "scientific method. " Qualitative theor ists {e.g.• Lincoln and t;uha , (']H '») use the term logical posiriuist to identify the domin ant paradigm. Even though the term is no! totally appropriate , we will usc "positivist" when referring to quantitative paradigm, in pan to distinguish it from quantitative data collection and analysis. The other par- adigm, rclcrrcd to by Lincoln and (;uba (l'JH'i) as naturalistic or constructivist, is usually associated with a qlfJIlitlltil't' approach to research. / lfn 'Plt'tI (kt oh,',. I . 11) 98. I ),: (;/i l/tT is () ( Olfd/ ;OI/, t! 11'I'Iilpl' .l1ul l :illll ",II;OI/ ,'lIrI I ),: i., 1: 1/11( /1/ ;011 ,/lui Human I >t·l'rla/' II/to /It. ( . ., I/Ol i/(io Sl'lft' Fort ( .idIi IlJ. I),: Harmon i.1 /'rok, ...-or ,{ ,/1/(1 I hl i,urit'" , ""I /ll'dti. I t u -ision a( llnirrrsitv or ( ,i,I" !>"I,, Sd",,,1"r Aluli- rinr. I srm-vr. / J,t' ,1I11/'Or.\ th.rnl: iV' l1Il:J' l'lummrr I ;, ,.I1M IIU. .;oip' /, rc!' oInu ioll . Pa rt, ,{ ,hii column an: "rI"plt'tI. with pt'","i. ,,- .;;,," [rom ,ht' pullli.dla ,lI if l thr authors, .FolII ( ,'/i llaJA ,l1if l .\! 01:f!."n (,/1 (ill prn .d , Ih'\l'.ln, .-h Ik \ign .md An.ilvvi -, in Applin i Sl'u illt.' : An I llt q !,r atnl I\ pprnar h. ALthu'fif,. A?: Frl lJtll oll . l 'crmi ssio») '0 rrprin: or IlIllpl ptlr' tln .. 01/ 11 "''' "/11 ,.1 J,t' ohldi llnlti· tJ111 l-rlb.uon. N"prim ""'I" I"f' f" / h: l Ltrmon . ( :1'/ 1 N"(/I/I .!A'O·f. 1:(:1 t.« : Box (:.!(,S- ).!. ·,l {)O Eu, Nintl. A " t'I IIIt ', 1 )011'('1: ( .'() 80l b.!. O.l·i21$O.l.OO/() " \ "' )') bv rlu- Amcri c.m A""knl\' of ( :hild .In.l Ad"I ""l 'nl l" y,hi. Ilr\'. j42 Lincoln and (;uba (1')H5 ) stated 'i axioms or issues. from their point of view as qualit ative researchers, which separate the 2 paradigms , Sinn' that t ime. they have modified their position somewhat , but it is in structive to examine these orig- inal 'i issues in some detail to gain a clearer picture of tlu- dif - ferl'nces between the 2 paradigms . Issue 1: The Nature of Reality Is there one I't'illiry 01" lire there multip]« mtiities?This issue is the most divisive because seemingl y there are either many realities (consrrucrivisrl or one (positivist); both positions cannot exist at the same t ime . We think that part of the issue comes from the lack of an agreed -upon definition of real ity. Phillips (")92) point s out that there is confusion because dif- fl:rent people and different cultures have different views about what is real (which seems unden iable). In addition, there is the issue of whether or not we can know which view is correct, or even whether there is a correct view. Because we might not be able to reach agreement. it does not lollow that there is more than one "real ity." Quantitative resear chers, recogniz ing that participants ha ve diHl:relll perspectives or p oint s of view. report those as uaria- In addition, qu ant itat ive studies often examine f;l ctors that are related to and perhaps cause different perception s. That a person states a beli ef does not necessarily mean that it is "real," even fllr that plTson. Sometimes people lie or arc con- fused about their beliefs. It seems to us that it is important to acknowledge that people have different perceptions, hut it is best to investigate why per ceptions seem to be diflcreur and to be cautious about assuming that what one says is what one really believes. Issue 2: The Relationship of Knower to Known C III illl'i'sti gJlto rs IlIId pnrt icip.ou« be ill II study unthout influencing each othrr? "Double-blind" studies acknowledge this issue. However, only nai ve investigators would argue that their observations are not influenced hy attributes of wh o is observed. The issue flH the positivist is to determine how much of the outcome might be due to these cllccrs. for the con structivists. how much difference does it make if the ob- server is a participant as compared to a "sile nt" obser ver ? We th ink that issue 2 separates the approaches on a relat ive rather than ab solute basis. The posit ivist is usuall y more confident I. AM AC A I l CIIIIII AIH 11 I'.S( . I' \ YC Il IAT RY. \ H: \ . M AR Cil I ')' )' )

A Tale of Two Paradigms

Embed Size (px)

Citation preview

Page 1: A Tale of Two Paradigms

CLINICIANS' G U I D E TO RESEARCH METHODS AND STATISTI CS

Dcpf(~Y Editor: Robert j. Harmon, M .D.

A Tale of 'fWD Paradigms

J ErFREY A. GUNER, I'II.D., GEORGE A. MORGAN, I'II.D., AN Il RORERT J. HARMON, M.D.

There is confusion in the social and health sciences about the

ditlcrcn cc between quarui nuivc and qual itative approaches to

research. Much of the confusion , as ind icated in the last article

in this series, comes from equat ing these paradigms with the

way data are collected . Investigators often associate "su b­

jective " methods of data collection, such as open-ended inter­

views, with the qualitative approach and "objective" methods

of data collection with the quantitative approach. The type of

data collection is nor adequate to distinguish between the 2

par adigms.

Paradigms

V;;/J,l/ iJ , I p,m1f/~l!,/Il? The term, coined hy Thomas Kuhn

(\970). rdl:rs to tl« hclii:# jAlrt'rl l~v mcmbrrs f{ '1 s/,t't"~/ic scirn­ti/ic community. It is not stric tly a methodology but more of a

philosop/~v that guides how the research is to he conducted . A

par;ldigm det ermines the typl' of questions that will he asked

and how they will he answered.

There currently are 2 major paradigms within the social

and health sciences. One paradigm. oftell referred to as th e

qu antitative paradigm, has been the dom inant one and is usually

associated with rhc so-called "scien tific method. " Qualitativetheorists {e.g.• Lincoln and t;uha, (']H '») use the term logicalposiriuist to ident ify the dominant paradigm. Even though the

term is no! totally appropriate , we will usc "positivist" when

referring to quantitative paradigm, in pan to distinguish it

from quantitative data collection and analysis. The other par­

adigm, rclcrrcd to by Lincoln and (;uba (l'JH'i) as naturalistic or

constructivist, is usually associated with a qlfJIlitlltil't' approach to

research.

/ lfn 'Plt'tI (ktoh,',. I . 11)98.

I ),: (;/il/ tT is Pro/~''iJm' t~r ()( Olfd/ ;OI/,t! 11'I'Iilpl' .l1ul l :illll",II;OI/ ,'lIrI I ),:J\lm:~JIIJ i ., l'rt~/i'H(lr t~r 1:1/11(/1/ ;011 ,/lui Human I >t·l'rla/' II/to/It. ( ..,I/Oli/(io Sl'lft'

l lllil'( 'nj~l" Fort ( .idIi IlJ. I) ,: Harmon i.1 / 'rok,...-or ,{ I J~l'l 'hi,It':J' ,/1/(1 Ihli,urit'"

,""I / ll'dti. I tu -ision a( ( .Y,ilrll'~ l'dIJ~lfrl'. llnirrrsitv or (,i,I" !>"I,,Sd",,,1"r Aluli ­rinr. I srm-vr.

/ J,t' ,1I11/ 'Or.\ th.rnl: iV' l1Il:J' l 'lumm rr I;, ,.I1M IIU..;o ip' /, rc!' oInu ioll. Pa rt, ,{ ,hii

column an : "rI"plt'tI. with pt'","i.,,-.;;,," [rom ,ht' pullli.dla ,lI if l thr authors, .FolII( ,'/i llaJA ,l1if l .\!01:f!."n (,/1 ( i ll prn .d , Ih'\l'.ln,.-h Ik \ ign .md An.ilvvi -, in Applin iSl'u illt.' : An I llt q !,ratnl I\ pprnar h. ALthu'fi f, . A?: Frl lJtll oll . l 'crmissio») '0rrprin: or IlIl lpl "'~ l' ptlr' (~r tln ..01/11"' '' "/11,.1J,t' ohldillnlti·tJ111 l -rlb.uon.

N"pri m ""'I" I"f' f" / h: l Ltrmon . ( :1'/ 1 N"(/I/I .!A'O·f. 1:(:1t.«:Box ( :.!( ,S-).! .·, l {)O Eu, Ni ntl. A " t' I IIIt ', 1 )011'('1: ( .'() 80l b.!.

OIl ')O- Il ~ (I~/'I'I/.IIlO.\ O.l·i21$O.l .OO/() " \"')') bv rlu- Amcric.m A""knl\'

of ( :hild .In.l Ad"I""l'nl l" y,hi.Ilr\'.

j42

Lincoln and (;uba (1')H5 ) stated 'i axioms or issues. from

their point o f view as qualitative researchers, which separate

th e 2 paradigms, Sinn' that time. they have modified their

position somewhat , but it is instructive to examine these orig­

inal 'i issues in some detail to gain a clearer picture of tlu- dif ­

ferl'nces between the 2 paradigms.

Issue 1: The Nature of Reality

Is there one I't'illiry 01" lire theremultip]« mtiities?This issue is

the most divisive because seemingly there are either manyrealities (co nsrrucr ivisrl or one (positivist); both positions

canno t exist at the same time. We think that part of the issue

comes from the lack of an agreed-upon definition of real ity.

Phillips ( " )92) points out that there is confusion because dif­

fl:rent people and different cultures have different views about

what is real (which seems unden iable). In addition, there is

the issue of whether or not we can know which view is correct,

or even whether there is a co rrect view. Because we might not

be able to reach agreement. it does not lollow that there is

more than one "real ity."

Quantitative researchers, recognizing that participants have

diHl:relll perspectives or point s of view. report those as uaria ­

bili~v. In addition, qu ant itat ive stud ies often examine f;lctors

that are related to and perhaps cause different perceptions.

That a person states a belief does not necessarily mean that it is

"real," even fllr that plTson . Som etimes people lie or arc con­

fused about their beliefs. It seems to us that it is important to

acknowledge that people have different perceptions, hut it is

best to investigate why per ceptions seem to be diflcreur and

to be cautious about assuming that what one says is what one

really believes.

Issue 2: The Relationship of Knower to Known

C III illl'i'stigJltors IlIIdpnrt icip.ou« be ill II study unthout

influencing each othrr? " D o ub le- blind" studies acknowledge

this issue. However, only nai ve investigators would argue that

their observations are not influenced hy attributes of wh o is

observed. The issue flH the positivist is to determine how

much of the outcome might be due to these cllccrs. for the

constructivists. how much differen ce does it make if the ob­

server is a participant as co m pared to a "silent" observer? We

th ink that issue 2 separate s the approaches on a relat ive rather

than absolute basi s. The posit ivist is usuall y more confident

I . AM AC A Il CIIIIII A IH 11 I'.S( . I' \ YC Il IAT RY. \ H: \ . M AR Cil I ')' )' )

Page 2: A Tale of Two Paradigms

that bias can be overcome. but bot h para dig ms need to besensitive to th is issue.

Issue 3:The Possibility of Generalization

Call the results ofan individual study be generalized to otherpopulations, setting), treatment variables. and mrnsurrment uari­ables? Linco ln and C uba (1985) arc correct tha t few studiesusing a posit ivist approach have employed proper samplingtech niqu es. Usua lly. part icipa nts are not randomly sampledfro m a target populati on. Instead . parti cip ant s arc usuallyob tained fro m a "co nvenie nce" samp le, often fou nd at theinvestigator's clinic.

The cons truct ivist professes to make no claims for general ­izing the results beyond what was found in the study, but manyqualitative studies seem to make conclusions beyond the spe­cifics of their findin gs. Meta-analyses allow quantitat ive inves­t igato rs to co mbine stu d ies to ma ke better generalizat ions.T his issue seems to separate th e 2 approaches o n a relativerather than absolute basis. Researchers from bot h approachesshould be cautious about generalizing their results .

Issue 4: The Possibility of Causal Linkages

Call we ide1ltifY causes? Regarding th is ph ilosophical issue,positivi sts believe th at un der th e proper experimental con­d itions, o ne ca n co ncl ude th at th e indepe nde nt var iable"caused" th e change (effect) in the dependent variable. Thesepro per cond itions incl ude the random assignment of part ic­ipants to gro ups. However, few positivists are willin g to makemore th an a pro babil ity sta tement about causes.

Constructiv ists rule out the concept of causality on manygro unds. but perhaps the most salient is thei r position thatmost event s have multiple causes. Lincoln and Cuba (1985)take a much stro nger stance on the issue of causality, suggest­ing th at it is impossible to separate cause and effect. Instead .they introduce th e concept of mu tua l simultaneous shapin g.T hey argue that ". . . everyth ing influences everything else. . . .Many elements are implicated in any given act ion. and eacheleme nt interacts with all of the others in ways th at changethem all . . ." (pp. 151-152).

There is no way to resolve differences between the 2 para­digms on issue 4. Lincoln and Cuba (1985) arc certainly cor­rect in pointing out th at mu ch be havio r is bo th cause andeffect , that there is mutual simultaneous shaping of behavior,and that causes and effects are di fficult. if not impossible. todistingu ish. Also, it is true that journalists, the publi c, and evenresearchers arc too loose in using words like "cause," "impact,"and "determinant. " In fact, one of the key points that we want

RESEARC H ME THO DS AND STAT IST ICS

to make in this series is that such words sho uld be used withcau tion and then prob ably onl y after performin g a stud y with atight , rand omi zed experimental design.

Issue 5: The Role of Values in Inquiry

Ca n researchers be unb iased? It is im po rta nt for all re­searchers to recogni ze th at research is not value-free. Onemight ask the positivist, "Who selected th e research problem,variables. sample, or pa rt icular treatment?" Of course, theinvestigato r selected all aspects of the study. T herefore, th econstructivis ts are correct in assum ing that research is notvalue-free. A part of a constructivist research report is to statethe biases of the investigator. or to "co me clean."

T hat th e investigator may have a bias, however, does notnecessarily mean that no thing can be don e abou t bias or thatth e outcome will be in th e direction of th e investigator'shypotheses. There arc probably far more studies that fail tofin d signi ficant d ifferenc es between treatment and co ntrolgro ups tha n those that find significant d ifferences. Fur ther­more, replication att empts by other researchers help to ensurethat somet hing ot her than the investigator 's bias is influencingth e results.

Issue 5 appears to separate the 2 parad igms on a relativebasis. The degree of subjectivity in most constructivist studies,in which the investigato r is the "instrument," seems to us fargreater th an th at in the positivist approach. H owever, th isd ifferen ce is not all-or-no rhing. Researche rs in both para­digms should acknowledge th at there will be subjectivity andbias; however, they sho uld do what they can to minimize th epot enti al effects of biases.

T hese issues lead to the following qu estion : In the healthsciences , is th e purpose of research to predict and discovercauses? T he answer to thi s question provide s a distinctionbetween the 2 paradi gms in how research should be carriedout. We believe that if the purpose of th e stu dy is to ide nt ifycauses and predict behavior, th en a methodology for carryingout thi s research devel oped from the tradition al scie ntificmeth od s should be un dert aken. Alte rnatively, if th e purposeof research is a full descripti on of the participants' "voices,"then a methodology based on constru ctivism seems to be welladapted to the problem.

REFERENCES

Kuh n TS (1970), The Strucmre of Scientific Reuoltaions, 2nd ed . C hicago:Universi ty of Chicago Press

Lincoln YS. C uba EC (1985 ). NaturalisticInquiry. Newbury Park. CA: SagePhill ips DC (19')2) . TheSocial Scientists Bestiary. Oxford. England: Pergamon

Watch our upcoming issues for the next two articles in this series:

April 1999: Evaluating rh e Va lid ity of a Resear ch Study GeorgeA. Morgan , j effreyA. Gliner; Robert j. Harmon

June 19 99: Resea rch Problems and Va riab les Robert j. Harmon. GeorgeA. M organ

J. AM . ACA D . C H I LD AD O LESC. PSYCH IAT RY. 3H : .\ . MAR C il I<J<J<J 343