SreruENn VAN EVERA GUIDE TO METHODS FOR STUDENTS OF POLITICAL SCIENCE Cc ‘Comnell University PressHypotheses, Laws, and Theories: A User's Guide What sa Theow? Distant ar hon” ti by pitas cil science are ypc and verse! I reconntea the following a simple framework that expres thir main mete ing while alo sping oat cements hy fen ei “Mores are general taterents hat see an explain the 1, Netti ea lt ht Teds ‘Sot Wl neue Pap of Sele ‘is temige MI se opal heey kl Sig Free Snes fe am Cae ae Pure Achy ween gna or bron = Sete nar are eae pene ceeenn maemo ‘descr and tna one pnenoncg htsl t SOrumiicaafearaignneeeary4 Geo nas eS ie ‘ses recs fess of phenom, They ar composed of ‘usa avs or hypethoss explanations snd sncedem eon ons Expinatons are also compeae of cul liso hype ‘es which are inturncompoced of dependent ane inert ‘arable, Fourteen definition bar mention tow An ohare euros betwen twophenamena Lam cn be ore prosbioe. The arma fae esha) Theat fer ane probs lone CH A then sometimes 8 with probity 3°) "Hard sinc is sy eri a Newly als Soner le prota Lacan be cus (Aca onan al (4a B aca by C bec areca bt ets ae he ther}? Our prime sah fr sal la We eae the poly hat far ‘ noncnsl mainy fo et oat me ‘inten tho pith herve ows spd nme et Sey eee eeeeatente oe oe Len gaasaaeearee Seo See eae ‘hse posh . opts arta Jp A conjured wtrabip between to phenomena Lake lw hype ca 90 typ heal (1 sie at ‘us and enc Terme tat A fo areas by Cane Aan eae ‘ome bt ater emses te the seo 2 eal lw (1 bane ented the A ‘ie ora cousl hyphen le Star cues ogee wha expan tian the nal ao hypothe hte ete how A caus B Note tele “naa theory” Is ote ea for mee ideranging tore, ut al thse ee detain genera to one degree planation ‘The causal ws o hypothe tat cope theca woe plocmenen binges, Mowinghow causation ccc Aca B ‘cu cnn whi ces ih sume) ‘teesent condition? A pheomenen whose pesenes ats ote ts 5 i 4 ye Se ie a at neyo Se er reheated Sie tsa peg a eal oe ap see lly has ci sa as Mri chy ae ee ee ‘hepa pond ae aca cpeehe Sacrpig mer eee” merase ree as hh pe a En a Poser Says 5, ti ot il Hee, f iF id Oder Soar HE nyt Sane Rw a fem ngeindw te oes Sa Pt magi the setion of nl aw oye Phi Withoutitcnetin opens rely (7A eae some Bi © abn moe BC te promo, “Surin ras gs grow, Dut ce ange growth ony in eee =i") orm a a CA ‘nutes BAC pram tere nt “Sante mes ae ebony ite abo tse it) ‘Wecan tite ated oon a cata aero ns Bi 1 prs not ail nas gs rom Bt only ea at ‘one since “Ansett nto clad “in teacion ten” “ital dione ee thling candies “eae conony” “ecamitons” “setting. sno? ‘grin condos” espns ‘serine cndeng” or" able [Accomp hat canhave vray Sh “dage a acy fn cousty 6 the "chro het pctv" rap ‘lpg. Independent arabe arable feming te eas nomen oy ‘facusl dhary hypothe In thy Pt “ery cases democracy” he fg of ery ie the indoprdent hls syatoe sets corte ee pa eke os a tat epenet varle A arablefnming he nse phen on ems thor or hypeties In ey thes “tteaey comes democracy the segee of democscy tthe dependent vate Inve vse en) ‘A sable faming nrvening phenome an incied ina eu! hry explana toner pre a esd by the TV and cae te DV Inthe theory ‘Sunshine camer photrythst cn, ms go” phan the ter ning aie ‘enon vse (CPA valseuming nate conn, he alusafendin vane gover he sina hw mpc tha 4 ots veo and other Ins. tthe hype “Sunshine mh gran pot oy sho get somes the me of a fal es cnion var study varie XA vrs wom cee o lit we ck ‘0 dicoee with aur mei A popes stay wri can bee TV DY a eV, line hpstss The otrarching ype tht ese relosomtip between tv’ nde ft and dependon vaioe 6. Wh iit doped Indi ce eg Fone che wth stan Renan the nda ese) “Seach ae cette and Scat artigo tp pera et SSE See Sa ey2 Ghost Sata Pe Se ‘plasty Hypothesis The need bypethss that conte 1 ory oxlnation® ‘he ype soko. Ako eld ‘he et pane” types Not: theory the, is nothing more than 9st of connected cal as or hypotheses We can always “arrow-diagram” si, ike thi In this diagram A isthe hooey’ independent variable Bis he dependent ysis, The eters y andr indate nervenig wat 1 er rt ty te a Seo pay ot ‘eo’ Nt Alla p's $s ak ‘Theses dl rel mee eh a eee eee ee ee er tidy ma a en a cramer ne Saris gear wae he cereron e Epona oan nates mame amt tet Se Gee nee eee islets ile narrate, “sta mrpind iypanses one! ocpian pra dasaerenee pt Lea ge a koeclrt tet icllinaietscan ce 2 tis, an Teas shies and comprise the thors explanation The popes “A isthe theory’s prime hypoth whe the proposal tat “A> 0 %q-9 1 and "t Baits explanatory hypotheses We cana condition variable nating them by wing the mulation symbol "0 Here Cis a contion variable the impact of Aon qsmagniicdby ahigh value on and eed by ow vase on An example would be nahine + pokes ge mth ‘One can playa theory's explanain a any level of desi. Here ve elaborate the ik Between rand to show explanatory variables sand ‘One can extn an explanation to define mare remote causes Here remote cause of &(Y and Z) ze detailed the malian ah ly ind ha eC mai int tare aay aps tO etl aigeto ese Saet ai rs c ‘We an deta dhe cass of onion vale as hee with the cause of ‘Thr isno into the umber of antecedent conditions we can frame, Here more corlitions (ate speed. Yao Aag rset ae xe : > 5 (One can sd more avenues of esustion betieen cal and ‘caused variables, Here tro chans of entation beeen and B (uring through intervening varbiesf and g) oe ade, Produce «thea theory ceeeere yay Sasrsrais xae ' ‘A teory hat cannot be arow-dagrammed is ot teary and es, am Te needs reframing to tere a theory (Acsoing to this tein uch polite scence theory” and “Uhorete? wingf not theo) What I «Specie planation? iplanatons of sprite eves (patil wars, revatlens leton cutee eam depressions. ands) use heres sed are nme ke thoes. A good explanation tells us wt ‘Spec anes produced aspecc phenomenon an ientifis the {recral phenomencr of which this pec cn I an example “Tivos eanepi beat netn: spsiccaplamtion An eaplanaon ct hn spesi foms tht ecount fora since event ke a he 0 descbes and expan cane ant ‘ec bot thse ies and fi ate Frm nse ere (Ts “np sn ease ggresio, casing a” ‘hey "German exansonizn ese Ge ‘man agrenon cusing Wed War it sjecie xplanton) Specie expats (nocd “prt eplinatong” ‘Spposnd “gel explanations”) Specilceplenatens mein we type The second ype (gene slaed spec explanation’ mor el angensalize pee A speci oxplanaim st doesnot deny paaon he theary shat th operating ene spl (Garman ena Weld War ‘Thecxplaatn dees not ser the(oe Moa St Pt tion of hat Germany a example»? ‘Aspclicenplinaton tates ie ‘oe dt gore operon Geren ‘prion tent Whe Wa 1° The peang cay, "Geman exer” ‘St cumte of expansions wih the Independent vain he hypo “ee users aes wa") peers pec ‘rian Spc explanaton ar cmnpone cn ase ater sh antecnent phenom “s ‘ul phenomenon Toph doing te cus ‘atid phenomenon The phnmnn bing ned, on) Iolerenngphenmees hunni xm he explanations a pParaon Thee ae cise bythe ca henerenen and cake the ee tote pnamenaPhenomens whose present ave ce “aD ‘moped cma sain of hel ‘ace explain pono sftp eer ie ‘PTR hn den a elon ed Ean he pr eet Rae ewecenr sae Sep pro ot eae Li an speriogca” a ‘Deena nt Sule” Ong pels plone ve fw Oak ‘Spel Be plinth dupe tng pm epoca foros” ots Ee w pte a, nT We arow-dlagrm spec explanation the sme way we do theo ” Athery spans + Aggeson + Wer Agermlle peiie German expansions > Gan spe ‘plantion son» Wit Wer ANeongeretalizst Germany > Outa of Shing on Sep Spesiicexpinaton tember 39» Wirt Marit What ea Good Theory? ‘Seven prime atbutes govern a theo’ iy. 1 Aghod theory has loge eplntory pouer The nyse pent vaable asa lage eet ona wide range of phenomena “under «wide rags of eonions. ase charts govern ‘explanatory power Inport: Does vance athe vlc on the Independent va she caso lege o small variance in the vale on the dependent ‘alae An important heey points oa cause tht large Impact—one that eases large variance on the dependent veh Sapte gg arr ‘rita e e ‘team eres gay oe Sine pe elegy ernie’ exactions mop vied pe a(io so Sas Ps so hie he grater the variance produce, the siete theory’ ‘explanatry power ‘Esplntory range: How many cases of phenomena does var ance inthe value on the theory's independent varie afc, Ince expan? The wider the range of ated phenomena, dhe rater the theory's explanatory power: Most ol uence theo fies have narow range, ut few gure explain many diverse ‘borin ‘Arpt. How common i he theory's case Inthe ea word? How common ar antecedent coos tat ctivate s ‘operation? ‘The moe prevalent the cases ane condos of tie theory the greater explanatory power” The prevalence of [ee rt ania wit et Say ae ok ‘ald npmang engl sin rer {5 a fos mel ol “upping poe delay ce ‘inet tary ea taraen rion ee pow aet ‘i tae he army Sonne ea 9 ‘it me cl as hi ih esta aoe ‘an arn Ss hea ea pa ae ‘Spina l seaa l n ey ‘Timm fe Mar Ota ea ige [it Ona ec re Roe fet Cpt i ey ‘BW may pyar a ‘natn Sat he Urey Ct fe SESAME RR EY he ie el ay 17 fol ote wet aoc ie play poner ia cn amen Be al wo ty mi ne Sets nee etna oe ‘hin econ pos theme 0 Patt a as ‘hose causes and condtons in the pt aver its power foo hin history. Ther curent and fatre prevalence gover is ower to expan present and fue even. 2: Good thooris elucidte by simpityng, Hen good theory spss. Tse fe vanabls imply eeanged exp is ‘lcs Gulning prsiony often rule sve arf ofenlanatory power, however that saci is too ange tse wert ‘whe, Wecan tolerate nme complet we need ito explain he worl 5 A god theory i “ting Ut it sates ur csi -Aihoory sunstetyng lever us wondering what ase the “nse proposed by tho theory. Ths happens hen vol point "fama cases hoe eases i ee a mystery polit tien on explained he lactones" did ge enoigh vot” ‘hiss truebutunsating We si want to know why she dd t {get enough votes ‘he farther moved ease stands foe its proposed ete, the more sting the theory. Ts “nit us fae” es sting han “changes in cea srl temperature case shits in amosphede wind pater, causing ahs in aes of heavy eal, nosing drove, esing famine ‘6A good theory Is dary rane. Otherie we can infer Prdictions from i testi oF apply io concrets tons ‘A clea famed theory fain ie varlables fen coneepss Uthat he theorist hs any dein. ‘A clay famed teary inchs fl tie ofthe theory's ‘explanation. Ttdoesnot eave us wonderinghow Aeuses Ths "changes In azn temperate cts fie” ie lene smplete than hanges in cee temperature este sis in ater wind patterns aus shifts In areas of Bea rif cnn droughts causing fine” ‘Aclesry famed theory Includes «statement of the antecedent coatons that enable operation and govern npat Other™ hie wt Sn ef Pai Shee wise wecanot el what cases he Ueory governs and scant infer seh pobey prescriptions Foreign ply dase often happen boone polcymalers apply valid theories to nappropriaecreumstances Consider the Typttass that “sppeasng eer statr mas them more ager ve causing war” The was tue wih Germany during 2938-39, atthe opposite is somtimes tre afm sand mak the oer moro spgrestv,couing wa To ave poy bck theres, Poiymahers ust kaw te anecedent conan that decide 2 fem san will make others more oles aggressive. Parallel problems arise is all pocymaking domaine and highlight the Importance of taming entedent condions deat Ny Aad theory iim prinpe fea: Dts that would fl ‘ay the theory ean be deta although i€ may not now be oa" "Theories that are not ely amo may be noniable be ‘aus ther wages prevents intentions fom inferring Pre ‘Scion from he, “Thora hae omnpeticsions that rei bao served events ate also nonalsiable Emp fess cannot cor ‘borate oni sich theorie Beste al evidence consistent ek them. Helou theres of phenomena have this quai appy outomes are God's wa dams ace Gx’ punish met erueies are G's tests of our faith and outcomes tht ‘le these broad eatgoele see God's mysteries. Some Marist “puma share this me-predictonsl at 1. Agen theory xp ipo pon tensive ques 2 Dmg tel Hg Py Ne ep eign tb, Ding a {meen nt Par ery fn al aye (Sy Seng Ges ay ty “Se Sy ‘cnn as ad Lenin an Fla le ‘Holy ta sv Srey fying i ASO ‘ei mo a tos a Tk honsthatnatior tothe wider word orithelpcthers answer sch _getone Theories ht answer whe question ae len wef fen they aner thee questions wel (Mach socal sence ‘hoor has te zealot elevance and th i thst) 7A ge theory hs pracpie be yee wel py sscorimendations. "A theory gains prescrip chines pontngtomanipulable ‘use ince maniplate cases might He controled by man seton Thus “cpt causes imperil, causing Wa is se (ef han “lfersive mltay postures and donee ous an” ‘ever if both doses are equally valid because the structure of atonal coors ses manipulabethan national military pos ‘ures anddocrines, “Teaching chauvinst history shoal eases war iseten mare seu since the content atonal eduction mere soy ajurted thn ations miltsy poly ‘Ahwory guns prescriptive richness by ident danger tht sould be averted or mate by timely ountermensines. Th Choos xpi the cases of husicanes provide no way 9 prevent them, bat they de elp forecasters warn trated com nie fo acu property and tke thet "A theory gine preserve hnes by iden att condions rested for operation aw pn). Theta thee ‘euilons re specie the greater ou abit o eid mia ingthe theory’ prescriptions to stations thatthe theory docs nt oven ‘That i no yoneralyacoapted recipe for making theres Some cus ie deduction, tering explanation om mone 2 Aig inet peeps hep te, Fe soak no eda ne Ss aka Tiyan “hmm ‘me tion enn sar ul eed cg2 (hes oe Sve Pt ee genera alesdy- Unfortunatly, this deerbes te predictions ve usealy work wih Inwrpretve disputes often arise fom disputes ove what ou ‘comes teres pret. Dos Rein make predictions ht ens ace wnt ge dt Se ing ‘i Cine avers pues a as contradicted by He end of tho col war? Sone schol yy essay no. Such dsagreements canbe narrowed ifthe ge Seay framed to bogin wih (ince vague theoretical statements Isa more room fr divergent predictions) and i ested pte og ate explained and tie. Inerpretve disputes ao arc fom quarcks over the unig nerand cette prdicns Is the prediction unig? Thats other theories or explanations pi the sae rau? Iso ‘paved tut i Jose tnpresive, The Recher school of historians gues that the December, 13, Garman "War coun an ‘er meeting between Kaser Wiel tan he mary leaders {encovered nly nthe gm sgl pot mong the German lie to instigate a mj war” Se ers newer hat the Kal ‘evs mercural peronaty explains his tlic tlk that ‘neting—he often Blew of seam by saying tng he i no ‘ea. In ahor they pitt a competing expanaon for eves ‘hat some Fishes claimed was "amokang un” fer thei hie ot theory the war The question hen reson the plasty ‘thi competing explant 1s the pricion axtin in other Words, uregatvoca? 1 ot: unked tests ae les damaging Some historians aga ht ‘he Spanish Amerian war of 848 ao hom a conpiacy of cept seekng US. ders who hoped oe he PHlppines trem Span Tho aon of any tnton of such a onepacy a these leaders! dares and private lters oti ofa archives ‘vince eters hat there Was nan. this iw the onepracy theory pedis with high certainty tate of eonapiacy shoul be found in Bese seconds Conspiracy thors answer Otten eres crm one are Bate ep Pate pe gem? ey Sorenstam claves Pr hry ‘Rn acento asym Cpe Be, sebtie gsehie nel ioo ine i Sus Pe Soe that god emoptr hide tel conepiade fen vig m0 secon The eropcy ory isla try ge tee, {henry pics ony ely tat spose ecard thir conpony hrc iackoteuch ord amen sain the wind” ta inset they oly meal The guoion ings ote heen bt depen of eet tuft ty rents spiny wel ve This caso gh the need ods the unig and cide of ted predicts when nerpretng eee Atevieresnetequl crm he prs ey rt ‘aval angu or eta Hen author sould cotton the que nd es te prions sng ee Wk fat at be or anh the theory For tap ec fom tts nde condto her cunteag cae ye se tht omer edited aon, Pape of uch ee ies eos he ys ha ge po tte tas high inp! Bat ad ety may nko os ‘ste courereng a nk ato Sac ts eas by meng ie teaineann the Imcsigaon ‘ind of htt Ba ese try Dons ps te ‘sth ee Anthro sper et ers theater cuties tht ck these cones ey ee ‘yea Agun the thoy ly to pay an oe sed i does. Page uggs tht tether hs ee ‘ciate on pry line Seana ‘rere of orbs vay Hower sn heya ‘at dae ta dont mah” Silanes ay conceive nctooteccaeaaea ‘Sralncsmmudia hedcay jose fatness pte es Tes Help Hints for Testing Theories “Theory shold lowe ncn: 1a many ofa hoy hypothe posi. Teg cniy sent of» ary’ ype bal pce bs ith yy oe Ay yng Themunber fete hypotension thet olin inathroy Coser be ey Agr complete est would ovate he thurs pee ypothet 31 the theory's explanntory hyptbeses(A= 9.99% and r=» 2B) andtheihybrig combinations (Arend >B} Ths eluce- Fnkthwory competes tal fs testblehypothees. An snl should explore theo all ime and energy perme. 2 lnfer and testes many predictions ofeach hypothesis at pombe. Moet hypotheses make sve testable previsins, 50 "ot be quickly content et wi oe. T nd more, consider ‘vat variance the hypothe profits acoss bth ne and space (that ig acon eon, groupe nett, ovals). Con Sider ls what decision posse (any) predict, an whit ‘pte individual speech and action pedi Praicon frame cervaton you expectto make ifthe theory ‘aval. Thy define expectations aout the incidence, sequen, Jomtien and arectare of penemena Avoid faming agin predictions tat forest simpy that we expect to absarve he theory in operation the theory valid 1 predict wo wil ob serve cause cng Hs eet). Thar the hypothe that ‘ray ll tpt i modi ds ha Bt a ia cy yg ecto ee ah [klar Seai ony applet em SEoemager ly ryt teos ee tes Sa ee Ss ae Se See epee Seah pears es mois Ttrtmaisimterrtconyanonnces aah As ed at eae canna oe acini corre Ss ciafonteeeenede ‘by fly explaining end defending thet pricione, Prdicions can be either genera (he Oost predic & Broad alter) or pec (he thee predicts dict fc ot eter single observations). Gneral prion are inte oh 5 tse tots genera hypothe windows feppertly tan walerbity dive sates twat, sats im late tang shuld inc more tan the share of war) Spe pote ‘ons arenes fm, ad ae wo et both general ats ‘est windows of opportunity and vara doves to x we should es pum have mor aggressively ars wine ‘ opprtnity pene ints favor tn 941) and ypc coplane ons ita window of pportnty dove Japan to wat hae me shoal nd eu of apres dso ser ting sg widow a son for war"), 4 Select data tha epresent, as accu as posi, the do- ‘mal fhe es hen ain lange test meds sect date that "present the universe defined by tested hypothe When eng Tete covering lav’ caual phenomena preset the ase we seek expan’ A spc explanation is lle ony ithe "ale on the independent arse othe ger ths on rich ‘hecxplantion rents egret than za. Even f fsa confirmed ‘se of it eannat explain nance of 8 that ocr whe A Sent "ven lf camomile depesons case wat they cant explain sears thatocarin pelos of preeperty Even capital crue lnpeition cannot expan communist x precast empites Asteroid impacts may cause extinctions, but cannot expan ex Spetone nt omar inthe enc fan impact 5. Are the covering Law's antecedent conitons met i the ‘ae? Theories cannot explain eutcomes im cn tml th cesar anecedent conditions. Dog bis spread rbie ithe dogs rabid tesbyanonabid dog carmot explain a ables se 1 Are the covering laws intervening phenoznena served in ‘he cast Phenomena Ut link the covering’ posted cate nu effect should be evident and appear in appropriate mes and ces hui an steroid espace the dinosaur lion Yur agp ve shout ind evidence of the catastrophic ling Process that a npc woul unleash, For example sere sn2 Gaeta Mt abt of Pt kc ‘is thorie that an impact wou ily praying the labo with olen rock gern fret is thal would dacken the sie wath smoke st out sunlght and fee the eae fs we should find the Sot fom these iin milion yeaa a. ‘men! worlwide. We should ako find evidence of recy lege (contnentszed or even gobl) molten rock shower ands very rapt dying of pecs ‘Thisfourth taps recssry because the fst hse steps are nat Aefitive If we aml tp eran penile hat the cove law that vuppots ur explanation i pbs and the ca at ‘hand among those whee it id at operate We lio shoul tes the explanations withi-case prions a edge agalnt the possiblity that oa faith in he covering law misplaced ard ‘hat he "lain act ae. ortho sos the eter the otis th cas conform ote dead withinese predictions ‘the explanation, the stenge the inference tht Un explanation ‘explain the cee SA a eee area nee Sorin ls aS Bes See acne art tes Sg ine ming wr eye aa os ete Sire feestmraraen some Seem seers pect iichee alae nee ace, ener ieee rene oS ee ee Speen onacemeecnnse ae ee a pts an Te Analysts aratlowed to infer the covering aw that ued the sec explanation of «given eve foc the event le The deals ofthe oven suggest a speci explaation the aay then {eames tha explanation in geneal ene that low tr gue oad database the explanation pasos these tet and the ana Ipslthen reaplis the theory to the pectic case, Th the sing the general Ucry ancl the explaling of «spec case canbe ‘done together and can rapport cach othe Methodology Myths ispaleny seu inuntons thatean bet gare. The fowing a ong em “Evidence ening tod conden oporanc - dence efing tres” Ke Popper thereon ice angoe tht cons aera wena ony leant Sorte my» tery a fr me spat oe traning Tei dn enrol eres th ood Sto There cet be ved sno bee we at ion and et every ron thy ak onde sty tvs ema tha an eimapined pein wl aly co tra ring can mre cymes ry. ce texto that inning tes ocr caning tow rec lfatieor pase many org ttt a ata revit ented pelle ths eu mene te hoy ‘pei dnd ener cot opie itindars eter cette en Sy en Tatcnait wbatticn nypotheig, though Si ak a ea al fogs Seatac eter ihea at rats Pal ee ‘We act reaming the theory to inca the anecdent cone ‘in, dhs namowing the scope ofthe hey him to exclude | the funk et In Popper’ tema we now have new thoory Iowevesallthe ets pase the ld dy ao cocoborie tie | ne leaving tin very tng shape stbith,Thscnfeming ets fell usa grat deal—about the old theory, about is reper placement ar aboxt any late versione Popper's contrary eg | ‘ment stems part from his strange assump tht onc theories | te sated they ae promplly accepted ™ hence evidence in thet favor i unimportant beesuse it merely ranforcs presi ‘ei in the theory. The opp moc en tue most ee ess ace haste prude even ater confining evidence scout | “Theories cannot be faleied Before their seplacement | ‘emenges" Ise Lakatos ar that “thee Iso flies fof | {hooey beoretheemergenceof abet thor and ascation | ‘ant precede the bt theo" This cal soo sweeping | pies oly to theorist fl soe ett tan sre plan story power. We could ain thse tues ual stronger te placement avs, But if testing shows tatty has oe lanatory power weshould pet whether ce not plement ‘hoor it than Many scene programs expe med tal research—edvance by routinely hang thoores again! mal hypotheses and jecting those at fi whether a ot rplce= sent ar ey eee poe eres ameny | SSR eames eer ‘Buln roe ta nes Speiny pcre pees eam cer eee {2th oie ny en” ana edn ch el wag et 6 yates a, a Ts Asking those wh calm to welts theories a explanations to propose paunbleeplacementscan serve aa chick prema ‘lam of ft, Tlscanexpee instances where the efuting, Investigate hed the theory toa wandard tht her wn expla in could not moat. This oggests intr thatthe standart was toohigh nother wondy tat he refer scott oie inthe dita a deve asying evidence agnnst the theory: However fog merit inthis crise a ar ry om grein tha hoo es annotbealstfied except bythe ester sees competing "erin Surely ee can kno whats wrong belo kowng what Iseigh 5. "The evidenoc tha inept a theory should not be reused to st” Thr aguent fen attached to wamings ot fo et ees withthe sre cases fom which they wee ined. It rts on a preference or Bnd testing? The amnion thi ‘tot want infra theory les lla en ietiga tor han used data, hance he fvectigator ing sed das Jee tempted o sample the dat slectvely. ‘und esting is aucefl check on dishonesty bt snot viable a «fwd rue Ts purpose i to prevent scholars fem choosing Cecrborating tte we mtg infin ones. Ba ipo Din-est ues on Soil science ct impose bec ir ‘estigatoreaely ays know something about te ata before ‘he test thet thoores and hasten havea goo ea what as vel show even the exe the dae that rie th is sae eee ‘rues cet eaten oe Snide cee caesar a Biss higee a gamer raat het bra etna tee ee Huse eit areas ace ee Hae eer nl ibed taneous? as 1 Rede Hep Py of Kt Sep.« Gudea os Sus Pe See ence we ned other baros aginst et aging Incing 3 sl enc profesons wide high tndandsol host isthe bes ‘sation ‘ “Donot lc cases on the dependent vai" that do ot scl caves of what you sock to explain for expe wars) without ao choosing eae ofthe contrary (pens), Stade a ‘he se method often repent hi warning Its not ail Selec Siow on the dependent variates appropriate ner any of he connor cnatons 4 can compare conditions in lected cases to known average situation The average testo on elicit well known ot to requre further desapive stad so, we can com Suir np ae se mean pede te Sele caraceeae meat eee eeee teres ‘tam eng to es? Won we est ete ayo yp ee eee oul dea wh he ny ar wh oe ely sa fe pees Gest nga tom Des Sal te en ps Svat coroueaersrans Sea cee Sn eens Sree anomie mean Zen iceman eooae ahh 3, ” SEioimoan poe an Tas pct eda te we orn ec tt Sonam rin mangereesemas TEA gee em ne tay stp ie cpap “tara i tg Si ete on ie caiemigneminactyncnoenae noe Shatin intone tas ben Settee pen SROs cmap aneey hon oem at UES cae Serioitaryenorstiataiasaitesa! cere ee nwt cae ipo taninteertpuenems caters Sho ptaocnecantaraeese Satelite acy st heres Secretion poedion Seater ene Ear Scstetattitypenrameeitnarcnrber ty cateieicrmar acannon ls =a See Seeeeepeaecreesemmaie eer fuemetar ee har elton no een igora ee Met tet Pil Smee Sind beter solution eto ve hone eto scholar who take on the hater tas of staying he lew aber able 6. "Counteiactual analysis can expand the number of beer: ‘ions aval fe Geary esting Janes aaron ages hi gument Counerfatal atement canna pone ite for empirical servations however Thoy can clay am explana- one I cai caused 40 aii ny clan, let explain Image ofa weld absent” They can also help east surface hypotheses brid in the own mld (ete secon “How Can ‘hoses Be Made?” ia his chaper But counterfactual sake ments arent data and cannot plac epi tan they tet ‘Rotate ea re dan ere Tg bein er ‘rer Soiree ha Ee Sei sane ‘veto Ths ater aon bp co ce pe Tana“ nd pay Pe ee We Pan aay al oe,