Yintoni oza kufunda ngokulandelayo

• Intshayelelo (icandelo 2.1)

Olunye uhlobo lokubheka okungafakiwe kule sahluko lu-ethnography. Ukufumana okungaphezulu kwi-ethnografi kwiindawo zedijithali, bona Boellstorff et al. (2012) , kunye nokunye okungakumbi kwi-ethnografi kwiindawo ezidibeneyo zedijithali nezizimele, jonga Lane (2016) .

• Idatha enkulu (icandelo 2.2)

Akukho ngcaciso enye "yedatha enkulu", kodwa iinguqu ezininzi zibonakala zijolise kwi "3 Vs": umthamo, udidi, kunye nokuhamba ( Japec et al. (2015) ., Japec et al. (2015) ). Jonga De Mauro et al. (2015) ukuhlaziywa kwengcaciso.

Ukubandakanywa kwam kwedatha yolawulo lworhulumente kwinqanaba leenkcukacha ezinkulu kunqabileyo, nangona abanye baye benza le meko, kuquka Legewie (2015) , Connelly et al. (2016) , kunye no- Einav and Levin (2014) . Ngolunye ulwazi malunga nexabiso lolawulo loorhulumente likarhulumente, khangela Card et al. (2010) , Adminstrative Data Taskforce (2012) , kunye Grusky, Smeeding, and Snipp (2015) .

Ngombono wolawulo lophando oluvela ngaphakathi kwenkqubo yoluhlu lwekarhulumente, ngakumbi i-US Census Bureau, bona Jarmin and O'Hara (2016) . Ngonyango lwangexesha elide lweengxelo zengxelo zolawulo kwiStatistics Sweden, bona Wallgren and Wallgren (2007) .

Kwisahluko, ndayifanisa ngokufutshane uphando lwendabuko olunjenge-General Social Survey (GSS) kunye nomthombo wolwazi weendaba zoluntu ezifana ne-Twitter. Ukuqhathaniswa ngokucokisekileyo kunye ngokucophelela phakathi kweengxelo zendalo kunye nolwazi lweendaba zoluntu, bona Schober et al. (2016) .

• Impawu eziqhelekileyo zeenkcukacha ezinkulu (icandelo 2.3)

Ezi nkalo ezili-10 zolwazi olukhulu zichazwe kwiindlela ezahlukeneyo ezahlukeneyo ngabalobi abahlukeneyo. Ukubhala okubangela ukuba ndicinga ngale miba kubandakanya Lazer et al. (2009) , i- Groves (2011) , i- Howison, Wiggins, and Crowston (2011) , i- boyd and Crawford (2012) , SJ Taylor (2013) , Mayer-Schönberger and Cukier (2013) , Golder and Macy (2014) , Ruths and Pfeffer (2014) , i- Tufekci (2014) , Sampson and Small (2015) , K. Lewis (2015b) , Lazer (2015) , Horton and Tambe (2015) , Japec et al. (2015) , kunye Goldstone and Lupyan (2016) .

Kulo lonke isahluko, ndisebenzise i-term traces yedijithali , endicinga ukuba ayihambisani nanye. Elinye igama elidumiweyo lokulandelelaniswa kwedijithali yimizila yedijithali (Golder and Macy 2014) , kodwa njengoHal Abelson, uKen Ledeen kunye noHarry Lewis (2008) ichaza ukuba igama elifanelekileyo ngakumbi liyiminwe yedijithali . Xa udala unyawo, uyazi oko kwenzekayo kwaye unyawo lwakho alukwazi ukulandelwa ngokubanzi kuwe. Okufanayo akunjalo neendlela zakho zedijithali. Enyanisweni, ushiya yonke indawo malunga nolwazi oluncinane. Kwaye, nangona le mizekelo ingabi negama lakho kubo, iyakwazi ukudibaniswa kuwe. Ngamanye amazwi, baninzi njengeminwe yeminwe: engabonakaliyo kunye nokuzibonakalisa ngabanye.

• Okukhulu (icandelo 2.3.1)

Ukufumana okunye ukuba kutheni iifasethi ezinkulu zibonelela ngeengxaki zeematriki, bona M. Lin, Lucas, and Shmueli (2013) McFarland and McFarland (2015) . Le micimbi ibangela ukuba abaphandi bajolise ngokubaluleka kokubaluleka kunokubaluleka kwamanani.

Ngezinye iinkcukacha malunga nendlela uRajar Chetty kunye nabalingane bakhe bafumane ngayo ukufikelela kwiirekhodi zerhafu, bona Mervis (2014) .

Iifasethi ezinkulu ziyakwazi ukudala iingxaki zoqhelwano eziqhelekileyo ezingaphezu kwamandla ekhompyutha enye. Ngako oko, abaphandi abenza iingcamango kwiidasta ezinkulu baninzi basasaza umsebenzi kwiikhomputha ezininzi, inkqubo ebizwa ngezinye iinkqubo ezibizwa ngokuba yinkqubo efanayo . Ukwenza isingeniso kwiprogram efanayo, ngokukhethekileyo ulwimi oluthiwa yiHadoop, bona Vo and Silvia (2016) .

• Kuhlala njalo (icandelo 2.3.2)

Xa uqwalasela njalo-kwi-data, kubalulekile ukuqwalasela ukuba uthelekisa abantu abafanayo ngexesha elide okanye nokuba uthelekisa iqela elitshintshayo labantu; ubone umzekelo, Diaz et al. (2016) .

• Ukungasebenzi (icandelo še sec: ayilisebenzisanga? )

Incwadi yeklasi kumanyathelo angasebenziyo Webb et al. (1966) . Imimiselo yale ncwadi ngaphambili kwexesha ledijithali, kodwa isakhanya. Imimiselo yabantu abaguqula indlela yabo yokuziphatha ngenxa yobuninzi bokubekwa kweliso, bona Penney (2016) kunye Brayne (2014) .

I-Reactivity ihambelana ngokufutshane nento abaphandi ababiza ngayo iziphumo zemfuno (Orne 1962; Zizzo 2010) kunye nomphumo we-Hawthorne (Adair 1984; Levitt and List 2011) .

• Ukungagqibekanga (icandelo 2.3.4)

Ukufumana okungakumbi kwi-link connection, bona Dunn (1946) kunye Fellegi and Sunter (1969) (ezembali) kunye Larsen and Winkler (2014) (namhlanje). Iindlela ezifanayo ziye zaphuhliswa kwi-computer yesayensi phantsi kwamagama afana nokudityaniswa kwedatha, ukuchongwa kwimeko, ukulingana kwamagama, ukufunyanwa kabili, kunye nokufunyanwa kwerekhodi (Elmagarmid, Ipeirotis, and Verykios 2007) . Kukho iindlela zokugcina ubumfihlo-bucala zokurekhoda ukudibanisa ezingafuneki ukuhanjiswa kolwazi (Schnell 2013) . I-Facebook iphinde iphuhlise inkqubo yokudibanisa iirekodi zabo kwiindlela zokuvota; Oku kwenziwa ukuba kuhlolwe uvavanyo endiza kukuxelela ngalo kwisahluko 4 (Bond et al. 2012; Jones et al. 2013) .

Ukufumana okungakumbi ekwakheni ukufaneleka, funda isahluko 3 Shadish, Cook, and Campbell (2001) .

• Ukungenakwenzeka (icandelo 2.3.5)

Ukufumana okungakumbi kwi-AOL search log debacle, khangela i- Ohm (2010) . Ndinika iingcebiso malunga nokusebenzisana neenkampani kunye noorhulumente kwisahluko 4 xa ndichaza iimvavanyo. Inani labalobi babonise ukukhathazeka malunga nophando oluthembele kwiinkcukacha ezingenakufikeleleka, bona Huberman (2012) kunye ne- boyd and Crawford (2012) .

Enye indlela elungileyo kubaphandi eyunivesithi ukuze afumane ufikelelo data kukuba ukusebenza ngexesha inkampani njengoko osafunda kwiziko okanye umphandi ukutyelela. Ukongezelela yokwenza ukufikelela data, le nkqubo iya kukunceda ukuba umphandi ukufunda okungakumbi ngendlela wadalwa idata, nto leyo ibalulekileyo uhlalutyo.

Ngokumalunga nokufikelela kwiinkcukacha zikaRhulumente, Mervis (2014) uxoxa ngendlela uRajer Chetty kunye nabalingane bakhe bafumane ngayo ukufikelela kwiirekhodi zerhafu ezisetyenziselwa uphando lwabo ekuhambeni kwentlalo.

• Abangabameli (umzuzwana: ongekho ummeli)

Ukufumana okungakumbi kwimbali ye "ukumela" njengengqiqo, bona Kruskal and Mosteller (1979a) , Kruskal and Mosteller (1979b) , Kruskal and Mosteller (1979b) , Kruskal and Mosteller (1979c) Kruskal and Mosteller (1980) .

Isishwankathelo sam somsebenzi weSiqhankqalazo kunye nomsebenzi weDolli kunye nesiHluma sasifushane. Ngolunye ulwazi ngomsebenzi we-Snow kwiKholera, funda i- Freedman (1991) . Ukufumana okungakumbi kwi-Doctors yaseBrithani, sifunda Doll et al. (2004) kunye Keating (2014) .

Abaphandi abaninzi baya kumangaliswa xa befunda ukuba nangona iDoll kunye neHolde beqokelele idatha evela oogqirha besifazane kunye noogqirha abangaphantsi kwe-35, abavumanga ngokusebenzisa le datha kwi-analysis yabo yokuqala. Njengoko babethi: "Kuba umdlavuza wamaphaphu awunqabile kubafazi kunye namadoda angaphantsi kwama-35, amanani aluncedo ayinakufumaneka kule maqela kwiminyaka ezayo. Ngaloo ngxelo yokuqala sesigxininise amehlo ethu kumadoda aneminyaka engama-35 nangaphezulu. " Rothman, Gallacher, and Hatch (2013) , enesihloko esichukumisayo esithi" Kutheni ukumela ukuphephelwa, "yenza ingxabano epheleleyo malunga nexabiso ngokudala ngenjongo yedatha engabonakaliyo.

Ukungabonakali kuyinkalo enkulu kubaphandi noorhulumente abanqwenela ukwenza ingxelo malunga nabantu bonke. Oku kuncinci kwenkxalabo yeenkampani, ezijoliswe kubasebenzisi bazo. Ngezinye iinkcukacha malunga nendlela i-Statistics yase-Netherlands ejonga ngayo imbambano yokungabonakali kweenkcukacha ezinkulu zoshishino, bona Buelens et al. (2014) .

Ngokomzekelo wabaphandi abavakalisa inkxalabo malunga nomntu ongabonakaliyo wemithombo yolwazi, bona i- boyd and Crawford (2012) , K. Lewis (2015b) , kunye Hargittai (2015) .

Ukuqhelanisa ngokubanzi iinkalo zophando zoluntu kunye nophando lwe-epidemiological, khangela Keiding and Louis (2016) .

Ukufumana okungakumbi kwiimvavanyo zokusebenzisa i-Twitter ukwenza i-generalizations malunga nabavoti, ngokukodwa kwimeko yonyulo Jungherr (2013) -2009, Jungherr (2013) Jungherr (2015) . Emva komsebenzi Tumasjan et al. (2010) abaphengululi behlabathi lonke baye basebenzisa iindlela zokuqhubela phambili-njengokusebenzisa uhlalutyo lwamazwi ukuhlukanisa phakathi kokuthetha okulungileyo kunye nokungalunganga kwamacandelo-ukuze kuphuculwe ikhono le-Twitter yedatha ukuxela iintlobo ezahlukeneyo zonyulo (Gayo-Avello 2013; Jungherr 2015, chap. 7.) . Nantsi indlela Huberty (2015) efingqa ngayo iziphumo zale mizamo yokuqikelela ukhetho:

"Zonke iindlela ezichazwayo ezisekelwe kwiimidiya zentlalo ziye zahluleka xa zixhomekeke kwiimfuno zecala lokujonga ukhetho lokujonga phambili. Ezi ntsilelo zibonakala zivela ngenxa yeempawu eziphambili zeendaba zentlalo, kunokuba kubekho ubunzima beendlela okanye i-algorithmic. Ngamafutshane, amajelo aseburhulumenteni ayenayo, kwaye mhlawumbi akayi kuyenza, anikela umfanekiso ozinzile, ongabonakaliyo, ummeleli we-electorate; kunye neesampuli ze-social media abanakho ulwazi olwaneleyo ukulungisa ezi ngxaki.

Kwisahluko 3, ndiza kuchaza isampuli kunye nokuqikelela kwiinkcukacha ezininzi. Nangona idatha ingabonakali, phantsi kweemeko ezithile, zinokulinganiswa ukuvelisa ukuqikelelwa okulungileyo.

• Ukukhutshwa (icandelo še sec: ukukhupha? )

Ukukhutshwa komnto kunzima kakhulu ukubona ngaphandle. Nangona kunjalo, iphrojekthi ye-MovieLens (exoxwa ngakumbi kwisahluko 4) iqhutywe iminyaka engaphezu kwe-15 yiqela lophando lwezemfundo. Ngaloo ndlela, baye bakwazi ukubhalwa kunye nokwabelana ngolwazi malunga nendlela inkqubo eye yaguquka ngayo ngexesha kunye nendlela oku kuchaphazela ngayo uhlalutyo (Harper and Konstan 2015) .

Inani labaphengululi lijolise ekuqhubeni kwi-Twitter: Liu, Kliman-Silver, and Mislove (2014) kunye Tufekci (2014) .

Enye indlela yokujongana nokuqhutyelwa kwabantu kubumba iphaneli yabasebenzisi, okuvumela abaphandi ukuba bafunde abantu abafanayo ngexesha elide, bona Diaz et al. (2016) .

• Ukhohlakele ngokugqithisileyo (icandelo ¿sec: ulungelelwano-udidekile? )

Ndaqala ukuva igama elithi "ulungelelwaniso olusisigxina" elisetyenziswe nguJon Kleinberg entetho, kodwa ngelanga andiyikhumbuli nini okanye kuphi na intetho. Ngethuba lokuqala ndibonayo igama eliprintiweyo Anderson et al. (2015) , enengxubusho enomdla malunga nendlela i-algorithm ezisetyenziselwa ngayo iziza ezinokuthandana zingenza ukuba abaphandi bakwazi ukusebenzisa idatha kulezizabhsayithi ukuze bafunde ukukhethwa kwentlalo. Le nkxalabo yaphakanyiswa K. Lewis (2015a) ngokuphendula Anderson et al. (2014) .

Ukongeza kwi-Facebook, i-Twitter iphinda icebise abantu kubasebenzisi ukuba balandele ngokusekelwe kwingcamango yokuvalwa kwe-triadic; ubona Su, Sharma, and Goel (2016) . Ngoko inqanaba lokuvalwa kwe-triadic kwi-Twitter kukudibanisa ukutyekela komntu ekuvaleni okungaqhelekanga kunye nolunye uhlobo lokuguqulwa kwe-algorithm.

Ukufumana okungakumbi malunga nokusebenza-ngokukodwa ingcamango yokuba ezinye iinkolelo zenzululwazi zentlalo "iinjini ezingekho ikhamera" (oko kukuthi, ziyilo ihlabathi kunokuba zichaze nje) -bona Mackenzie (2008) .

• Ezingcolileyo (icandelo 2.3.9)

Ii-arhente zorhulumente zedata zibiza umnxeba wokucoca idatha yokuhlela idatha . De Waal, Puts, and Daas (2014) ichaza iindlela zobuchule bokuhlela idatha eziphuhliswe kwiinkcukacha zophando kwaye zihlolisise ubungakanani Puts, Daas, and Waal (2015) kwimithombo yolwazi emikhulu, kunye Puts, Daas, and Waal (2015) enika ezinye iimbono ezifanayo abaphulaphuli abaqhelekileyo.

Ukujonga ngokubanzi kwi-bots bots, bona Ferrara et al. (2016) . Kweminye imizekelo yezifundo ezijoliswe ekufumaneni ugaxekile kwi-Twitter, khangela Clark et al. (2016) kunye Chu et al. (2012) . Ekugqibeleni, i- Subrahmanian et al. (2016) ichaza iziphumo zeDarPA Twitter Bot Challenge, intsebenziswano yobuninzi eyenzelwe ukuqhathanisa iindlela zokufumana ibhola kwi-Twitter.

• Efanelekileyo (icandelo 2.3.10)

Ohm (2015) ukuphonononga uphando kwangaphambili malunga nengcamango yolwazi olubukhali kwaye unikeze uvavanyo lwezinto ezininzi. Izinto ezine ezicebisayo zibukhulu bengozi, ubunokwenzeka bengozi, ubukho bobudlelwane obuyimfihlo, nokuba ingaba ingozi ibonisa ukukhathazeka okukhulu.

• Ukubala izinto (icandelo 2.4.1)

Ukufunda kuka-Farber kweetekisi e-New York kwakusekelwe kwisifundo sangaphambili Camerer et al. (1997) esetyenziselwa iisampuli ezintathu zokuhamba ezifanelekileyo zephepha lamaphepha. Olu phofu lwakusasa lufumene ukuba abaqhubi babonakala bejolise ekufumaneni abaqashiweyo: basebenze ngaphantsi kweentsuku apho umvuzo wabo uphezulu.

Kwimisebenzi elandelayo, uKumkani kunye nabalingane baye bahlola ngokubanzi ukucelwa kwe-intanethi eChina (King, Pan, and Roberts 2014, [@king_how_2016] ) . Bamman, O'Connor, and Smith (2012) intanethi eChina, bona Bamman, O'Connor, and Smith (2012) . Ngezinye iinkcukacha kwiindlela zokubala ezifana neyasetyenziswa King, Pan, and Roberts (2013) ukuqikelela ukuvakalelwa kwezikhundla ezigidi ezili-11, bona Hopkins and King (2010) . Ukufumana okungakumbi kwi-learning supervised, bona James et al. (2013) (ubuchwepheshe obungaphantsi) kunye Hastie, Tibshirani, and Friedman (2009) (ubuchwepheshe obungaphezulu).

• Ukuxelwa kwangaphambili kunye nokuhlaselwa (icandelo 2.4.2)

Ukubikezela inxalenye enkulu yesayensi yezentengiselwano (Mayer-Schönberger and Cukier 2013; Provost and Fawcett 2013) . Olunye uhlobo lokubikezela oluqhelekileyo lwenziwa ngabaphengululi bezenhlalakahle lubangelwa ngabantu; sibone, umzekelo, Raftery et al. (2012) .

Iintlobo zeGoogle Flu yayingeyona yiprojekthi yokuqala yokusebenzisa idatha yokukhangela kwi-influenced current influenza. Enyanisweni, abaphandi e-United States (Polgreen et al. 2008; Ginsberg et al. 2009) kunye neSweden (Hulth, Rydevik, and Linde 2009) baye bafumanisa ukuba amagama athile okucinga (umz. idatha ngaphambi kokuba ikhishwe. Ngokugqithisileyo, ezininzi iiprojekthi ziye zazama ukusebenzisa idatha yedijithali yokubona ukuhlolwa kwezilwanyana; sibone Althouse et al. (2015) ukuhlaziywa.

Ukongezelela ekusebenziseni idatha yedijithali yokubalisa iziphumo zempilo, kuye kwaba nomlinganiselo omkhulu womsebenzi usebenzisa idatha ye-Twitter ukuxela kwangaphambili iziphumo zowunyulo; ukuphonononga kubone Gayo-Avello (2011) , Gayo-Avello (2013) , Jungherr (2015) (isahluko 7), kunye Huberty (2015) . Ukuhlaselwa kwezibonakaliso zezoqoqosho, njengemveliso emveliso yasekhaya (GDP), nayo iyaqheleka kwiibhanki eziphambili, bona Bańbura et al. (2013) . Itheyibhile 2.8 ifaka imizekelo embalwa yezifundo ezisebenzisa uhlobo oluthile lwe-digital trace ukuqikelela uhlobo oluthile lomcimbi kwihlabathi.

Ekugqibeleni, uJon Kleinberg kunye noogxa (2015) baye bachaza ukuba iingxaki ezibikezelayo ziwela kwiindidi ezimbini ezihlukeneyo kwaye iintlobo zenzululwazi zentlalo zijonge ukugxila kwelinye kwaye zinyamekele enye. Cinga ngomnye umenzi womgaqo-nkqubo, ndiza kumthiya u-Anna, ojongene nesomiso kwaye kufuneka anqume ukuba aqeshe i-shaman ukwenza umdaniso wemvula ukwandisa amathuba emvula. Omnye umenzi womgaqo-nkqubo, ndiza kumthiya uBetty, kufuneka anqume ukuba uthathe i -ambule ukuba asebenze ukuze aphephe ukuhlamba endleleni. Bobabini uAnna noBetty bangenza isigqibo esingcono xa baqonda imozulu, kodwa bafuna ukwazi izinto ezahlukeneyo. U-Anna kufuneka aqonde ukuba imdaniso yemvula ibangela imvula. U-Betty, ngakolunye uhlangothi, akadingeki ukuba aqonde nantoni na ngokuphathelele imeko; Ufuna nje ukubikezela ngokuchanekileyo. Abaphandi bezentlalo bahlala begxila kwiingxaki ezifana no-Anna-oko uKleinberg kunye nabo bahamba nabo bathi "iingxaki zomgaqo-nkqubo wokudanisa imvula" -kuba zibandakanya imibuzo yecala. Imibuzo enjengaleyo ebhekene nayo noBetty-e-Kleinberg kunye noogxa bayo bathi "iingxaki ze-umbrella-like" zingabalulekanga kakhulu, kodwa zifumene ingqalelo engakumbi kubaphandi bezenhlalakahle.

• Uvavanyo lokulinganisela (icandelo 2.4.3)

Iphepha leNzululwazi yezoPolitiko ye-PS luneenkomfa yenkcazelo enkulu, i-causal inference, kunye ne-theory yaseburhulumenteni, kunye no- Clark and Golder (2015) isishwankathela inxaxheba nganye. Iphephabhali Inkcazo yeSizwe seSizwe seSayensi e-United States yaseMelika ine-symposium malunga neengxelo kunye nolwazi olukhulu, kwaye Shiffrin (2016) ifingqa intsingiselo nganye. Ukufumana iindlela zokufunda ngomatshini ezama ukuzenzekela ngokuzenzekelayo iimvavanyo zendalo ngaphakathi kwemithombo yolwazi emikhulu, Jensen et al. (2008) , i- Sharma, Hofman, and Watts (2015) , kunye ne- Sharma, Hofman, and Watts (2016) .

Ngokwezilingo zemvelo, i- Dunning (2012) inikeza unyango olusenkulumeni, ubude benemizekelo emininzi. Ngombono ongathembekiyo wovavanyo lwendalo, bona Rosenzweig and Wolpin (2000) ( Sekhon and Titiunik (2012) ) okanye Sekhon and Titiunik (2012) (isayensi yezopolitiko). Deaton (2010) kunye Heckman and Urzúa (2010) bathi ukugxila kwizilingo zemvelo kungakhokelela abaphandi ukuba bagxile ekuqikeleleni iziphumo ezingabalulekanga; Imbens (2010) arguments with optimistic view of the value of experiments.

Xa kuchaza indlela umphandi angahamba ngayo ukusuka ekuqikeleleni umphumo wokubhalwa kwiphumo lokukhonza, ndandichaza ubuchule obubizwa ngokuba yiintlobo eziguqukayo . Imbens and Rubin (2015) , kwizahluko zabo ezingama-23 no-24, banikezela isingeniso kwaye basebenzise i-lottery ebonwayo njengomzekelo. Iimpembelelo kwinkonzo yemikhosi kwiinkampani ngamanye amaxesha zibizwa ngokuba yi-compress average causal effect (CACE) kwaye ngamanye amaxesha umphumo wonyango wendawo yokwelapha (LATE). Sovey and Green (2011) , i- Angrist and Krueger (2001) , kunye ne- Bollen (2012) inikezela ukuhlaziywa kokusetyenziswa kwezinto ezibonakalayo kwisayensi yezopolitiko, ezoqoqosho, kunye noluntu, kunye no- Sovey and Green (2011) unikeza "uluhlu lokuhlola" ukuphonononga uphando ngokusetyenziswa kwezixhobo.

Kubonakala ukuba i-lottery yango-1970 yayingekho, ngokwenene ngokulandeleleneyo; kwakukho ukuphambuka okuncinci kwizinto ezingaqhelekanga (Fienberg 1971) . Berinsky and Chatfield (2015) bathi le mpazamo encinci ayibalulekanga kakhulu kwaye ixoxe ngokubaluleka kokulandelelaniswa ngokufanelekileyo.

Ngokubhekiselele, funda i- Stuart (2010) yokuphononongwa ngokuthe tye, kunye no- Sekhon (2009) ukuphononongwa kokungathembeki. Ukufumana okuninzi ngokuhambelana njengoluhlobo lwesitambo, bona Ho et al. (2007) . Ukufumana umdlalo omnye opheleleyo kumntu ngamnye kudla nzima, kwaye oku kufaka inani lezinto ezinzima. Okokuqala, xa iiminganiso ezikhoyo zingekho, abaphandi kufuneka banqume indlela yokulinganisa umgama phakathi kweeyunithi ezimbini kwaye ukuba umgama owenziwe uphelele ngokukhawuleza. Kuvela ubunzima besibini ukuba abaphandi bafuna ukusebenzisa imilinganiselo emininzi kwimeko nganye kwiqela lonyango, kuba oku kungakhokelela ekuqikeleleni okuchanekileyo. Zomibini le micimbi, kunye nabanye, ichazwe ngokucacileyo kwisahluko 18 ka- Imbens and Rubin (2015) . Bona kwakhona iNgxenye II ye ( ??? ) .

Jonga Dehejia and Wahba (1999) ngomzekelo apho iindlela ezihambelana nazo zaziyokwazi ukuvelisa ukuqikelela okufana nezo zivela kuvavanyo olulawulwa ngokungahleliwe. Kodwa, khangela i- Arceneaux, Gerber, and Green (2006) kunye ne- Arceneaux, Gerber, and Green (2010) ngemimiselo apho iindlela ezihambelanayo zahluleka ukuvelisa ibhenkmark yokulinga.

Rosenbaum (2015) kunye Hernán and Robins (2016) banikeza amanye amacebiso ngokufumanisa ukuthelekiswa okuxhamlayo kwimithombo enkulu yedatha.