Yini okufanele ufunde ngokulandelayo

• Isingeniso (isigaba 2.1)

Olunye uhlobo lokubheka lokho akufakiwe kule sahluko yi-ethnography. Ukuze uthole okwengeziwe nge-ethnography kuzikhala zedijithali, bheka i- Boellstorff et al. (2012) , futhi ngokuningi kokuthinta ngezindlela ezixubile zedijithali nezomzimba, bheka Lane (2016) .

• Idatha enkulu (isigaba 2.2)

Ayikho incazelo eyodwa yokuvumelanisa "idatha enkulu," kodwa izincazelo eziningi kubonakala zigxila "ku-3 Vs": ivolumu, ukuhlukahluka, nokuhamba (isib., Japec et al. (2015) ). Bheka De Mauro et al. (2015) ukubuyekezwa kwezincazelo.

Ukufakwa kwami ​​kwedatha yokuphatha kahulumeni esigabeni semininingwane enkulu kuyinto engavamile, nakuba abanye baye benza leli cala, kuhlanganise Legewie (2015) , Connelly et al. (2016) , no- Einav and Levin (2014) . Ukuze uthole okwengeziwe ngokubaluleka kwedatha yokuphatha kahulumeni yokucwaninga, bheka Card et al. (2010) , i- Adminstrative Data Taskforce (2012) , ne- Grusky, Smeeding, and Snipp (2015) .

Ngombono wocwaningo lokuphatha oluvela ngaphakathi ohlelweni lukahulumeni lwezibalo, ikakhulukazi i-US Census Bureau, ubone Jarmin and O'Hara (2016) . Ukuze uthole ukwelashwa ubude bezincwadi zamakhodi wokuphatha e-Statistics Sweden, bheka i- Wallgren and Wallgren (2007) .

Esikhathini esahlukweni, ngaqhathanisa ukuhlola okwenziwe ngendabuko njenge-General Social Survey (GSS) nomthombo wedatha yemithombo yezenhlalo njenge-Twitter. Ukuze uqhathanise ngokucophelela nokucophelela phakathi kokuhlolwa kwendabuko kanye nedatha yezokuxhumana, bheka Schober et al. (2016) .

• Izici ezivamile yedatha enkulu (isigaba 2.3)

Lezi zici ezingu-10 zedatha enkulu ziye zachazwa ngezindlela ezihlukahlukene ngezindlela ezihlukahlukene zababhali. Ukubhala okwangithonya ukucabanga kwami ​​ngalezi zindaba kuhlanganisa Lazer et al. (2009) , Groves (2011) , Howison, Wiggins, and Crowston (2011) , boyd and Crawford (2012) , SJ Taylor (2013) , Mayer-Schönberger and Cukier (2013) , Golder and Macy (2014) , Ruths and Pfeffer (2014) , i- Tufekci (2014) , i- Sampson and Small (2015) , K. Lewis (2015b) , K. Lewis (2015b) Lazer (2015) , i- Horton and Tambe (2015) , i- Japec et al. (2015) , Goldstone and Lupyan (2016) .

Kulo lonke isahluko, ngisebenzise igama elithi traces digital , engicabanga ukuthi alihambisani nhlobo. Elinye igama elidumile lokulandelana kwe-digital yizinyathelo zedijithali (Golder and Macy 2014) , kodwa njengoHal Abelson, uKen Ledeen noHarry Lewis (2008) baveza ukuthi igama elifanele kakhulu cishe yiminwe ye-digital . Uma udala izinyathelo, uyazi okwenzekayo futhi izinyathelo zakho azikwazi ukulandelwa kuwe ngokwakho. Okufanayo akulona iqiniso ngemiphumela yakho ye-digital. Eqinisweni, ushiya umkhondo ngaso sonke isikhathi onolwazi oluncane kakhulu. Futhi, nakuba lezi zimpawu zingenalo igama lakho kuzo, zingase zixhunywe kuwe. Ngamanye amazwi, afana neminwe yeminwe: engabonakali futhi ekhomba ngokwayo.

• Okukhulu (isigaba 2.3.1)

Ukuze uthole ukuthi kungani ama-dataset amakhulu enza izibalo zokuhlolwa kwezibalo, zibone M. Lin, Lucas, and Shmueli (2013) McFarland and McFarland (2015) . Lezi zindaba kufanele ziholele abacwaningi ekubhekiseni okubaluleke kakhulu kunokubaluleka kwezibalo.

Ukuze uthole ukuthi u-Raj Chetty nabalingani bakhe bathola ukufinyelela kanjani kumarekhodi yentela, bheka Mervis (2014) .

Ama-dataset amakhulu angakha futhi nezinkinga zokuxilonga ngokuvamile ezingaphezu kwamakhono ekhomputha eyodwa. Ngakho-ke, abacwaningi abenza ucwaningo kuma dataset amakhulu bavame ukusakaza umsebenzi kumakhompyutha amaningi, inqubo ngezinye izikhathi ebizwa ngokuthi izinhlelo ezifanayo . Ukuze uthole isingeniso sokuhlela okufanayo, ikakhulukazi ulimi olubizwa ngeHadoop, bheka Vo and Silvia (2016) .

• Ukuhlala njalo (isigaba 2.3.2)

Uma ucabangela njalo-ku-data, kubalulekile ukucabangela ukuthi ngabe uqhathanisa nabantu abanjalo ngokuhamba kwesikhathi noma ukuthi uqhathanisa yini iqembu elishintshayo labantu; bheka isibonelo, Diaz et al. (2016) .

• I-non-active (isigaba ½ sec: ayisebenzi ? )

Ibhuku lezincwadi ezinyathelo ezingenzi lutho yi- Webb et al. (1966) . Izibonelo zaleso ncwadi ngaphambi kwesikhathi sedijithali, kodwa zisakhanya. Ukuze uthole izibonelo zabantu abaguqula ukuziphatha kwabo ngenxa yobuningi bokubhekwa, bheka Penney (2016) Brayne (2014) .

Ukuziphendulela kuhlobene eduze nalokho abacwaningi abakubiza ngokufuna imiphumela (Orne 1962; Zizzo 2010) kanye nomphumela we-Hawthorne (Adair 1984; Levitt and List 2011) .

• Akuphelele (isigaba 2.3.4)

Ukuze uthole okwengeziwe kokuxhumanisa amarekhodi, bheka Dunn (1946) Fellegi and Sunter (1969) (umlando) Larsen and Winkler (2014) (okwamanje). Izindlela ezifanayo nazo zakhiwe kumakhompiyutha wesayensi ngaphansi kwamagama anjengokudambululwa kwedatha, ukuhlonza isenzo, ukufaniswa kwegama, ukuhlolwa okuphindwe kabili, nokutholakala kwerekhodi eyimpinda (Elmagarmid, Ipeirotis, and Verykios 2007) . Kukhona futhi izindlela zokulondoloza ubumfihlo zokurekhoda ukuxhuma okungadingi ukudluliselwa kolwazi lokukhomba (Schnell 2013) . I-Facebook iphinde ithuthukise inqubo yokuxhumanisa amarekhodi abo ekuziphatheni kokuvota; lokhu kwenziwa ukuhlola ukuhlolwa engizokutshela ngakho esahlukweni 4 (Bond et al. 2012; Jones et al. 2013) .

Ukuze uthole okwengeziwe Shadish, Cook, and Campbell (2001) ukufaneleka, bheka isahluko 3 Shadish, Cook, and Campbell (2001) .

• Akufinyeleleki (isigaba 2.3.5)

Ukuze uthole okwengeziwe ku-AOL search log debacle, bheka i- Ohm (2010) . Nginikeza iseluleko mayelana nokusebenzisana nezinkampani kanye nohulumeni esahlukweni 4 lapho ngichaza ukuhlolwa. Abalobi abaningi baye babonisa ukukhathazeka ngocwaningo oluthembele kudatha elingenakufinyeleleka, bona Huberman (2012) kanye boyd and Crawford (2012) .

Enye indlela enhle yokusebenza kubacwaningi enyuvesi ukuthola ukufinyelela idatha ukusebenza enkampanini njengoba intern noma umcwaningi evakashele. Ngaphezu olwenza ukufinyelela idatha, le nqubo ngeke aphinde asize umcwaningi ufunde okwengeziwe mayelana nokuthi idatha yadalwa, okuyinto ebaluleke for analysis.

Mayelana nokuthola idatha kahulumeni, Mervis (2014) uxoxa ngendlela uRajar Chetty nabalingani bakhe abathola ngayo ukufinyelela kumarekhodi entela asetshenziselwa ucwaningo lwabo ngokuhamba komphakathi.

• Abangekho abamele (isekhondi: okungewona omele)

Ukuze uthole okwengeziwe emlandweni we "ukumelela" njengombono, bheka Kruskal and Mosteller (1979a) , Kruskal and Mosteller (1979b) , Kruskal and Mosteller (1979c) , Kruskal and Mosteller (1980) .

Izifinyezo zami zomsebenzi we-Snow kanye nomsebenzi weDolli ne-Hill zazifushane. Ukuze uthole okwengeziwe ngomsebenzi we-Snow on cholera, bheka Freedman (1991) . Ukuze uthole okwengeziwe kwi-British Doctors Study bheka i- Doll et al. (2004) Keating (2014) .

Abacwaningi abaningi bayomangala lapho bezwa ukuthi nakuba iDoll no Hill beqoqa idatha kudokotela besifazane nakudokotela abangaphansi kweminyaka engama-35, abavumanga ukusebenzisa le datha ekuhlaziyweni kwabo kokuqala. Njengoba bephikisana ngokuthi: "Njengoba umdlavuza wamaphaphu awunqabile kubantu besifazane namadoda angaphansi kuka-35, izibalo eziwusizo cishe azitholakali kula maqembu iminyaka eminingana ezayo. Kulesi sivumelwano sokuqala sesisindise amadoda eseneminyaka engama-35 nangaphezulu. "U- Rothman, Gallacher, and Hatch (2013) , onesihloko esicasulayo esithi" Kungani ukumelela kufanele kugwenywe, "yenza ingxabano evamile yokubaluleka ngenhloso yokwakha idatha engabonakali.

Ukungaboni kahle kuyinkinga enkulu kubaphenyi nohulumeni abafisa ukwenza izitatimende mayelana nendawo yonke. Lokhu kungaphansi kokukhathazeka kwezinkampani, ezivame ukugxila kubasebenzisi bazo. Ukuze uthole ukuthi imininingwane ye-Statistics Netherlands ibheka kanjani ukungabonakali kwemininingwane enkulu yebhizinisi, bheka i- Buelens et al. (2014) .

Ukuze uthole izibonelo zabacwaningi abonisa ukukhathazeka mayelana nemvelo engeyona K. Lewis (2015b) yemithombo emikhulu yedatha, bheka boyd and Crawford (2012) , K. Lewis (2015b) , Hargittai (2015) .

Ukuze uthole ukuqhathaniswa okuningiliziwe kwemigomo yokuhlola kwezenhlalakahle kanye nokucwaninga ngezifo, bheka Keiding and Louis (2016) .

Ukuze uthole okuningi mayelana nemizamo yokusebenzisa i-Twitter ukwenza ukukhiqizwa okuvela kwesampula mayelana nabavoti, ikakhulukazi icala kusukela kokhetho lwaseJalimane ka-2009, bheka Jungherr (2013) Jungherr (2015) . Ngokulandela umsebenzi Tumasjan et al. (2010) abacwaningi emhlabeni wonke baye basebenzisa izindlela zokuqasha-njengokusebenzisa ukuhlaziywa kwamagama ukuhlukanisa phakathi kokukhuluma okuhle nokubi kwamalungu-ukuze kuthuthukiswe ikhono le-Twitter idatha yokubikezela izinhlobo ezahlukene zokhetho (Gayo-Avello 2013; Jungherr 2015, chap. 7.) . Nansi indlela Huberty (2015) efingqa ngayo imiphumela yale mizamo yokubikezela ukhetho:

"Zonke izindlela ezibikezelayo ezisekelwe emithonjeni yezokuxhumana zihlulekile uma zihambisana nezidingo zokubikezela okhethweni okuqhubeka phambili. Lokhu kuhluleka kubonakala kungenxa yezinto eziyisisekelo zemithombo yezenhlalo, kunokuba kube nezinkinga zokusebenza noma ze-algorithmic. Ngamafuphi, abezindaba bezenhlalakahle abenzi, futhi mhlawumbe ngeke baphinde, banikeze isithombe esisimeme, esingenabulungiswa, esimele sabantu abakhethiweyo; futhi amasampula elula omphakathi wezenhlalo ayinakho idatha eyanele yokulungisa lezi zinkinga ngokuthumela i-hoc. "

Esahlukweni 3, ngizochaza ukulinganisa nokulinganisa ngemininingwane eminingi. Ngisho noma idatha ingabonakali, ngaphansi kwezimo ezithile, ingabalelwa ukukhiqiza izilinganiso ezinhle.

• Ukuqhutshwa (isigaba ½ sec: ukukhukhumeza? )

Ukudonsa kwesistimu kunzima kakhulu ukubona ngaphandle. Noma kunjalo, iphrojekthi ye-MovieLens (exoxwe kabanzi esahlukweni 4) iye yaqhutshwa iminyaka engaphezu kwengu-15 yiqembu labacwaningi bezemfundo. Ngakho-ke, bakwazi ukukopisha nokwabelana ngolwazi mayelana nendlela uhlelo luye lwashintsha ngayo ngokuhamba kwesikhathi nokuthi lokhu kungathinta kanjani ukuhlaziywa (Harper and Konstan 2015) .

Izazi eziningi ziye zagxila ekuthunjweni e-Twitter: Liu, Kliman-Silver, and Mislove (2014) Tufekci (2014) .

Enye indlela yokubhekana nokudonsa abantu ukudala iphaneli yabasebenzisi, okuvumela abacwaningi ukuba bafunde abantu abafanayo ngokuhamba kwesikhathi, bona Diaz et al. (2016) .

• I-algorithm ididekile (isigaba ½ sec: ngokuzenzakalelayo-edidekile? )

Ngaqala ukuzwa igama elithi "algorithmically confounded" elisetshenziswe nguJon Kleinberg enkulumweni, kodwa ngeshwa angikhumbuli ukuthi kuphi inkulumo. Isikhathi sokuqala lapho ngibona leli gama liphrintiwe ku- Anderson et al. (2015) , okuyinto engxoxweni ezithakazelisayo mayelana nokuthi izilungiswa ezisetshenziswe ngamasayithi okuphola zingenza kanjani abacwaningi bakwazi ukusebenzisa idatha kusuka kwala mawebhusayithi ukuze batadishe izintandokazi zomphakathi. Lokhu kukhathazeka kwaphakanyiswa K. Lewis (2015a) ngokuphendula ku- Anderson et al. (2014) .

Ngaphezu kwe-Facebook, i-Twitter iphinde iphakamise abantu kubasebenzisi ukuthi balandele ngokusekelwe emcimbini wokuvalwa kwe-triadic; bheka i- Su, Sharma, and Goel (2016) . Ngakho izinga lokuvalwa kwe-triadic ku-Twitter luyingxube yethonya elithile lomuntu ekuvaliwe kwe-triadic nokunye ukuthambekela kokuhleleka kokukhuthaza ukuvalwa kwe-triadic.

Ukuze uthole okunye mayelana nokusebenza-ikakhulukazi umbono othi ezinye izinkolelo zesayensi yezenhlalo "ama-engesiwo amakhamera" (okungukuthi, abumba umhlaba kunokuba nje achaze) -bheka Mackenzie (2008) .

• Ukungcola (isigaba 2.3.9)

Ama-ejenti wokubala ohulumeni abiza idatha yokuhlanza idatha yokuhlela idatha . De Waal, Puts, and Daas (2014) achaza amasu okuhlela idatha elandelwayo ukuze athole idatha yocwaningo futhi ahlole ukuthi asebenza kanjani emithonjeni emikhulu yedatha, futhi ama- Puts, Daas, and Waal (2015) athile eminye imibono efanayo izilaleli ezivamile.

Ukuze uthole ukwaziswa okuningiliziwe kwe-bots social, bheka Ferrara et al. (2016) . Kwezinye izibonelo zezifundo ezigxile ekutholeni ugaxekile ku-Twitter, bheka Clark et al. (2016) Chu et al. (2012) . Ekugcineni, i- Subrahmanian et al. (2016) uchaza imiphumela yeDarPA Twitter Bot Challenge, ukubambisana okukhulu okwenzelwe ukuqhathanisa izindlela zokuthola ibhola ku-Twitter.

• Izwela (isigaba 2.3.10)

Ohm (2015) ibuyekeza ukucwaninga kwangaphambili emcimbini wolwazi olubucayi futhi inikeza ukuhlola okuningi kwezinto. Izici ezine aziphakamisayo yizikhulu zokulimala, amathuba okulimaza, ukutholakala kobudlelwano obuyimfihlo, nokuthi ngabe ingozi ibonisa ukukhathazeka okukhulu.

• Ukubala izinto (isigaba 2.4.1)

Ukufunda kukaTerber kwamatekisi e-New York kwakusekelwe ekuhloleni kwangaphambili Camerer et al. (1997) esasebenzisa amasampula amathathu ahlukene wephepha lamapheshana ohambo. Lolu cwaningo lwangaphambili lathola ukuthi abashayeli babonakala behlose abaqashile: basebenza kancane ezinsukwini lapho umholo wabo wawuphakeme khona.

Emisebenzini elandelayo, iNkosi nabasebenza nabo baqhubeka behlola ukucwaninga kwe-inthanethi eChina (King, Pan, and Roberts 2014, [@king_how_2016] ) . Ukuze uthole indlela ehlobene yokulinganisa ukucutshungulwa kwe-inthanethi eChina, bheka Bamman, O'Connor, and Smith (2012) . Ukuze uthole izindlela eziningi zezibalo ezifana nalezi ezisetshenziswa King, Pan, and Roberts (2013) ukulinganisa ukuzwa kwezikhala eziyizigidi eziyi-11, bheka i- Hopkins and King (2010) . Ukuze uthole okwengeziwe kokufunda okuqondisiwe, bheka James et al. (2013) (ubuncane obuchwepheshe) Hastie, Tibshirani, and Friedman (2009) (ngaphezulu kwezobuchwepheshe).

• Ukubikezela kanye nokuqokwa kwamanje (isigaba 2.4.2)

Ukubikezela kuyinhlangano enkulu yesayensi yezohwebo (Mayer-Schönberger and Cukier 2013; Provost and Fawcett 2013) . Uhlobo olulodwa lokubikezela okuvame ukwenziwa ngabacwaningi bezenhlalakahle ukubikezelwa kwabantu; bheka, isibonelo, Raftery et al. (2012) .

Izinkombandlela ze-Google Flu akuyona iphrojekthi yokuqala yokusebenzisa idatha yokusesha kwi-currentcast influenza. Eqinisweni, abacwaningi base-United States (Polgreen et al. 2008; Ginsberg et al. 2009) naseSweden (Hulth, Rydevik, and Linde 2009) bathole ukuthi amagama athile okusesha (isib. "Umkhuhlane") abikezela ukubhekwa kwezempilo kazwelonke idatha ngaphambi kokuthi ikhishwe. Ngokulandelayo, amanye amaphrojekthi amaningi aye azame ukusebenzisa idatha ye-digital trace yokuthola ukuhlolwa kwesifo; bheka i- Althouse et al. (2015) ukuze kubuyekezwe.

Ngaphezu kokusebenzisa idatha ye-digital trace ukubikezela imiphumela yezempilo, sekubuye kube nemali enkulu yomsebenzi usebenzisa idatha ye-Twitter ukubikezela imiphumela yemiphumela; izibuyekezo zibona Gayo-Avello (2011) , Gayo-Avello (2013) , Jungherr (2015) (isahluko 7), Huberty (2015) . Ukuchithwa kwezinkomba zezomnotho, okufana nomkhiqizo omningi wasekhaya (GDP), kubuye kuvame kakhulu emabhange asezindaweni eziphakathi, bheka Bańbura et al. (2013) . Ithebula 2.8 liqukethe izibonelo ezimbalwa zezifundo ezisebenzisa uhlobo oluthile lwe-digital trace lokubikezela uhlobo oluthile lomcimbi emhlabeni.

Ekugcineni, uJon Kleinberg kanye nozakwethu (2015) baye babonisa ukuthi izinkinga zokubikezela ziwela emikhakheni emibili ehluke kakhulu futhi ukuthi ososayensi bezenhlalakahle baye bahlose ukugxila kwelinye bese bengayinaki enye. Cabanga ngomunye umenzi wezinqubomgomo, ngizombiza ngokuthi u-Anna, obhekene nesomiso futhi kumele anqume ukuthi uzoqasha um shaman ukwenza umdanso wemvula ukwandisa amathuba emvula. Omunye umenzi wezinqubomgomo, ngizombiza ngokuthi uBetty, kumele anqume ukuthi uthathe isambulela ukusebenza ukuze agweme ukugola endleleni. Bobabili u-Anna noBetty bangenza isinqumo esingcono uma beqonda isimo sezulu, kodwa badinga ukwazi izinto ezahlukene. U-Anna kudingeka aqonde ukuthi ukudansa kwemvula kubangela imvula. UBetty, ngakolunye uhlangothi, akudingeki ukuba aqonde noma yini mayelana nesimo; udinga nje isibikezelo esinembile. Abacwaningi bezenhlalakahle bavame ukugxila ezinkingeni ezinjengalezo ebhekana no-Anna-okuyinto iKleinberg kanye nosebenza nabo ababiza ngokuthi "izinkinga zemigomo yokudansa imvula" -kubandakanya imibuzo yokukhathazeka. Imibuzo efana neyayibhekene noBetty-okuyinto uKleinberg kanye nosebenza nabo ababiza ngokuthi "izinambulela-like" izinkinga zenqubomgomo-ingabalulekile nakakhulu, kodwa ziye zathola ukunakekelwa okuncane kakhulu kubacwaningi bezenhlalakahle.

• Ukulinganisa ukuhlolwa (isigaba 2.4.3)

Iphephabhuku le- PS Political Science lalinomhlangano ohlanganiselwe ngemininingwane enkulu, i-causal inference, ne-theory ehlelekile, Clark and Golder (2015) ngokufingqa umnikelo ngamunye. Iphephandaba i- Proceedings of the National Academy of Sciences yase-United States of America lalinomhlangano ochungechunge mayelana nokucatshangelwa kanye nemininingwane enkulu, kanti Shiffrin (2016) ufingqa umnikelo ngamunye. Ngendlela yokufunda yemishini ezama ukuzitholela ngokuzenzakalelayo ukuhlolwa kwemvelo ngaphakathi kwemithombo emikhulu yedatha, bona i- Jensen et al. (2008) , i- Sharma, Hofman, and Watts (2015) , ne- Sharma, Hofman, and Watts (2016) .

Ngokuhlolwa kwezemvelo, i- Dunning (2012) inikeza ukwelashwa kokuqala, ubude bezincwadi nezibonelo eziningi. Ukuze uthole umbono wokungabaza mayelana nokuhlolwa kwemvelo, bheka i- Rosenzweig and Wolpin (2000) (ezokwezimali) noma Sekhon and Titiunik (2012) (isayensi yezombangazwe). Deaton (2010) Heckman and Urzúa (2010) bathi ukugxila ekuhlolweni kwemvelo kungabangela abacwaningi ukuba bagxile ekuqhathaniseni imiphumela engabalulekile ye-causal; Imbens (2010) lezi Imbens (2010) ngombono onethemba lokubaluleka kokuhlolwa kwemvelo.

Lapho echaza indlela umcwaningi angase ahambe ngayo ekubaleni umphumela wokubhalwa komphumela wokukhonza, ngangichaza inqubo ebizwa ngokuthi i- instrumental variables . Imbens and Rubin (2015) , ezahlukweni 23 no-24, banikeza isingeniso futhi basebenzisa i-lottery ehleliwe njengesibonelo. Umphumela wemisebenzi yezempi kuma-compliers ngezinye izikhathi ubizwa ngokuthi i-complier average causal effect (CAcE) futhi ngezinye izikhathi umphumela wendawo yokwelapha isilinganiso (LATE). Sovey and Green (2011) , u- Angrist and Krueger (2001) , no- Bollen (2012) banikezela ukubuyekezwa kokusetshenziswa kwezinguquko zezinsimbi ezenzweni zesayensi yezombangazwe, ezokwezomnotho, nezomphakathi, kanti Sovey and Green (2011) unikeza "uhlu lokufunda" ukuhlolisisa izifundo ngokusebenzisa iziguquko zezinsimbi.

Kuvela ukuthi i-lottery eyabhalwa ngo-1970 yayingekho, empeleni ihlelwe ngendlela efanele; kwakukhona ukwehlukana okuncane okuvela ku-randomness yangempela (Fienberg 1971) . Berinsky and Chatfield (2015) bathi le Berinsky and Chatfield (2015) encane ayibalulekile kakhulu futhi ixoxisana nokubaluleka Berinsky and Chatfield (2015) .

Ngokuhambisana, bheka i- Stuart (2010) yokubuyekezwa okuhle, ne- Sekhon (2009) ngokubukeza okungathí sina. Ukuze uthole okuningi ngokufanisa njengendlela yokusika, bheka Ho et al. (2007) . Ukuthola umdlalo owodwa ophelele kumuntu ngamunye kuvame ukulukhuni, futhi lokhu kufaka izinkinga eziningi. Okokuqala, uma ukufanelana okuqondile kungatholakali, abacwaningi kudingeka bathathe isinqumo sokukala ibanga phakathi kwamayunithi amabili futhi uma ibanga elinikeziwe lisondele ngokwanele. Kuvela isicatshulwa sesibili uma abacwaningi befuna ukusebenzisa imifino eminingi yocala ngalinye eqenjini lezokwelapha, ngoba lokhu kungaholela ekulinganisweni okuqondile. Zombili lezi zindaba, kanye nabanye, zichazwe ngokuningiliziwe esahlukweni 18 sika- Imbens and Rubin (2015) . Bheka futhi Ingxenye II ye ( ??? ) .

Bheka u- Dehejia and Wahba (1999) ukuze uthole isibonelo lapho izindlela ezifanayo zikwazi ukukhiqiza izilinganiso ezifanayo nalabo ezivela ekuhlolweni okulawulwa ngokungahleliwe. Kodwa, bona i- Arceneaux, Gerber, and Green (2006) ne- Arceneaux, Gerber, and Green (2010) ngezibonelo lapho izindlela ezifanako zahluleka ukukhiqiza uphawu lokuhlola.

Rosenbaum (2015) no- Hernán and Robins (2016) banikeza ezinye izeluleko zokuthola ukuqhathaniswa okuwusizo phakathi kwemithombo enkulu yedatha.