Key:
[ , ] Esahlukweni, mina wawumuhle kakhulu ngokudidiyela babhekane nezinga elikhulu. Nokho, awusebenzi ngaso sonke isikhathi ngcono ikhwalithi nokuqagula. Ukwakha isimo lapho i besho-babhekane nezinga elikhulu kungehlisa izinga nokuqagula. (Ukuze mfana, bheka Thomsen (1973) ).
[ , , ] Design and aqhube inhlolovo non-okungenzeka on Amazon MTurk ukubuza kwezibhamu ( "Uyazi, noma ingabe ubani emndenini wenu, umnikazi isibhamu, isibhamu sohlobo lwe-noma Pistol? Ingabe wena noma omunye umuntu emndenini wakho?") Futhi lekutsatfwa ngayo control gun ( "ucabanga kubalulekile kunokunye-ukuvikela ilungelo baseMelika ukuba ngumnikazi izibhamu, noma ukulawula kwezibhamu?").
[ , , ] Goel kanye nozakwabo (2016) zaphathwa inhlolovo non-okungenzeka-based ehlanganisa 49 imibuzo multiple-choice ukuziphatha nokucabanga sisuselwa General Social Survey (GSS) bese ukhetha izinhlolo-vo ezenziwa Pew Research Center on Amazon MTurk. Babe ulungise for the non-representativeness idatha usebenzisa imodeli-based post-babhekane nezinga elikhulu (uMnu P), futhi uqhathanise nezilinganiso kushintshwe nalabo kulinganiselwa besebenzisa okungenzeka-based GSS / Pew survey. Ukuqhuba lolu cwaningo efanayo phezu MTurk bese uzama ukwenza Figure 2a futhi Figure 2b ngokuqhathanisa tilinganiso yakho olungisiwe ne ezakhishwa imizuliswano yamuva iningi GSS / Pew (Bheka Isithasiselo Table A2 for uhlu lwemibuzo 49).
[ , , ] Izifundo eziningi zisebenzisa self-umbiko izinyathelo mobile idatha yomsebenzi phone. Lokhu siyilungiselelo ezithakazelisayo lapho abacwaningi qhathanisa kokuzilimaza libike ukuziphatha ungene (bheka isib, Boase and Ling (2013) ). izindlela zokuziphatha ezimbili ezivamile ukubuza enizibiza kanye imiyalezo, namaseya amabili ozimele isikhathi esivamile "Izolo" futhi "ngesonto eledlule."
[ , ] Schuman futhi presser (1996) baphikisa ngokuthi oda mbuzo nendaba izinhlobo ezimbili zemali ubudlelwane phakathi imibuzo: ingxenye-ingxenye imibuzo lapho kunemibuzo emibili ezingeni elifanayo of ngqo (isib izilinganiso kwababili kamongameli); futhi imibuzo ingxenye-wonke lapho umbuzo jikelele kanje umbuzo othile more (sib becela "Waneliseke kangakanani ngomsebenzi wakho?" lilandelwa ngelithi "Waneliseke kangakanani ngempilo yakho ngokupheleleyo?").
Baphinde bagcwele ezimbili izinhlobo umbuzo oda umphumela: ukungaguquguquki imiphumela zenzeka lapho izimpendulo embuzweni kamuva balethwa closer (kuka abengeke abe) ukuze abanikwe umbuzo ngaphambili; aqhathanise imiphumela zenzeka lapho kukhona umehluko okukhulu phakathi izimpendulo zemibuzo emibili.
[ , ] Ukwakhela phezu umsebenzi Schuman futhi presser, Moore (2002) uchaza isici esihlukile umbuzo oda nomphumela: okufakiwe kanye subtractive. Nakuba Ngokuphambene bomsebenzi imiphumela kukhiqizwa ngenxa kwamahhotela wabaphenduli 'izinto ezimbili maqondana nomunye, okufakiwe ngamunye kanye nemiphumela subtractive akhiqizwa lapho wabaphenduli zenziwa bazwelane kakhudlwana uhlaka ezinkudlwana sokucutshungulwa imibuzo ebhekiswa. Funda Moore (2002) , khona-ke ukuklama futhi ugijime sokuhlolwa inhlolovo on MTurk ukuze bakhombise okufakiwe noma subtractive imiphumela.
[ , ] Christopher Antoun kanye nozakwabo (2015) benza ucwaningo eqhathanisa amasampula lula etholakala ezahlukene online ukuqasha imithombo emine: MTurk, Craigslist, Google AdWords kanye Facebook. Ukwakha inhlolovo elula futhi athole ababambiqhaza ngokusebenzisa ezahlukene ukuqasha online imithombo okungenani ezimbili (bangaba nobuhlobo naye babe imithombo ehlukene emithonjeni emine asetshenziswe Antoun et al. (2015) ).
[ ] YouGov, i-inthanethi-based ucwaningo emakethe niqinile, baqhuba ngezinhloko inthanethi kwethimba labantu abangaba ngu-800,000 abaphendula UK futhi wasebenzisa uMnu P. ukubikezela umphumela EU Inhlolovo (ie, Brexit) lapho abavoti UK bavote noma ukuhlale e noma ashiye European Union.
A incazelo eningiliziwe imodeli YouGov sika izibalo lapha (https://yougov.co.uk/news/2016/06/21/yougov-referendum-model/). Cishe ukukhuluma, YouGov ngodonga abavoti zibe izinhlobo okusekelwe 2015 okhethweni lwango ivoti choice, ubudala, ticu, ubulili, usuku interview, kanye wabavoti abahlala kulo. Okokuqala, ayesebenzisa idatha eqoqwe kusukela YouGov panelists ukulinganisa, phakathi kwalabo abavotela, inani labantu ngamunye hlobo Umvoti abahlose ukuvota Ikhefu. They balinganisela abeze ngamunye hlobo ukuvota ngokusebenzisa 2015 British Election Study (Bes) post-ukhetho ubuso nobuso inhlolovo, okuyinto sasebenza abeze kusukela imiqulu lokhetho. Ekugcineni, bakala bangaki abantu kukhona ngamunye hlobo ukuvota e labavoti esekelwe Census zakamuva kanye Wonyaka Population Survey (usuphethe ukwaziswa okungamsiza kwalokho kusukela Bes, idatha inhlolovo YouGov ezivela emhlabeni ukhetho jikelele, kanye nokwaziswa on bangaki abantu bavotela for party ngalinye e wokhetho ngamunye).
Ezinsukwini ezintathu ngaphambi komhlangano ivoti, YouGov wabonisa ukuhola ezimbili iphuzu Ikhefu. On the eve zokuvota, i-poll wabonisa close kakhulu ukubiza (49-51 Hlala). I final on-the-day isifundo ababikezela 48/52 esivuna Hlala (https://yougov.co.uk/news/2016/06/23/yougov-day-poll/). Eqinisweni, lesi sibalo ngaphuthelwa umphumela wokugcina (52-48 Ikhefu) by ngemaphesenti lamane.
[ , ] Bhala sekulingisa ukuze sicacise ngamunye amaphutha isethulo Figure 3.1.
[ , ] Ucwaningo of Blumenstock kanye nozakwabo (2015) abathintekayo ekwakheni a machine oyisibonelo yokufunda ukuthi ayengase awasebenzise idatha trace sedijithali ukubikezela izimpendulo inhlolovo. Manje, uya ukuzama into efanayo yedathasethi ehlukene. Kosinski, Stillwell, and Graepel (2013) bathola ukuthi Facebook likes ongabikezela kwezimfanelo ezingabalulekile ngabanye kanye nezimfanelo. Ngokumangalisayo, lezi zibikezelo kungaba olunembile nakakhulu kunalezo kwabangani nozakwabo (Youyou, Kosinski, and Stillwell 2015) .
[ ] Toole et al. (2015) Ukusetshenziswa call imininingwane kumarekhodi (CDRs) Movie kuya ukubikezela aggregate ukutlhogeka kwemisebenzi izitayela.