4.6.1 Create sifili m kudin data

The kewayawa don guje manyan gwaje-gwajen da aka tuki your m kudin sifili. Mafi hanyoyin da za a yi wannan ne aiki da kuma zayyana m gwaje-gwajen.

Digital gwaje-gwajen da za a iya cika fuska daban-daban kudin Tsarin da wannan sa masu bincike gudu gwaje-gwajen da suke ba zai yiwu ba a baya. More musamman, gwaje-gwajen kullum da biyu main iri halin kaka: gyarawa halin kaka da kuma m halin kaka. Kafaffen halin kaka ne halin kaka cewa ba su canja dangane da yadda mutane da yawa mahalarta kana da. Alal misali, a cikin wani Lab gwaji, gyarawa halin kaka zai yi da kudin hayar da sararin samaniya da kuma sayen furniture. M halin kaka, a daya hannun, canji dangane da yadda mutane da yawa mahalarta kana da. Alal misali, a cikin wani Lab gwaji, m halin kaka zai zo daga biyan ma'aikata da kuma mahalarta. A general, analog gwaje-gwajen da low gyarawa da kuma halin kaka high m halin kaka, da digital gwaje-gwajen da high gyarawa halin kaka da kuma low m halin kaka (Figure 4.18). Tare da ya dace zane, za ka iya fitar da m kudin your gwaji duk hanyar zuwa sifili, kuma wannan zai iya haifar da m bincike damar.

Figure 4.18: Schematic na kudin Tsarin cikin analog da dijital gwaje-gwajen. A general, analog gwaje-gwajen da low gyarawa da kuma halin kaka high m halin kaka, alhãli kuwa digital gwaje-gwajen da high gyarawa halin kaka da kuma low m halin kaka. A daban-daban kudin Tsarin nufin cewa digital gwajen iya gudu a sikelin da cewa ba zai yiwu tare da analog gwaje-gwajen.

Akwai biyu main abubuwa na m cost-biya wa ma'aikata da kuma biya wa mahalarta-da kowane daga cikin wadannan za a iya kore wa sifili amfani da daban-daban dabaru. Biya ga ma'aikata kara daga aikin da bincike mataimakansa kada a jawo ra'ayinsu mahalarta, haihuwa jiyya, da kuma aunawa sakamakon. Alal misali, analog filin gwaji na Schultz da kuma abokan aiki (2007) a kan zamantakewa norms da wutar lantarki da ake bukata amfani bincike mataimakansa tafiya zuwa kowane gida ya sadar da magani da kuma karanta lantarki mita (Figure 4.3). Dukan wannan} o} arin da bincike mataimakansa nufin cewa ƙara wani sabon iyali da binciken dã kara wa cost. A daya hannun, domin digital filin gwaji na Restivo kuma van de Rijt (2012) a kan lada a Wikipedia, masu bincike zai iya ƙara ƙarin ƴan takara a kusan wani farashi. A general dabarun domin rage m administrative halin kaka ne ya maye gurbin dan Adam aiki (wanda yake shi ne m) da kwamfuta aiki (wanda yake shi ne cheap). Wajen, za ka iya ka tambayi kanka: iya wannan gwaji gudu yayin da kowa da kowa a kan bincike tawagar yana barci? Idan amsar ita a, ka zo da wani babban aiki da aiki da kai.

Na biyu main irin m kudin ne biya wa mahalarta. Wasu masu bincike sun yi amfani da Amazon Mechanical Turk da sauran online aiki kasuwanni don rage biya da ake bukata domin mahalarta. Don fitar da m halin kaka dukan hanyar sifiri, duk da haka, wani daban-daban m ake bukata. Ga wani dogon lokaci, masu bincike sun tsara gwaje-gwajen da aka haka] in da suka yi ya biya mutane su shiga. Amma, abin da idan ka iya ƙirƙirar wani gwaji cewa mutane suna so su kasance a cikin? Wannan na iya sauti m debo, amma zan ba ka da wani misali a kasa daga kaina aiki, kuma akwai fiye da misalai a Table 4.4. Ka lura da cewa wannan tsarin kula da zayyana m gwajen Makamantan bayanai da wasu daga cikin jigogi a Babi na 3 game da zayyana mafi m safiyo da kuma a Babi na 5 game da zane na taro haɗin gwiwar. Saboda haka, ina ganin cewa yar jin dãɗin-da abin da zai ma a kira mai amfani kwarewa-za su kasance da samun muhimmanci na bincike zane a digital shekaru.

Idan kana so ka ƙirƙiri sifili m halin kaka-gwajen da za ku ji so don tabbatar da cewa duk abin da yake da cikakken sarrafa kansa da kuma cewa mahalarta ba su bukatar wani biya. Domin nuna yadda wannan mai yiwuwa ne, zan bayyana ta dissertation bincike kan nasarar da gazawar da al'adu da kayayyakin. Wannan misali ya nuna cewa sifili m kudin data ne ba kawai game da yin abubuwan da rahusa. Maimakon haka, shi ne game da kunna gwaje-gwajen da cewa ba zai yiwu in ba haka ba.

My dissertation aka m da m yanayin nasara ga al'adu da kayayyakin. Hit songs, m sayar da littattafai, da kuma blockbuster movies suna da yawa, fiye da nasara fiye da talakawan. Saboda wannan, da kasuwanni na wadannan kayayyakin sukan da ake kira "lashe-sha-duka" kasuwanni. Duk da haka, a lokaci guda, wanda musamman song, littafin, ko movie za ta zama m ne mai wuce yarda unpredictable. The screenwriter William Goldman (1989) elegantly summed up kuri'a na ilimi da bincike da cewa, a lokacin da ta je tsinkaya nasara, "babu wanda ya san wani abu." The unpredictability na lashe-sha-duka kasuwanni sanya ni mamaki yadda da yawa daga nasara ne a sakamakon na inganci da nawa ne kawai luck. Kõ bayyana dan kadan daban, idan za mu iya haifar da layi daya halittu, kuma suna da su duka halittu farfadowa da kansa, zai guda songs zama mashahuri a kowace duniya? Kuma, idan ba, da abin da zai zama mai inji cewa yana sa wadannan bambance-bambance?

Domin amsa wadannan tambayoyi, za mu-Peter Dodds, Duncan Watts (ta dissertation shawara), da kuma I-gudu jerin online filin gwaje-gwajen. Musamman, za mu gina wani website da ake kira MusicLab inda mutane zai iya samu sabon music, kuma mun yi amfani da shi ga wani jerin gwaje-gwajen. Mun dauka mahalarta da guje banner talla a kan wani matasa-sha'awa website (Figure 4.19) da kuma ta hanyar ambaci a cikin kafofin watsa labarai. Mahalarta isa a mu website bayar informed yarda, kammala wani ɗan gajeren baya tambayoyi, kuma aka da ka sanya daya daga biyu gwaji yanayi-m da kuma zamantakewa tasiri. A cikin m yanayin, mahalarta sanya yanke shawara game da abin da songs saurare, ba kawai sunayen makada da songs. Duk da yake sauraron wani song, mahalarta aka tambaye su Rate shi bayan da suka samu damar (amma ba wajibi) to download da song. A cikin zamantakewa tasiri yanayin, mahalarta da wannan kwarewa, fãce su iya ganin yadda sau da dama kowane song aka sauke daga baya mahalarta. Bugu da ƙari kuma, mahalarta a cikin zamantakewa tasiri da yanayin da aka sanya da ka ga daya daga takwas a layi daya halittu kowanne daga abin da ya samo asali da kansa (Figure 4.20). Amfani da wannan zane, mu gudu biyu related gwaje-gwajen. A cikin farko, za mu gabatar da mahalarta da songs a cikin wani unsorted Grid, wanda azurta su mai rauni sigina na shahararsa. A karo na biyu gwaji, mun gabatar da songs a ranked list, wanda ya bayar da yawa da suka fi karfi sigina na shahararsa (Figure 4.21).

Figure 4,19: Wani misali na banner ad cewa ta abokan aiki, kuma ina amfani da su kurtu mahalarta ga MusicLab gwajen (Salganik, Dodds, and Watts 2006) .

Figure 4.20: gwajin zane ga MusicLab gwajen (Salganik, Dodds, and Watts 2006) . Mahalarta da aka da ka sanya a cikin daya daga biyu yanayi: m da kuma zamantakewa tasiri. Mahalarta a m yanayin sanya su zabi ba tare da wani bayani game da abin da wasu mutane suka yi. Mahalarta a social tasiri da yanayin da aka da ka sanya a cikin daya daga takwas a layi daya halittu, inda za su iya ganin shahararsa-auna downloads na baya mahalarta-kowane song a cikin duniya, amma ba su iya ganin wani bayani, kuma ba su yi ma sani game da wanzuwar, wani daga cikin sauran halittu.

Mun gano cewa, da shahararsa na songs sãɓã wa jũna a fadin halittu bayar da shawara muhimmiyar rawa sa'a. Alal misali, a daya duniya da song "An kulle" by 52Metro zo a cikin 1st, kuma a wata duniya ta zo a cikin 40th daga 48 songs. Wannan shi ne daidai wannan song gasar da duk daya songs, amma a daya duniya ya samu sa'ar da kuma a cikin wasu da ya aikata ba. Bugu da ari, ta hanyar kwatanta sakamakon fadin biyu gwaje-gwajen da muka gano cewa, zamantakewa tasiri take kaiwa zuwa more unequal nasara, wanda watakila halitta bayyanar wani hasashen. Amma, neman fadin halittu (wanda ba za a iya yi a waje, irin wannan layi daya halittu gwaji), za mu gano cewa, zamantakewa tasiri a zahiri ya karu da unpredictability. Bugu da ari, abin mamaki, shi ne songs of mafi roko da cewa suna da mafi unpredictable sakamakon (Figure 4.22).

Figure 4,21: Screenshots daga zamantakewa tasiri yanayi a cikin MusicLab gwajen (Salganik, Dodds, and Watts 2006) . A cikin zamantakewa tasiri yanayin a gwajin 1, songs, tare da yawan m downloads, da aka gabatar wa mahalarta shirya a 16 X 3 rectangular Grid, inda matsayi na songs aka da ka sanya ga kowane ɗan takara. A gwajin 2, mahalarta a cikin zamantakewa tasiri da yanayin da aka nuna da songs, download da kirga, gabatar a daya shafi a saukowa domin yanzu shahararsa.

Figure 4.22: Results daga MusicLab gwajen nuna dangantaka tsakanin roko da kuma nasara (Salganik, Dodds, and Watts 2006) . The x-axis ne kasuwa rabo daga song a zaman duniya, wanda hidima a matsayin ma'auni na roko na song, da y-axis ne kasuwar share na wannan waƙa a 8 zamantakewa tasiri halittu, abin da hidima a matsayin ma'auni na nasarar da songs. Mun gano cewa, kara da zamantakewa tasiri cewa mahalarta dandana-musamman, da canji a layout daga gwaji 1 to gwaji 2 (Figure 4.21) -caused nasara ya zama mafi unpredictable, musamman ga mafi roko songs.

MusicLab ya iya gudu a gaske sifili m cost saboda hanyar da aka tsara. Na farko, duk abin da aka cikakken sarrafa kansa don haka shi ne iya gudu yayin da nake barci. Na biyu, da diyya da aka free music haka babu m yar diyyar kudin. Yin amfani da music matsayin diyya kuma nuna yadda akwai wani lokaci da wani ciniki-kashe tsakanin gyarawa halin kaka da kuma m halin kaka. Amfani music karu da gyarawa halin kaka, saboda ina da suke ciyarwa lokaci kullawa izni daga makada da kuma shirya rahotanni ga makada game mahalarta dauki ga music. Amma, a cikin wannan harka, kara gyarawa halin kaka domin rage canji halin kaka da hakkin ya abu da ya yi. wancan ne abin da ya sa mu mu yi gudu da wani gwaji da aka game da 100 sau ya fi girma fiye da misali Lab gwaji.

Bugu da ari, MusicLab gwajen nuna cewa sifili m cost ba dole ba ne ya kasance mai ƙarshen a kanta. m, zai iya zama a wajen guje wani sabon irin gwaji. Ka lura cewa ba mu yi amfani da duk mu mahalarta gudu wani misali zamantakewa tasiri Lab gwaji 100 sau. A maimakon haka, za mu yi wani abu daban-daban, abin da za ka iya tunanin yadda ya sauya sheka daga m gwajin zuwa sociological gwaji (Hedström 2006) . Maimakon mayar da hankali a kan mutum yanke shawara, mun mayar da hankali mu gwaji a shahararsa, a gama sakamako. Wannan canji zuwa gama sakamako nufi cewa muna da ake bukata game da 700 mahalarta su samar da guda data nufi (akwai 700 mutane a kowane daga cikin layi daya halittu). Wannan sikelin ne kawai zai yiwu saboda da kudin tsarin da gwaji. A general, idan masu bincike so ya yi nazarin yadda gama sakamakon bayyana daga mutum yanke shawara, kungiyar gwajen kamar MusicLab ne sosai m. A baya, sun kasance logistically wuya, amma waɗanda matsaloli suna Fading saboda yiwuwar sifili m kudin data.

Bugu da ƙari, dake bayyana amfanin sifili m kudin data, da MusicLab gwaje-gwajen kuma nuna wani kalubale da wannan dabarar: high gyarawa halin kaka. A yanayin, Na musamman m iya aiki tare da talented yanar gizo developer mai suna Peter Hausel watanni shida zuwa yi gwajin. Wannan shi ne kawai zai yiwu saboda ta bada shawara, Duncan Watts, ya samu dama da tallafin taimaka wa irin wannan bincike. Technology ya inganta tun da muka gina MusicLab a shekarar 2004, kuma shi zai zama sauƙin gina wani gwaji, kamar wannan yanzu. Amma, high gyarawa cost dabarun ne da gaske kawai zai yiwu ga masu bincike da suka iya ko ta yaya rufe wadanda halin kaka.

A ƙarshe, digital gwaje-gwajen da za a iya cika fuska daban-daban kudin Tsarin fiye analog gwaje-gwajen. Idan kana so ka gudu da gaske manyan gwaje-gwajen, ya kamata ka yi kokarin rage your m kudin kamar yadda zai yiwu da kuma fi dacewa duk hanyar zuwa 0. Za ka iya yin haka ta automating da makanikai na gwaji (misali, ya maye gurbin dan Adam lokacin da kwamfuta lokaci) kuma zayyana gwaje-gwajen da cewa mutane suna so su kasance a. bincike wanda zai iya tsara gwaje-gwajen da wadannan siffofin za su iya gudanar da sabon iri-gwajen da suke ba zai yiwu ba a baya.