4.5.1.2 gina your own gwaji

Gina your own gwajin zai yi m, amma zai taimaka maka ka halitta gwaji cewa kana so.

Bugu da ƙari, ado gwaje-gwajen a saman data kasance yanayin, za ka iya gina your own gwaji. Babban amfani da wannan dabarar ne iko. idan kana gina gwaji, za ka iya haifar da yanayi da jiyya da ka ke so. Wadannan bespoke gwaji muhallin iya ƙirƙirar damar jarraba theories cewa su ne ba zai yiwu ba a gwada a halitta abin da ke faruwa halin wurare. Babban drawbacks gina your own gwaji ne cewa shi zai iya zama tsada da kuma cewa yanayi da ka iya haifar da zai ba da labarun almara mai sauƙi abin da ke faruwa da tsarin. Masu bincike gina nasu gwaji kuma dole ne a yi dabarun domin a jawo ra'ayinsu mahalarta. Lokacin aiki a data kasance tsarin, masu bincike suna da gaske kawo gwaje-gwajen su mahalarta. Amma, a lokacin da masu bincike gina nasu gwaji, suna bukatar kawo mahalarta da shi. Abin farin, ayyuka kamar Amazon Mechanical Turk (MTurk) zai iya samar da masu bincike a dace hanyar kawo mahalarta su gwaje-gwajen.

Daya misali da cewa nuna falalan bespoke muhallin for gwada m theories ne digital Lab gwaji da Gregory Huber, Shitu Hill, kuma Gabriel Lenz (2012) . The gwaji duba yiwuwar m iyakancewa ga aiki na tsarin dimokaradiya. Tun da farko ba gwaji karatu na ainihin zaben bayar da shawarar cewa masu kada kuri'a ba su iya daidai tantance wasan kwaikwayon na tabbatacce yan siyasa. Musamman ma, masu jefa} uri'a ya bayyana a sha daga uku biases: 1) mayar da hankali kan 'yan maimakon tarawa yi. 2) manipulatable da rhetoric, siffatawa, da kuma sayar da. da kuma 3) ya rinjayi events alaqa tabbatacce yi, kamar nasarar gida wasanni tawagar da weather. A cikin wadannan baya karatu, duk da haka, shi ne wuya a ware wani daga cikin wadannan dalilai daga dukan sauran kaya da ya faru a hakikanin, m zabe. Saboda haka, Huber da kuma abokan aiki halitta sosai Sauki zabe yanayi domin ware, sa'an nan kuma gwaje-gwaje karatu, kowane daga cikin wadannan uku yiwu biases.

Kamar yadda na bayyana gwaji sa-up a kasa shi da yake faruwa don sauti sosai wucin gadi, amma ka tuna cewa hakikanci ba wata manufa a Lab-style gwaje-gwajen. Ã'a, manufa shi ne a fili ware tsari da kake kokarin nazarin, kuma wannan m kadaici ne, wani lokacin ba zai yiwu ba a karatu da more hakikanci (Falk and Heckman 2009) . Bugu da ari, a cikin wannan batu, da masu bincike jãyayya da cewa idan masu jefa} uri'a ba zai iya yadda ya kamata kimanta yi a cikin wannan sosai Saukake saitin, sa'an nan sũ, ba su faruwa a iya yin shi a cikin wani more idon basira, more hadaddun saitin.

Huber da kuma abokan aiki amfani Amazon Mechanical Turk (MTurk) kurtu mahalarta. Da zarar takara azurta informed yarda da wuce a takaice gwajin, ta gaya mana cewa ta shiga a cikin wani 32 zagaye game aikatãwa tsõkaci da za a iya tuba a cikin real kudi. A farkon wasan, kowane ɗan takara aka gaya cewa ta aka sanya wani "allocator" da zai ba ta free tsõkaci kowane zagaye da cewa wasu allocators kasance mafi m fiye da wasu. Bugu da ari, kowane ɗan takara ya kuma gaya mana cewa, ta yi da damar da ko dai ci gaba da ta allocator ko a sanya wani sabon daya bayan 16 akai-akai na wasan. Ganin abin da kuka sani game da Huber da kuma abokan aiki 'bincike a raga, za ka iya ganin cewa allocator wakiltar gwamnati kuma wannan zabi wakiltar wani za ~ e, amma mahalarta kasance ba ku sani ba daga cikin general raga na gudanar da bincike. A cikin duka, Huber da kuma abokan aiki dauka game da 4,000 mahalarta da suka biya game da $ 1.25 domin wani aiki da ya game 8 minutes.

Ka tuna da cewa daya daga cikin binciken daga baya bincike shi ne, masu jefa} uri'a, sakamako da azabtar ai ga sakamakon cewa su ne a fili bayan da iko, kamar nasarar gida wasanni teams da weather. Don tantance ko mahalarta zabe yanke shawara za a iya rinjayi zalla bazuwar al'amura a saitin, Huber da kuma abokan aiki kara da cewa wani irin caca ga gwaji tsarin. A ko dai 8th zagaye ko 16th zagaye (watau dama da damar maye gurbin allocator) mahalarta aka da ka sanya a cikin wani irin caca inda wasu lashe 5000 da maki, wasu lashe 0 maki, da kuma wasu rasa 5000 da maki. Wannan irin caca da aka yi nufi ga mimic mai kyau ko mummuna labarai cewa shi ne mai zaman kanta da wasan kwaikwayon na siyasa. Ko da yake mahalarta aka baro-baro ya gaya cewa irin caca ya alaqa da wasan kwaikwayon na su allocator, sakamako ne na irin caca har yanzu tasiri mahalarta yanke shawara. Mahalarta cewa amfana daga irin caca kasance mafi kusantar su ci gaba da allocator, kuma wannan sakamako ya fi karfi a lõkacin da caca da ya faru a zagaye 16-dama kafin sauyawa yanke fiye da lokacin da ya faru a zagaye 8 (Figure 4.14). Wadannan sakamakon, tare da sakamakon da dama wasu gwaje-gwajen a cikin takarda, ya jagoranci Huber da kuma abokan aiki a kammala da cewa ko da a cikin wani Saukake saitin, masu jefa} uri'a da wahala yin m yanke shawara, a sakamakon cewa tasiri nan gaba da bincike game da masu jefa} uri'a yanke shawara Making (Healy and Malhotra 2013) . The gwajin na Huber da kuma abokan aiki ya nuna cewa MTurk za a iya amfani kurtu mahalarta for Lab-style gwaje-gwajen a daidai jarraba sosai musamman theories. Har ila yau, ya nuna darajar gina your own gwaji yanayi: yana da wuya su yi tunanin yadda wadannan guda matakai iya an ware haka cleanly a wani wuri.

Figure 4.14: Results daga Huber, Hill, and Lenz (2012) . Mahalarta cewa amfana daga irin caca kasance mafi kusantar su riƙe su allocator, kuma wannan sakamako ya fi karfi a lõkacin da caca da ya faru a zagaye 16-dama kafin sauyawa yanke fiye da lokacin da ya faru a zagaye 8.

Bugu da ƙari, gina Lab-kamar gwaje-gwajen da masu bincike kuma iya gina gwaje-gwajen da cewa su ne mafi filin-kamar. Alal misali, Centola (2010) ya gina digital filin gwaji yi nazarin sakamako na zamantakewa na cibiyar sadarwa tsarin kan yaduwar hali. Da bincike tambaya da ake bukata shi ne don tsayar da wannan hali yada a alƙarya da cewa yana da daban-daban na zamantakewa na cibiyar sadarwa Tsarin amma sun kasance in ba haka ba fayyace. Iyakar hanyar yin wannan tare da wata bespoke, custom-gina gwaji. A wannan yanayin, Centola gina yanar gizo na tushen kiwon lafiya al'umma.

Centola dauka game da 1,500 mahalarta da talla a kan kiwon lafiya yanar. A lokacin da mahalarta isa online jama'a-wanda aka kira da Healthy Salon Network-su auren sanar yarda sa'an nan kuma aka sanya "lafiya buddies." Saboda hanyar Centola sanya wadannan lafiya buddies ya ya iya saƙa tare daban-daban na zamantakewa na cibiyar sadarwa Tsarin cikin daban-daban kungiyoyin. Wasu kungiyoyi da aka gina a yi bazuwar networks (inda kowa da kowa yana daidai da m da za a haɗa) da kuma sauran kungiyoyin da aka gina a yi nonnan networks (inda sadarwa ne mafi gida mai yawa). Sa'an nan kuma, Centola gabatar da wani sabon hali a cikin kowane cibiyar sadarwa, da damar yin rajistar wani sabon website tare da ƙarin lafiya bayani. A duk lokacin da kowa ya sanya hannu up for wannan sabon website, duk ta kiwon lafiya buddies samu wani email bayyana wannan hali. Centola tarar, cewa wannan hali-sayi-up don sabon website-yada kara da sauri a cikin clustered cibiyar sadarwa fiye da bazuwar cibiyar sadarwa, a binciken da ya saba wa wata data kasance theories.

Overall, gina your own gwaji ya ba ka fiye da iko. shi sa ka ka yi mafi kyau zai yiwu yanayi don ware abin da kuke so karatu. Yana da wuya su yi tunanin yadda za ko dai daga cikin wadannan gwaje-gwajen dã an yi a cikin wani riga data kasance yanayi. Bugu da ari, gina your own tsarin rage-rage da'a damuwa a kusa da gwaji a data kasance tsarin. A lokacin da ka gina naka gwaji, duk da haka, ka gudu zuwa cikin da dama daga cikin matsalolin da ake fuskanta a Lab gwaje-gwajen: jawo ra'ayinsu mahalarta da kuma damuwa game da labarun almara. A karshe downside shi ne cewa gina your own gwajin zai iya zama m da kuma lokacin-cinyewa, ko da yake kamar yadda wadannan misalai nuna, da gwaje-gwajen da za a iya Range daga gwada sauki muhallin (kamar nazarin zabe da Huber, Hill, and Lenz (2012) ) to gwada hadaddun muhallin (kamar nazarin networks da contagion da Centola (2010) ).