Guide-yakabatana mibvunzo · Pakati Nhanho

Maitiro eKuongorora Mhedzisiro Yekuongorora neAI Quiz

Miedzo inovaka bhuku rekodhi remhinduro dzakavhurika, kuchengetedza maperesenti akasungirirwa kune chaiwo kuverenga, uye kuongorora AI-inokurudzirwa makotesheni inopesana nekwakabva data.

Nzira dzekutungamira dzinoenderanaMaitiro Ekuongorora Mhedzisiro Yeongororo neAI
Mubvunzo 1 ye 8

Chii chinonzi codebook mukuongorora ongororo?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. Chii chinonzi codebook mukuongorora ongororo?
    • A list of respondents' login credentials
    • A spreadsheet formula for calculating averages
    • A set of named themes with definitions, used to label open-ended answers
    • A software license for survey tools
  2. Chidimbu cheAI chinoti 'inenge 40% yevakapindura vakataura mutengo.' Unofanira kuitei?
    • Recalculate the figure from the labeled responses with a spreadsheet or code
    • Accept it, because language models count accurately
    • Round it to 50% for simplicity
    • Delete the number and add a chart instead
  3. Ndeipi nzira yakanakisa yekusimbisa kuti AI-inopihwa quote ndeyechokwadi?
    • Ask the AI whether it is sure
    • Check whether it sounds like something a respondent would say
    • Rewrite it in plainer language
    • Search the original responses for the exact text, using its response ID
  4. Ko inductive coding inorevei?
    • Using only categories decided before reading any responses
    • Letting themes emerge from what respondents actually wrote
    • Coding only the multiple-choice questions
    • Having the AI rate its own accuracy
  5. Sei uchibatanidza 'mamwe' uye 'asina kujeka' makodhi mubhuku rako rekodhi?
    • They make the final chart more colorful
    • They are legally required for all surveys
    • They stop the model from forcing unclear answers into themes that do not fit
    • They increase the survey's response rate
  6. What does Cohen's kappa measure?
    • Agreement between two coders, corrected for agreement expected by chance
    • The average answer on an agreement scale
    • The survey's response rate
    • The average length of open-ended answers
  7. Before pasting survey responses into a chatbot, which privacy step matters most?
    • Sorting responses by length
    • Translating them into English
    • Converting them to PDF
    • Removing names, emails and other personal details and replacing them with IDs
  8. Why report counts alongside percentages for subgroups?
    • Percentages cannot be calculated for subgroups
    • So readers can see when a percentage rests on very few people
    • Counts are always more accurate than code
    • Survey software requires it