3.3 Example 2

  • Population: Tweets; Google searches
  • Sample: non-random sample of tweets; random sample of Google searches
  • Concept:
    • Sentiment (tweets): [Negative, Positive]
    • Topic (searches): [Politics, Sports, Culture, Science]
  • Measure:
    • Classifier (Sentiment): Assigns sentiment scores to tweets (based on content)
    • Topic model (topic): Assigns topics to searches (based on content)
  • Variable/scale:
    • Sentiment: ..has values between 0-1 (negative-positive)
    • Topic: Has for categories [Politics, Sports, Culture, Science]
  • Measurement process
    • Assign sentiment values to tweets
    • Assign topics to searches
  • Data: Sample of tweets/searches grouped according to values (cells)
    • …again each value/topic represents a cell
  • Q: What is could measurement error be here?