#datacollection #statistics #quantitativemethod
[The Hindu] very very important article relevant for Unit 3 in paper 1.
ANALYSIS:
"Model collapse" and the "liar's dividend" come from the article (via WEF). These are important pitfalls of synthetic data.
Reliability and validity: synthetic data can look consistent (reliable) while not measuring reality (invalid).
Douglas, Atkinson, Cicourel: Official statistics are products of how data gets recorded and categorised, not neutral facts. This fits well with the article's point that synthetic data inherits the limits of its source data.
Weber (Verstehen): meaning and context are lost when a model fills the gaps.
Malinowski (fieldwork): the case for ground-level qualitative evidence that a statistical model can't generate.
[The Hindu] very very important article relevant for Unit 3 in paper 1.
ANALYSIS:
"Model collapse" and the "liar's dividend" come from the article (via WEF). These are important pitfalls of synthetic data.
Reliability and validity: synthetic data can look consistent (reliable) while not measuring reality (invalid).
Douglas, Atkinson, Cicourel: Official statistics are products of how data gets recorded and categorised, not neutral facts. This fits well with the article's point that synthetic data inherits the limits of its source data.
Weber (Verstehen): meaning and context are lost when a model fills the gaps.
Malinowski (fieldwork): the case for ground-level qualitative evidence that a statistical model can't generate.
❤2