Proposes consistent estimators for average and local treatment effects on treated and respondent users in A/B tests with non-response bias in sentiment surveys, evaluated via simulations.
Overlapping experiment infrastructure: More, better, faster experimentation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ME 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
User Sentiment as a Success Metric: Persistent Biases Under Full Randomization
Proposes consistent estimators for average and local treatment effects on treated and respondent users in A/B tests with non-response bias in sentiment surveys, evaluated via simulations.