{"blocks": [{"key": "7a3ab410", "text": "Scenario", "type": "header-two", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "9114554b", "text": "Meta Data Scientist onsite round focused on statistical reasoning behind product experimentation.", "type": "unstyled", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "6ec1cdf5", "text": "Question", "type": "header-two", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "f1cf0f9f", "text": "Explain the difference between Type I and Type II errors in A/B testing and how you would choose acceptable levels for each at Meta. If the metric distribution is highly skewed with outliers, how would you estimate treatment lift and construct a confidence interval?", "type": "unstyled", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "ea54e3c3", "text": "Hints", "type": "header-two", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "0035cc24", "text": "Discuss α vs β, power analysis, non-parametric tests, bootstrapping or transforms for skewed data.", "type": "unstyled", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}], "entityMap": {}}