A marketing team put up billboards in several cities and now wants to know whether the campaign increased brand awareness versus cities without billboards.

Design an experiment to measure the billboard campaign’s effect, starting from hypothesis definition and sampling strategy. How would you handle potential biases such as city-level differences in population or existing brand affinity? After data collection, what statistical test or model would you use and why?

Think randomization, stratification, pre-post matching, and difference-in-differences.

Pinterest Data Scientist interview question: Measure Billboard Campaign Impact: Design, Bias, Test Strategy. {"blocks": [{"key": "e977ad5d", "text": "Scenario", "type": "header-two", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {...