Data Analysis & Interpretation Techniques Flashcards

Overall conversion rate rose, yet every individual traffic segment's conversion rate fell. What phenomenon best explains this?

Answer: Simpson's paradox caused by a shift in traffic mix

Simpson's paradox occurs when aggregate trends reverse because the proportion of segments changed.

Which analysis groups users by the date they first visited and tracks their behavior over subsequent weeks?

Answer: Cohort analysis

Cohort analysis compares groups sharing a common starting event over time to reveal retention patterns.

A report shows a 40% jump in sessions on a single day with no marketing activity. What should the analyst do FIRST?

Answer: Check for anomalies such as bot traffic, tracking duplication, or a tagging change

Unexplained spikes should be validated against data-quality causes before being interpreted as real performance.

Which metric is the most appropriate for comparing the efficiency of two campaigns with very different traffic volumes?

Answer: Conversion rate

Rate-based metrics normalize for volume, allowing fair comparison between unequal groups.

When the mean time on page is 4 minutes but the median is 40 seconds, what does this most likely indicate?

Answer: A few extremely long sessions skew the mean upward

A mean far above the median signals a right-skewed distribution driven by outliers.

An analyst finds that pages with more images have higher conversion rates. What is the most defensible conclusion?

Answer: There is a correlation, but causation requires a controlled test

Observational correlation does not prove causation; an experiment is needed to confirm cause.

What is the main purpose of establishing a baseline before analyzing the effect of a site change?

Answer: To provide a reference point for measuring change

A baseline gives the pre-change performance level against which results are compared.