Experimental Design CERTIFIED-DATA-SCIENCE-PRACTITIONER Practice Questions (Page 1)

Questions 1–5

1. 1foundation · easy

   In the context of data science model testing, what is the primary purpose of experimental design?

   • ATo eliminate the need for statistical analysis after the experiment
   • BTo structure the testing process so that observed effects can be attributed to the changes being tested rather than to external factors
   • CTo maximize the number of features used in the model to improve accuracy
   • DTo ensure the model's predictions are always correct on unseen data
   Reveal Solution Discussion Select an answer first
2. 2application · medium

   A data scientist is designing an experiment to test a new feature in a mobile app. They are concerned that user engagement might vary by day of the week. To ensure the treatment and control groups are comparable, they decide to assign users to groups randomly within each day of the week. This approach is an example of:

   • AConfounding, because the day of the week is not controlled.
   • BRandomization, because users are assigned to groups by chance.
   • CBlocking, because the day of the week is controlled as a nuisance factor.
   • DReplication, because the experiment is repeated on multiple days.
   Reveal Solution Discussion Select an answer first
4. 3application · medium

   A company is testing a new website layout to see if it increases purchase conversion. They are concerned that the effect might be different for users coming from different marketing channels (e.g., organic search, paid ads, social media). They have enough users in each channel. Which approach best controls for the channel as a confounding variable?

   • AStratify users by channel and randomly assign the layout within each channel group.
   • BExclude users from channels with low traffic to simplify the analysis.
   • CUse a factorial design with channel as a factor and layout as the other factor.
   • DRandomly assign users to the new or old layout without considering their channel.
   Reveal Solution Discussion Select an answer first
5. 4foundation · easy

   Why is replication important in experimental design?

   • AIt guarantees the model will perform well in production
   • BIt eliminates the need for randomization
   • CIt increases the reliability of results by reducing the impact of random variation
   • DIt reduces the number of subjects needed
   Reveal Solution Discussion Select an answer first
6. 5foundation · easy

   In an experiment to test a new pricing model, a data scientist notices that users from a certain geographic region behave differently. To ensure this regional difference does not confound the results, which method should be used?

   • AIncreasing the sample size
   • BUsing a factorial design
   • CRemoving all users from that region
   • DStratified randomization
   Reveal Solution Discussion Select an answer first