datacamp-logo
Session Ready
Exercise

A Bad Argument for As-If Randomization with Katie Perrie

Although it is wholly misleading, one way that unethical scientists justify their arguments that their studies are "as-if" natural experiments is to only report descriptive information about variables in their datasets that are balanced (i.e. with basically random assignment). That's bad science, but let's give it a try, just one time!

The dataset KatiePerrie contains covariate data from a natural experiment about how repeated exposure to music performed by pop star Katie Perrie (measured as owning a Katie Perrie album) is associated with the debilitating psychological disorder called "choreomania." The human rights activists who gathered this data want to publish an article claiming that Katie Perrie's music should be banned since it appears to be harmful to the public's mental health.

However, the control and treatment groups in this "natural experiment" are not balanced across most key variables, so it's not a valid experiment. Nonetheless, let's ignore that and help the human rights activists find one variable, any variable, that shows balance so they can incorrectly justify their assumption that this is an as-if natural experiment:

Instructions
100 XP
  • 1) Use what you have learned about balance and statistical inference and look through each variable in the dataset KatiePerrie. Name the variable that is balanced between the datasets.