A recent Washington Post analysis by Philip Bump asserts Democrats are more likely to say they approve of President Donald Trump’s job performance when they respond to a poll conducted online or via IVR than by telephone. In turn, Bump argues, the president’s job approval rating is higher in polls conducted online rather by phone.

In this episode, the Poll Hub team debunks Bump’s analysis through an in-depth discussion of social desirability, mode effects, and a study from the Pew Research Center and explains why omitting house effects from the analysis of survey research is dangerous.