Skip to content

Assuming normality #1

Description

@bschneidr

Hi Daniel,

This app is really well put together and is a nice learning resource. I did want to suggest though that you'd want to drop the Step 1 page (checking whether the outcome is normally distributed) from the app or at least update the text.

As you probably know, the actual assumption for regression isn't that the outcome variable is normally-distributed, but that the residuals are normally distributed. This is an important but confusing distinction for many people new to regression, and the normality checks in Step 1 make this more confusing by making it seem like it's necessary for the outcome (and not just the residuals) to be normally distributed.

This SO post elaborates on the point a bit:
https://stats.stackexchange.com/questions/105239/regression-on-a-non-normal-dependent-variable

This app is otherwise great, it's just this aspect which is a bit misleading for students.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions