How to detect Spurious Correlations in R using Linear Models

Storks bring babies - probably the most famous example in statistics for a spurious correlation. While it may be easy here to recognize the association between storks and babies (which has been empirically confirmed) as spurious because we know a more convincing theory about where babies come from, we may find that harder in other use cases, where theory might not be that clear. So how can we test for spurious correlations in a statistical way?

We use the level of industrialization of a region as a control variable and create three linear models, using the number of babies as the dependent variable each time:

1. A model that only uses the number of storks as a predictor (independent variable),
2. a model that only uses the level of industrialization as a predictor,
3. a model that uses both the number of storks and the level of industrialization as predictors.

This is a fine example where a non-significant result can be highly interesting and a very useful finding!

You can find data and code here: https://github.com/fjodor/spurious_hi...

Contact me, e. g. to discuss (online) R workshops / trainings / webinars:

LinkedIn:   / wolfriepl  
Twitter:   / statistikindd  
Xing: https://www.xing.com/profile/Wolf_Riepl
Facebook:   / statistikdresden  

https://statistik-dresden.de/kontakt
R Workshops: https://statistik-dresden.de/r-schulu...
Blog (German, translate option): https://statistik-dresden.de/statisti...

Playlist: Music chart history
   • Music Chart History