- Thread starter Azevedo
- Start date

The black dots are the individual data points for the groups. The red dots are the means for the groups. Each black line is a rough best fit line within each group (they all have negative slope). The red line is the best fit line for the aggregated data. So it's possible that the relationship 'flips' depending on what you actually care about. Note that even in this example if we had the raw data and ignored group the overall trend would have a positive slope. If we included the grouping variable to at least give a different intercept we would end up with a negative slope though.