Validate independence of residues with Autocorrelation or Ellipse plot

[#1]

Hi,

I'm trying to validate the independence of residues by reading the stats summary from distribution and auto-correlation showed result is 0.27 (within threadshold +/- 0.6). But when I plotted ellipse plot (residue values vs. row) and use coverage (0.95), r=-0.63 (above threadshold +/- 0.6), and I can see a clear downward trend across the x-axis. I was assuming these results should be similar, can someone explain the difference and which one should I choose?

[#2]

Can you share screenshots of these results for evaluation? Can you tell us which platforms you used for each result?

[#3]

The first result is the auto-correlation of the data lagged by one on itself. The second is the (overall) correlation between the centered observations and the row numbers. They are not the same thing.

There is a downward trend if the observations are entered in a time sequence (row number).