Analysis:

We have two measurements of NFL per person (NFL3 and NFL4). We want to convert NFL4 to NFL3.

We ran a regression and a LME (random intercept) of NFL4 predicting NFL3.

We would like to compare results, to see if the LME or LM approach is better.

1. Plot compares the predicted NFL values derived from the LM vs the LME.

2. Plot compares the log transformed predicted NFL values derived from the LM vs the LME.

add in 3, but not tansformed.

3. Plots (2) compare the predicted NFL values derived from the LM vs the LME. Color indicates the total number of visits completed by that participant.

—

plot 1: LM predicted vs LME predicted

—

plot 1: log(LM) predicted vs log(LME) predicted

4. Plots (2) compare log transformed NFL3 values to log transformed predicted values from the LM and LME.

plot 1: log NFL3 vs log LM

—

plot 2: log NFL3 vs log LME

5. Plots (4) compare NFL3 to differences scores (NFL3 - predicted value).

—

plot 1: NFL3 vs ( NFL3 - LM )

—

plot 2: NFL3 vs ( NFL3 - LME )

—

plot 3: log(NFL3) vs [ log(nfl3) - log(LM) ]

—

plot 4: log(NFL3) vs [ log(nfl3) - log(LME) ]