Varying measures of Covariates

[andkov]

@annierobi and @GracielaMuniz , i'm drawing your attention to the fact that in the current state the models from ELSA don't conform to the typical structure of the covariates. To illustrate, consider the output for ELSA's b1_female_aehplus_fev_gait. Education is defined by a collection of variables

MODEL RESULTS
                                                    Two-Tailed
                    Estimate       S.E.  Est./S.E.    P-Value
 IP         ON
    BAGE              -0.010      0.001    -13.380      0.000
    EDUCN             -0.107      0.025     -4.331      0.000
    EDUCSEC           -0.034      0.027     -1.251      0.211
    EDUCDE             0.000      0.030      0.011      0.992
    EDUCHE            -0.006      0.030     -0.216      0.829
    EDUCF             -0.076      0.027     -2.786      0.005
    EDUCO              0.006      0.048      0.131      0.896
    HEI2               0.005      0.001      6.235      0.000
    SMOKE1             0.033      0.011      3.075      0.002
    DIAB1             -0.122      0.024     -5.105      0.000
    CARDIO1           -0.079      0.018     -4.479      0.000

instead of a single variables, like in the rest of the studies, for example RADC's b1_female_aehplus_fev_gait

MODEL RESULTS

                                                    Two-Tailed
                    Estimate       S.E.  Est./S.E.    P-Value

 IP         ON
    BAGE              -0.023      0.002    -14.764      0.000
    EDUC               0.012      0.004      3.426      0.001
    HEIGHT             1.626      0.216      7.539      0.000
    SMOKHIST          -0.087      0.023     -3.737      0.000
    DIAB              -0.024      0.039     -0.609      0.542
    CARDIO            -0.063      0.050     -1.269      0.204

If detailed results on the covariate loading are required as #86 prescribes then ELSA may not be included. I also suspect that this issue is not unique and will be encountered in the future. Please advise in this thread.

[ampiccinin]

Yeah, we discussed this - Annie, Graciela, Sean and I.

There is a similar issue in ILSE.

There is no good solution.

Since our emphasis is on the S-S correlation, rather than on the covariate associations, I'm thinking we just live with it for now. Yes, it won't fit in that table, but both ELSA and ILSE have barely three waves, along with other issues, so it is a matter of how much time and effort it is worth, especially at this point. This is all stuff that should have been worked out in February, not two weeks before we submit a paper. It is too late and we will just have to somehow make it work the way it is.

It IS something to deal with before the Physical-Cognitive paper, however (yes, not ideal to do it one way for one paper and another for another, but I have too many other pieces to pull together to worry about a small inconsistency that (I HOPE!) will not make a big difference in the end).

...so, does this mean you are making tables?

Thanks!

[andkov]

I'm giving it my best shot this weekend. It'll be like launching a shuttle: many things can go wrong