Your recent help with my model specification has been invaluable. I also followed your advise on reading the book by Berry (1993), which led me to additional explorations so I can have BLUE estimates

So, all my assumptions are met (as supported by the specification tests). Breusch-Pagan test rejects null of constant variance. I attribute that to the intentionally retained outliers (Cook's Ds and DFBETAs are within threshold of 1 for 99% of the data points). I decided to examine

#1 My initial remedy was to use robust SE (White), but they don't allow me to "go BLUE" :)

#2 I ran the OLS estimation of the model with analytic weight [aweight=Y] (following a document I found by Kousser from UCSD), which gave me (a) consistent (with #1) coefficient estimates and more efficient (i.e. smaller) SE and confidence intervals, (b) 10%(!) higher R squared, and (c) even better distribution of the fitted residuals (best one so far).

#3 Additionally, I followed recommendation by Berry (1993) to use GLS for BLUE estimators. So, I calculated omega as follows (using STATA):

- ran regression, saved residuals, sorted Y;

- by Y: egen omega=mean(r1^2)

Then I used variance-weighted least squares (vwls) with the newly created omega as conditional SD. The new coefficients estimates were very close to those in #1 and #2 and the SEs were the most efficient (small) out of all 3 estimations.

Please let me know if my logic and procedures are correct in all this. Thank you in advance :)

A short warning for a very long post!

Im currently working on my master thesis in mechanical engineering. I need to do, for me, an advanced regression analysis on some data - and i therefore seek the experience of you pro's :)

I have some experimentally obtained motion data for a big container ship, which i compare to theoretical derived data. An example of such can be seen here:

The uncertainty between the theoretical derived data compared to the experimental data can be defined as: where is experimental data and is theoretical data, and is the difference between the data.

phi can then be a simple form or and more advanced form , where omega is the frequency as seen on the photo above. a and b is then determined in both cases.

The data is obtained for various wave headings (U), speed (V) and draft(D) of the ship, and i think therefore it's obvious to make an regression analysis to determine how a, b and c change when wave heading, speed and draft changes.

I set up a simple linear regression model for the simple phi first, i.e.:

and an adjusted R^2 value of 0.33 and three not so good looking residual plots. I therefore change the model to a quadratic second order model, which i think is non-linear??, i.e:

and gets a adjusted R^2 value of 0.83 and great looking residual plots.

I then tried to test the more advanced , but i fail to understand how to do it. Do i just set and then do the calculation for various omega values??

Il hope i made my self clear, if not please let me know :)

Thank you very much ]]>

I have a multivariable logit regression model that has two terms showing two-way interaction (the interaction also makes biological sense.)

So what am I to do about it?

Do I just mention it in my findings?

Do I remove one of the variables? (How do I decide which one?)

I understand that interactions weaken the goodness-of-fit of a model. True?

Many thanks!

Johann ]]>