Body size (total length or body mass) estimates, derived from linear measurements of skeletal elements (e. g. skull length, molar length, long bone diameter), are commonly used for paleobiological studies.
There is a trend to use Reduced Major Axis (RMA), instead of Ordinary Least Squares (OLS), due to the asumption of error distributed betweeen both the independent and dependent variable in RMA, instead of being considered only for the dependent variable as in OLS. However, some authors (e.g. Kilmen & Rodriguez, 2017) consider RMA has additional problems, and OLS can be used for allometry studies if the error of the independent variable can be ensured as low.
Please note that my question applies only to body size estimates derived from linear measurements. The choice might be different for other applications of linear regression.
Kilmer, J.T. and Rodríguez, R.L. (2017), Ordinary least squares regression is indicated for studies of allometry. J. Evol. Biol., 30: 4-12. doi:10.1111/jeb.12986
Jorge -
I think you might research "errors
-in-variables" models which I think look at errors in the independent variables by making repeated measures. Not my area, but it sounds promising here.
Also, have you considered heteroscedasticity?
For regression of form y = y* + e:
https://www.researchgate.net/project/OLS-Regression-Should-Not-Be-a-Default-for-WLS-Regression
Cheers - Jim
Given the 1st answer you may like to see the attached reprint...
Hint: do the correlation matrix of your predictors. There are likely to be high correlations....further take the inverse of the above matrix. the diagonal elements of the inverse matrix are called Variance Inflation Factors (VIF)...if any of the VIFs is greater than (say 10) then OLS equation is likely to be faulty. Ridge regression and/or PCR: principle component regression are then options...
Do you mean to actually estimate the body size of an organism? Or analyze data on body size for allometric analyses, species comparisons or some other question?
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics