For example: ROA (%) = 0.25 + 0.15 (Debt/Equity) +0.25 (Market to Book ratio) + e
Muhammad Haseeb
I am expecting that you are using multiple regression using OLS.
With that I don't find any problem, if you regress the data using dependent variable as percentage (25%) the Beta coefficients will adjust to reflect the impact of it, In case you use normal number(0.25 for same 25%) then your Beta coefficient will adjust as per that, assuming that your independent variable format is ratio and data is same in both the iterations.
You can use/run the natural logarithms of all variable to bring them to a common base.
Dear Muhammad Haseeb,
We have published research on the systems classification and systems integration of open set and closed set models (Mallick, Krichel & Novarese (2010)) on www.researchgate.net/profile/Soumitra_Mallick, in which we have Econophysically and both experimentally and mathematical-statistically, derived the relation between open and closed models. This leads to many mathematical, statistical, physical and economic issues in intrepretation or in other words systems integration of the experiment using categories discovered and hence you can safely I think multiply the r.h.s. variables by 100 and do the scaling (somewhat quantum scaling) as we have argued elsewhere. However, the results will be different but can be intrepreted.
Soumitra K Mallick(CA)
for RHMHM School
The results are not affected by whether you use percentage or ratio. The coefficients will adjust themselves. What is more important is the significance of each of the variables in terms of the predictive power of the model.
If it is multiple OLS then One unit change in independent variable causes beta change in dependent variable. In other words, one unit change in Debt/Equity (whatever % or without) increase 0.15% ROA (if you measure in %).
Thanks Umar Farooq Attari, I was also doing the same interpretation but was confused.
Remember dependent and independent variables must be in the same unit of measurement to ease interpretation of research result, and make the interpretation meaningful. In this case, natural logarithms of all variables will help to bring them to a common unit of measurement. Good luck.
- Badges
- Science topic
- Similar topics
- Books