Generating continuous variables from catagorised data
Posted: Wed Mar 16, 2011 6:30 am
Hi,
I am struggling with data, I have data on income for 2 different groups (for simplicity call them groups A and B). Group A is fine, it has the income in continous variable i.e. the exact figure. Group B is posing a massive problem to me, income is banded into 11 different bands for example: '150-199per week/7800-10399 annual', I need to change this data into something suitable for regressing in the format group A= c +B*group B. Some academics fit a singh madalla / Burr distribution but I can't find that on Eviews. My actual lecturers have mentioned using dummy variables, which has confused me a little. Even then how would i make the data in each cell say the midpoint e.g '9000' instead of '150-199pw/7800-10399 pa' which clearly is meaningless.
I appreciate any help in advance.
Stuart.
I am struggling with data, I have data on income for 2 different groups (for simplicity call them groups A and B). Group A is fine, it has the income in continous variable i.e. the exact figure. Group B is posing a massive problem to me, income is banded into 11 different bands for example: '150-199per week/7800-10399 annual', I need to change this data into something suitable for regressing in the format group A= c +B*group B. Some academics fit a singh madalla / Burr distribution but I can't find that on Eviews. My actual lecturers have mentioned using dummy variables, which has confused me a little. Even then how would i make the data in each cell say the midpoint e.g '9000' instead of '150-199pw/7800-10399 pa' which clearly is meaningless.
I appreciate any help in advance.
Stuart.