In the data I have one dependent variable, "alternate travelling mode"，having 04 choices
01= Walking
02= Cycling
03= Both
04 = Neither
is MNLR model is valid for this case ?
I was thinking on the 03 i.e both walking and cycling it is related to the upper two choices 01 and 02.
if MNLR is not valid then how to solve this case?
You shouldn't use "both" and "neither" simultaneously, it is against Gauss-Markov suggestion on independence of the input variables. They are linearly dependent: 03+04 = 1 in all possible cases. Actually, the 4th variable appears to be equal to 1 in the case of all other variables are equal to 0. In fact you don't need even the 03, because if 03 = 1 then 01 = 02 = 1, and if 03 =0 then 01 = 02 = 0. So to code 4 considering cases you need just 2 binary variables.
@Vladislav Shchekoldin thank you i want little more clarification;
if I have
01= Walking
02= Cycling
03= Both 01 and 02
04 = Other transportation modes
as you mentioned if 03 = 1 then 01 = 02 = 1 but if 03 =0 then 01 or 02 is 0 or 1.
now 03 can be considered or I have to omit 01 and 02 or 03.
Hello, Muhammad
you have 4 cases:
01 = 0, 02 = 0 --> no walking, no cycling --> this is the case 03 =0
01 = 1, 02 = 0 --> walking, but no cycling
01 = 0, 02 = 1 --> no walking, but cycling
01 = 1, 02 = 1 --> walking and cycling --> this is the case 03 =1
So you dont need to use 03 at all.
initially you mentioned 04 means "neither". but if it is other kind of transportation, you could code this case as 01 = 0 and 02 = 0, because I can't imagine the another situation instead of walking, cycling, or other transpostation modes, only if there are no transportation at all, but maybe it is not the case )
thank you it was very helpfull.
if all the above four choices are presented to a respondent as discrete choice only that is respondent has to chose one option only, then in such case 03 will be valid or not?
you don't need to ask respondent on 03 or 04 because their values are evident after determination of 01 and 02 values. So, 03 and 04 seems unnecessary, and you don't need to exaggerate your questionnaire with extra questions.
Dear Mr. Usama,
One way is to set up alternate hypothesis to test the validity of your regression equations with respect to your i.i.a. assumptions. You can check on our research paper Mallick, Hamburger & Mallick (2016, 2017, 2018) and Mallick (2011) which tegether have been used to prove the Millenium Prize problems and also the quantum mechanics of String Theory in AGGNNNetworks which are fundamental Physical Networks as our school has discovered. This is the String Duality which in some sense can quantum mechanically jump the irrelevant alternatives using our discovered Higgs-Mallick quantum mechanics in the Econophysics cybernetic circuit field of the RHMHM School. Sorry for a convoluted answer.
Soumitra K. Mallick (CA)
for Soumitra K. Mallick, Nick Hamburger, Sandipan Mallick and others
RHMHM School of Mathematical Sciences Thinking
USA, Japan, India
- Badges
- Science topic