# Quantitative or qualitative independant variable question

• May 17th 2011, 04:51 AM
TheRobster
Quantitative or qualitative independant variable question
I'm performing a linear regression analysis of the form y = mx + c where y is the amount of water a house uses and x is the number of occupants (in the range of 1 to 6).

My question is: is the occupancy rate counted as quantitative or qualitative? Clearly I can't have 2.5 people but as each value relates numerically to others maybe this means it is quantitative and not categorical?

Can I do a 'normal' regression analysis with my data, e.g. using Excel to create a trendline and associated formula. Would the corresponding R^2 value mean anything? Would it actually explain the % variation in my model?

-Rob
• May 17th 2011, 06:27 AM
CaptainBlack
Quote:

Originally Posted by TheRobster
I'm performing a linear regression analysis of the form y = mx + c where y is the amount of water a house uses and x is the number of occupants (in the range of 1 to 6).

My question is: is the occupancy rate counted as quantitative or qualitative? Clearly I can't have 2.5 people but as each value relates numerically to others maybe this means it is quantitative and not categorical?

Can I do a 'normal' regression analysis with my data, e.g. using Excel to create a trendline and associated formula. Would the corresponding R^2 value mean anything? Would it actually explain the % variation in my model?

-Rob

Since the mean of categorical data is meaningless ask yourself does the mean ocupancy have any meaning?

CB
• May 17th 2011, 07:28 AM
TheRobster
Yes it does (mean occupancy rate for housing is used all the time in demographics).
• May 20th 2011, 06:42 AM
TheRobster
Wouldn't the data actually be considered to be ordinal rather than nominal? Couldn't it therefore be ranked and then assessed? (Since there is a relationship between the occupancy rates, i.e. 2 is twice as many as 1, 3 is more than 2 etc).
• May 20th 2011, 09:52 AM
CaptainBlack
Quote:

Originally Posted by TheRobster
Wouldn't the data actually be considered to be ordinal rather than nominal? Couldn't it therefore be ranked and then assessed? (Since there is a relationship between the occupancy rates, i.e. 2 is twice as many as 1, 3 is more than 2 etc).

The number of occupants of a building is a numeric variable, just like a length (you can rank lengths ..)

CB