# Correlation

#### maltesedog

##### New Member
I have a list of vehicle models and a number of costs associated with them.

For example:

Mazda 1000.50 euro
Peugeot 2000.23 euro
Mazda 1233.43 euro
Citroen 1323.34 euro
Mazda 1345.34 euro

What statistical tool should I analyse to verify if there is any relationship between the vehicle make and these costs.

Regards

#### bourne

##### New Member
well you need to have a brand image score as well. you can ask people about this. gather a survey that how much they like mazda (1 to 5 points for example)
and all other brands. then average out the score of each brand.
i think this is the simplest way. don't know any other way

#### maltesedog

##### New Member
Its not relating to liking.. these are costs associated with the car..

##### New Member
I don't know if this is doable for you but why don't you just input the data into a spredsheet like excel or minitab and do the correlation in that software. Its a lot easier. I don't know if youhave these applications available but you can download a 30 day trial of minitab wich is my favorite.

Sorry I don't really know what you mean by "statistical tool". But typically to my understanding if you want to know if there is a meaningful relationship between to variable you do a correlation test where the correlation coefficient or r-squared should range between +1 & -1 if it is 0 than there is no relationship, other wiser they could be strongly or weakly related in either a positive or negative way, hence the +1, -1. The closer to 1 either positive or negative the stronger the relationship.

Hope this helps, hope I'm right (I think I am).

#### Papabravo

##### New Member
It is always important to point out that correllation does not imply causality. Just because two things are correllated does not imply that they have any meaningful relationship.

There is no "statistical tool" that will provide you with a relationship between a character string and a number except the lookup table which you presented.

Last edited:

#### lekei

##### New Member
Arrange the makes and corr prices, then compare means for each statistically. Example it it were 2 makes I would use t test. If you have more than 2 you can use relevant test, I am not sure - analysis of variance.....

#### lekei

##### New Member
Following the test the higher mean value and lower mean value will give reflection of make and price

#### Papabravo

##### New Member
Arrange the makes and corr prices, then compare means for each statistically. Example it it were 2 makes I would use t test. If you have more than 2 you can use relevant test, I am not sure - analysis of variance.....
I don't think this is correct. There are three Mazdas, one Peugeot, and one Citroen. It is a bit of a strech to take a sample of one and say it approximates the mean of a population. Are you suggesting that if I name a price you will use the presumed correlation between price and model to predict the model with some degree of confidence? Pretty tenuous on this data.

##### New Member
If you have already determined that there is some type of relationship between the variable I think you should a F-test to test if the relationship if significantly relavent

#### lekei

##### New Member
Ok True. I thought there were several on each brand. If you have one price for each one brand then the test will not apply. Ill thin it over and will come back.

#### lekei

##### New Member
But I have seen from your list of example you have 3 brands of mazda. I think if you have about 1000 vehicles different brands mean values for each brand will reflect

#### maltesedog

##### New Member
There are around 50 cars in the population I have.
Some may be of the same brand.

#### maltesedog

##### New Member
Lekei I have not determined whether there is some kind of relationship.
I want to test whether there is any relationship!