one numerical and three categorical variables - which test?

Neta_Friedman

New Member
I am looking at the relation between the distance from archaeological sites to their nearest water source.
I have three types of water sources (the categorical data) and a list of many sites, each with a numerical distance value to the closest water sources.
I want to know if there's a connection between water types and the sites' distances to them.
Which test do you think would be best?

P.S.
I have the Jamoui software if that helps

Karabiner

TS Contributor
I am looking at the relation between the distance from archaeological sites to their nearest water source.
So each water source can possibly be the nearest to multiple sites in your dataset,
not just one?

How many sites and how many water sources are there in your dataset?

With kind regards

Karabiner

Last edited:

Miner

TS Contributor
If I understand your study design correctly, you have one categorical independent variable (Water Source type), which has three levels, and one continuous dependent variable (distance). A 1-way ANOVA would be a good place to start.

BTW, how do you know whether the water sources today are the same water sources that existed at the time those sites were active?

Neta_Friedman

New Member

@Karabiner - As you wrote, each water source may have one or more site connected to it.
There are 556 sites (though I might decide to divide this number into several smaller groups later), and there are 194 water sources that are divided into three categories (55, 58 and 81).

@Miner - I think you understood correctly, I'll go ahead and do as you suggested.
To answer your excellent question - I don't know, I'm relying on educated guesses. Assuming the basic rules of nature are the same, climate, hydrology, geomorphology and other influences follow specific physical laws. I looked at previous research (performed by people much smarter than me), focused on reconstructing the environment of the period I'm studying. This helped me reconstruct the environmental conditions quite accurately (I hope). It was a little complicated, but I think I now have as good a clue as one can get, for now.
In archaeology (as with proper science), we often have to make do with what we have, in many cases, very little, and correct ourselves when more data become available.

Karabiner

TS Contributor
Not completely sure, but I suppose that you have to account for the
clustering of sites with regard to sources. That would mean that you
create a multilevel model for your analysis where the distances are
nested within sources.

With kind regards

Karabiner

Neta_Friedman

New Member
Not completely sure, but I suppose that you have to account for the
clustering of sites with regard to sources. That would mean that you
create a multilevel model for your analysis where the distances are
nested within sources.

With kind regards

Karabiner
I'll look into that,
Thank you