Comparing two groups of dates

[Chris Chinnock (Chris)]

Hi all, 

Riding on the wave of the helpful responses I received yesterday, I have another question.

One o fthe projects I am working on is a 10th-11th century cemetery in the UK. I have received a series of C14 dates spread across the cemetery, targeted at trying to get a good spatial distribution. The dates all look good and I have been able to plug them into a Phase model easily enough. 

The cemetery is mostly clustered around a single cell chapel though there is a distinct group of burials that appear to be clustered around another feature, possibly a tree. During excavation it was suggested that the tree may have been the original focus of burial with the chapel being added at a later date. The cemetery was only in use for a short period of time and the dates from both groups look broadly similar, superficially. What I want to know is whether there is a model I can build that would be able to ascertain with any confidence whether one group of burials may be earlier than the other?

Thanks in advance

Best

Chris

[Derek Hamilton]

Hi Chris,

One way to do this is to model the group around the tree independently of the remaining burials - so the equivalent of two 'overlapping' Phases in OxCal.

Then use the Order parameter to compare the start and end probabilities for each group.

Depending on the number of dates for each group and how precise the model results are this might prove informative for your narrative - potentially enabling you to make a statement such as "There is a XX% probability Y burials began before Z burials."

I hope that helps.

Best wishes,

Derek

[Chris Chinnock (Chris)]

Thanks Derek, 

That sounds like what Im looking for. I'll look into it.

Thanks

Chris

[Chris Chinnock (Chris)]

Hi Derek, 

I think I have been able to construct a model which works. i found a couple of earlier conversations around using the order parameter and have followed those conclusions to get the following (see below for code). I think I have done this right so that the model isnt assuming that one phase is earlier than the other, right? its just that the results seem a bit too good to be true!

I think that the order command has also worked. My only problem is I dont know how to read the results. I click on the little raw data icon which brings up the probability table but Im afraid my very limited stats knowledge is being stretched beyond capacity. im sure its a dumb question but how do I convert this to the "There is a XX% probability Y burials began before Z burials", you mentioned above?

Thanks again

Chris

 Plot()
 {
  Sequence()
  {
   Boundary("Start Tree");
   Phase("Tree cemetery")
   {
    R_Date("Wk-10654/B41",1110,60);
    R_Date("Wk-10655/B43",1090,60);
    R_Date("SUERC-126968/B34",1114,27);
    R_Date("SUERC-126972/B50",1159,27);
    R_Date("SUERC-126973/B33",1101,27);
    Span("Use - tree");
   };
   Boundary("End Tree");
  };
  Sequence()
  {
   Boundary("Start chapel");
   Phase("Chapel cemetery")
   {
    R_Date("Wk-10657/B67",1100,70);
    R_Date("WK-10656/B65",1060,70);
    R_Date("SUERC-126967/B30",1033,27);
    R_Date("SUERC-126969/B9",1050,27);
    R_Date("SUERC-126970/B53",1087,27);
    R_Date("SUERC-126971/B22",1045,27);
    R_Date("Beta 507486/B6",1010,30);
    Span("Use - chapel");
   };
   Boundary("End chapel");
  };
  Order("Tree vs chapel")
  {
   Date("=Start Tree");
   Date("=End Tree");
   Date("=Start chapel");
   Date("=End chapel");
  };
 };

[Derek Hamilton]

So - I quickly copy-paste and ran your model and my table is attached (hopefully).

To change the probabilities to percentages just multiply by 100.

From this table you can say:

"There is a 98% probability Start Tree occurred before Start chapel and a 70% End Tree occurred before Start chapel."

For me, based on your data, a narrative supported by the model is one where burial began around the tree initially and then spread out from there.

Derek

[Chris Chinnock (Chris)]

Hi Derek, I was just in the middle of responding. You've confirmed what I suspected was the case. Thanks a lot for laying it out. This is a real turn up for the books. I'm very pleased with those results.

Thanks again for the help.

Chris