How to to test statistical agreement of calibrated dates?

[kuri...@gmail.com]

Hello all,

I have two sets of human bone radiocarbon dates. These are two group burials, and in each one, each  individual has from 1 to 4 dates (15 in total). I‘ve made combinations for both group burials. My command included R-Combine function for dates from the same individual and Combine function for different individuals. Now I have two calibrated dates for both group burials. But the archaeological context allows me to assume that these two burials are very likely synchronous (same day or so), or this is exactly what I need to test as a possibility.

Is there a way to test the statistical agreement of calibrated dates (for this is what I have now, as the Combine function returns calibrated dates, not BP)? Or maybe I can statistically compare both sets prior to combination? What command could I use?

Thank you in advance,

Laurynas

[MILLARD, ANDREW R.]

I think the best way is to use the Difference command to calculate the difference in time between the two calibrated dates. If the resulting range includes zero, then there is no evidence for a difference in time.

 

Best wishes

Andrew

--

Dr. Andrew Millard

Associate Professor of Archaeology, and

Designated Individual under the Human Tissue Act,

Durham University, UK

Email: A.R.M...@durham.ac.uk 

Personal page: https://www.dur.ac.uk/archaeology/staff/?id=160

Scottish Soldiers Project: https://www.dur.ac.uk/scottishsoldiers

Dunbar 1650 MOOC: https://www.futurelearn.com/courses/battle-of-dunbar-1650

 

 

From: ox...@googlegroups.com <ox...@googlegroups.com> On Behalf Of kuri...@gmail.com
Sent: 08 December 2020 07:50
To: OxCal <ox...@googlegroups.com>
Subject: How to to test statistical agreement of calibrated dates?

 

 

[EXTERNAL EMAIL] Do not open links or attachments unless you recognise the sender and know the content is safe. Otherwise, use the Report Message button or report to phis...@durham.ac.uk.

--
You received this message because you are subscribed to the Google Groups "OxCal" group.
To unsubscribe from this group and stop receiving emails from it, send an email to oxcal+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/oxcal/dc0b9e2a-d2f2-4c40-b31b-4cbd248fbd9bn%40googlegroups.com.

[kuri...@gmail.com]

Thank you, Andrew. Started reading about it. I think Difference command is exactly what I need. I suppose I have to build a Sequence. But I cannot figure out if I can include two Combined dates, which, on their part, are combinations of both R_Dates and R_Combined dates? Would be grateful for a reference to any example of a script.

All the best,

Laurynas

[MILLARD, ANDREW R.]

Laurynas,

 

You don’t need a sequence. This structure would suffice

 

Combine(“A”){ ….. };

Combine(“B”){ ….. };

Difference(“Difference A and B”, ”A”, ”B”);

To view this discussion on the web visit https://groups.google.com/d/msgid/oxcal/5fa15be5-8139-4661-89c7-47a0bc147780n%40googlegroups.com.

[kuri...@gmail.com]

Thank you,

This works, and Difference returned the Unmodelled (BC/AD)value from 11 to160. This, I understand, means that the dates are different, and the burials were not simultaneous. Will have to think what happened there.

Best wishes,

Laurynas

[Erik Marsh]

I would also try combining the two dates, because it will run a chi-square test. If they pass the chi-square test, then you have no statistical reason to believe they happened at different times. If the difference result of 11–160 years is at 95%, then it sounds like you're on the edge of them being statistically synchronous.

Another option: the Order command. This will give you the probabilities that one was prior to the other and vice versa. If the result is close to 50/50 they are more likely synchronous. Finally, you could try Correlation, very similar to Difference and with the same syntax, but with a nice graphical output.The manual has examples of the code for all of these queries (and you can run them all at once). If you're having trouble with the code, it's useful to post what you have so far and people can suggest corrections.

I would try this set of queries with the two group dates that you have, but also a separate trial with all 15 individuals. You may get interesting results comparing the individual burials without assuming they belong in two groups (this approach would able to better detect an outlier individual that might skew comparisons between the group dates). And in my book, models with fewer assumptions are preferred.

Erik

[kuri...@gmail.com]

Hello,

Thank you for your suggestions.

The attempt to Combine the dates from the two group burials failed with the message: Warning! X-Test fails at 5% - Combine burials 162 and 336/337 X2-Test: df=1 T=11.243(5% 3.8); Warning! Poor agreement - n=2 Acomb= 7.4%(An= 50.0%).

The Order command gave this result:

Order burials 162 and 336/337

Probability t1 < t2

I understand these show that the burials are unlikely to be simultaneous. The Correlation gives a blank graph.

Will continue to work with all dates from one group burial, all dates from both burials, etc.

All the best,

Laurynas

[kuri...@gmail.com]

My mistake, re Order results:

Order burials 162 and 336/337

Probability t1 < t2

t1                                            t2

                              Burial 162   Burial 336/337

Burial 162            0                   0.02827

Burial 336/337    0.9717        0

Laurynas