Data, data everywhere

Apparently I really was doing something horribly wrong a couple months back when at least three dozen of my strontium samples failed miserably in the SEM. One of the geochemists emailed me a crapload of data this afternoon: at least 50 new results. Only one sample failed, but he's rerunning it tonight to see if the other half can give good data. Now I have 111 unique data points from a combined population of about 215 ancient Roman skeletons. And I have no idea what to do with them.

If I put all 100+ results on a scatterplot, there are only 5 clear outliers: three lower than the Roman range, and 2 higher than that range. 4 of the 5 are males older than 30, 1 is a kid around 11-15. But there has got to be fancier statistical analysis that one usually does with these numbers. I mean, given that there was both short-distance and long-distance migration in Rome, the question becomes: how do I sort out the short-distance migrants from the locals? And given that Sr ratios vary more or less continuously from north to south along the Italian peninsula, shouldn't I expect my data to fall into a more or less continuous range? Maybe I'll do some cluster analysis tomorrow. Or just go talk to the geochemist and see what suggestions he has for analysis.

It'll be sad if I only get 5 immigrants out of 6 months of lab work and $5,600 of grant money, mostly because I really want to write a dissertation on immigrants, but also because these results don't bode well for future Sr analysis in Rome. If the largest Sr study ever done in archaeology only manages to find an immigration rate of 5% (when we know from ancient demography that it should be more like 20%) with 100% sample coverage, what hope is there for people who can only afford to sample 10% of their population?


Unknown said…
hey thanks for the nod! I appreciate your encouragement and am proud to be among such a list of cool people.
dissertation conclusion samples

Popular Posts