3
$\begingroup$

I have 5 categories rock samples and I am trying to understand how their fossil preservation differs and whether the difference is statistically significant. I determined the presence or absence of fossils using a spectroscopy tool in rocks representing 5 sample categories. I obtained the percentage of positive occurrences. I then repeated the measurement 10 times in each category and obtained a mean percentage for each category.

  • What test do I use to understand (1) if and (2) how the means of the 5 categories are statistically different from each other?
  • Can I then rank the categories in their preservation potential to contain fossils?
  • Is repeating the test 10 times per category enough?
$\endgroup$
  • $\begingroup$ Do I understand correctly that you have 10 yes/no measurements from each of 5 different types of rocks, & you want to know if the proportion yes differs? Re how much data you need, that is a question about power-analysis; you don't have enough information here to answer it, but if you click the link you can read our threads on the topic. $\endgroup$ – gung - Reinstate Monica Mar 28 '15 at 2:28
  • $\begingroup$ Gung, I have 5 categories of samples and 10 is an arbitrary number of measurements I was going to make on each to determine a yes/no presence of a phenomenon. Yes, I want to know if the categories have a statistically significant difference in "yesses" or presence of that phenomenon. $\endgroup$ – Anonymous1717 Mar 29 '15 at 5:46
2
$\begingroup$

Your data are appropriate for a chi-squared test. You can form a contingency table by entering the counts of each type that did and did not have fossils in them. For example:

            marble  quartz  granite  sandstone  mica
w/ fossils       5       1        6          8     1
no fossils       5       9        4          2     9

N.B., if those were your numbers, the results would be:

        Pearson's Chi-squared test

data:  as.table(x)
X-squared = 15.9278, df = 4, p-value = 0.003118
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.