# Thread: Statisticl analysis of 2 sets of data, HELP?!

1. ## Statisticl analysis of 2 sets of data, HELP?!

Hi.

For my work I have to compare 2 sets of data but I'm not sure what statistical test to perform on it?

The hypothesis is: Despite C. diff causing more deaths annually than MRSA, MRSA is more frequently reported in UK National Newspapers.

The data is:

Microorganism Reported in Media Associated Deaths
MRSA 889 1593
C. diff 511 8324

Is there any sort of statistical test on this?

Thank you so much!!

2. Are your numbers for one year, or the sums over more than one year? If your numbers are just for one year, then your hypothesis would just be for that year, and I do not think you would have to do any further statistics. Just say, "For 2008, MRSA killed more people, but was reported on less often." If you have more than one year of data, or can break it down by month, you might find interesting trends (maybe news reports come out a short time after an increase in deaths, and C.Diff reports are catching up to an increase in deaths while MRSA reports are still high despite a recent decline in deaths). You might do a 2-way ANOVA with time and deaths as the independent variables.

On more of a methodology note, I would recommend looking at hospitalizations, health care costs, and days of lost productivity in addition to just deaths, if you can get that data. Perhaps you could do a regression on what variables most determine frequency of reporting in newspapers. I am sure that other factors, such as age of patients, are important. At least in the US, MRSA causes big scares in schools, and people worry a lot about children (who typically have low mortality rates), so it is newsworthy, but C. Diff is far more common among the elderly, and since we expect the elderly to suffer more illness and have higher mortality rates, it is less newsworthy.

3. Unfortunately I wasn't given enough time to take into account hospitalisations etc, since this part of my dissertation was sprung upon me!
I had that written down for future work, in case some other poor person gets stuck with a similar dissertation.

Thanks for your help! I didn't think there was any test I could do, but my dissertation manager was pretty adamant there was something that could be done to prove that the data was "statistically significant". But I've searched and can't think of anything, and if you don't think there's anything either, I think I'm fighting a losing battle.
To be honest, I think the data alone shows a difference. but as I've noted in the discussion, the number of deaths doesn't take into account the age of the person dying, whether the infection was hospital or community-acquired etc, so more tests would have to be done anyway!