Two-tailed t-test help please!

**adiski** · 03-30-2015, 10:05 AM

I need to test the hypothesis that the annual global carbon sink is weakening over time by testing whether this hypothesis is supported by the data through comparing decadal means using a two-tailed t-test.

I have data of carbon sink uptake 1961-2011, so split the decades into 1961-1970, 1971-1980 etc and took the mean of the sink values and now have 2 columns like this:

1961-1970 1.410792418
1971-1980 1.559115845
1981-1990 1.800177877
1991-2000 2.212384716
2001-2010 2.500734866

With what variables do I perform the t-test? And how do I know my confidence level?

Thanks so much in advance!

**MrShorty** · 03-30-2015, 10:31 AM

It is sometimes difficult to give an answer to these kind of questions, because I am not sure if your question is a question about Excel or a more generic question about statistics.

It often seems that these questions are more about statistics than Excel. If that is the case, rather than re-write an entire lesson on t-tests, I will point you to this tutorial (or put something like "t-test difference of means" into your favorite search engine): http://stattrek.com/hypothesis-test/...-in-means.aspx From this, the main variables that go into this kind of t-test are A) the two means and B) the standard deviation/variance of each mean. The null hypothesis is usually "mean1-mean2=0" and the alternative hypothesis is "mean1-mean2<>0".

If you are looking for help with Excel's built in statistics functions (including the various T.xxx() functions, I would first suggest the main Excel help files: https://support.office.com/en-us/art...__toc309306716

Review that information. If there is something you are still having trouble understanding and implementing, let us know where you are stuck and we will help as best we can.

**adiski** · 03-30-2015, 12:02 PM

Thank you.
I have read both and there were quite useful but I am still unsure whether my variables to find a decadal mean is correct.
Maybe I am asking a purely stats questions and need to find a stats forum...

**MrShorty** · 03-30-2015, 12:26 PM

I am still unsure whether my variables to find a decadal mean is correct.

Without the raw data you used to compute these means, there is no way for us to know if yoiu calculated them correctly, either. Means are usually fairly simple =AVERAGE() or =AVERAGEIF() or =AVERAGEIFS() functions. If needed you can do SUMIFS()/COUNTIFS(). Maybe some other variation, depending on the the source data. Since simple arithmetic means are usually fairly easy to calculate, what is your uncertainty regarding your method of calculating these means?