Calculating t-statistic without raw data

[Yoha]

Can anybody help please!

Hi all,

I wonder if anybody would help answering my question.
If a researcher found a significant result using One-way ANOVA (F1,11 =15.29, P˂.01), What is the equivalent t-statistic value for F1,11 =15.29? what is the one-tailed and two tailed p value for the t-statistic?

A value of p<.05 was considered statistically significant.

I couldn't figure out how to calculate t-statistic because the author didn't provide the mean and standard deviation of both groups.

 

[chiro]

Hey Yoha and welcome to the forums.

It might help if you give a bit more context for your question.

ANOVA is used to compare means in a way that you avoid the Type I errors associated with t-tests in this context.

This page shows you how the values are calculated for the F-distribution given the data.

http://en.wikipedia.org/wiki/Analysis_of_variance#Logic_of_ANOVA

If you haven't been given any information about the variance/standard deviation or the means for the two groups, I don't think you will be able to get the parameters for your t-distribution.

You would have to expand out the definitions to see if you get at least two independent equations for the two variables of variance and mean of your associated t-distribution.

 

[Yoha]

Hi Chiro,

Thanks very much for your reply.

It’s a part of an experiment conducted to see the plasma concentration of Oxytocin after prolonged release of Oxytocin . The author also mentioned that there was a significant -ve correlation (pearson correlation test) = -.60, p<.05 between plasma and adipocyte (fat cells) size, and he provided a graph for both.
With regard to the concentration graph , he mentioned underneath the graph the following:
Plasma concentration of Oxy. In Westar rats treated with either Oxyt. (n=6) or saline =control(n=7) for 2 weeks . Data are expressed as means+- SEM, p<.01.
That’s all what he provided, there is no baseline data and no treatment data.
Here is the article link if you’re interested in knowing more

http://www.ncbi.nlm.nih.gov/pubmed/20846187

Thanks for your help!

 

[chiro]

Hey Yoha.

I had a look at the article and I noticed this:

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3031065/figure/fig03/

After reading the rest of the paper I couldn't find any specific information for the mean and variance, but if you can't get the exact numbers you might want to estimate them from the graphic above.

It's very odd that they do not include this data: it's very irresponsible in my view to hold back this kind of data for the reasons a) it gives other people to perform the calculation to check the results and b) it also provides a way for other researchers with domain experience to check the data for integrity.

Is this the norm in health or is this a one off kind of thing?

 

[Yoha]

Hi Chiro,

Thanks very much for spending a time reading the article.
You're absolutely right, I was going to do what you suggested in order to estimate the means and variances. With regards to the paper, it's weird! they didn't provide the raw data of their work which is a question mark, but, they published in a very good journal.I don't know how they accepted it.!

Anyway, many thanks
I appreciate your help.
Cheers!