I did not understand what you have. You have 30 different samples of 30 observations? Can you please explain?
Hi all,
please help me I am stuggling doing my data analysis of my poject and my supevisor is on holiday but I have to finish all the ananlysis in two week time.
I am looking for which test is more suitable to test my hypothesis. .I have one group and the comarison is between samples in this goup, what i want to confirm is that the variation of results between my samples is differ significantly. For example, I have 30 samples and I measured the number of cells in each sample and there was huge differences between them, some have vey high number and some have very low. So, I want to pove that this difference between the samples is significant. All the tests used are comparing two sets of data, and even for one goup test is comparing befoe and after or compare the mean of sample with previously mensioned mean but I only have one goup of data with one measuement.
The only way I can think of is to calculate the mean and then compare the data with it and see if the spread is significant using one sample t-test? is that true? if not So which test is more suitable?
Please help if you have any info
I did not understand what you have. You have 30 different samples of 30 observations? Can you please explain?
...So you want to show you have variability in your data? Is that really what you want to show?
yes Dason that what I want to show, there is significant variation between my samples.
Define "significant" - If all you mean is that there is a non-0 variation then you don't even really need to test because... hey you have non-0 variation between your samples.
So yeah... what do you mean by significant variation
There is no difference in what you mention. If you had two groups you could look for differences in the means but as it is right now you aren't looking at differences.
Maybe you could do well by taking a step back and explaining what your research question is. What are you trying to show. Tell us about that kind of stuff.
If all you're trying to show is that the cell counts you get for different samples is not the same then you're not going to tell anybody anything they didn't already know.
In clinical pactise we found that some people after recieving blood have very quick cell count recovrey whereas some people have very slow recovrey which may takes years and they may die because of that. Many studies analysed different factors such as the age, sex, prophylaxis and other but non was correlated. Therefore, our research was aiming to see if the number of the responsible cells vary in the blood between our samples. We know they are not the same of course but we wanted to see if there is a big difference, so that people recieving high dose recover quickly and verse versa (this could be done in a different study). Do you have any other suggestion? if I reported the SD only, would that be infomative? if not how can I make it infomative???
You have 30 samples, how many observations in each sample on average?
we test each sample twice in order to increase the accuracy of our results
So you have 15 samples each tested twice How many observations in each sample????
no 30 was an example, I have actually around 50 samples and each was tested twice (so 100 in total). It was done only by me so one observation. )
Ok, finally we have it, you have 50 observations measured twice and want to see if the variance of the first and or the second measures is significant. What I suggest you do is for each sample construct a 95% confidence interval for the variance. Also perform a test to see if your values in each group come from a chi-square distribution. Then, the question is, how large is variance or the standard deviation to be thought of as large? Do you knwo from previous studies?
sorry why do we need to do chi-square distribution? do you mean that we need to see if the data are normally distributed?
and no there is no any data is available from any previous studies
Tweet |