Hi All,
we would like to know the statistical analysis of an experiment .
We have done two different methods in 5 samples(different sizes as x axis) and got yield of these two methods for each Sample.Is it possible to do the T-Test for this in excel or we have to go for ANOVA. Could you please guide us if anyone has a solution.Thanks in advance .
What are the characteristics of the variables within your "5 samples"? Are they categorical or numerical?
More info is needed, Betty.
Hello,
Please add more info about your data.
At a first glance, the t-test would be ok for the analysis (two methods=two groups to be compared) and you should be able do it in Excel. However, I must say that the sample size (5) looks small and most likely you'll get unreliable results.
Since you have only two experimental conditions being compared, ANOVA and t-test will have similar results. If the treatment, I dont know if I understood correctly, is a size, so you have a numerical effect and can apply the linear regression to know the slope. All statistics can be calculated throgh the Excel, but ANOVA requires the instalation of one extension or suplement, may be, I dont know how it appears in english version, called "Analysis Tools" or somethig like that. It can be installed in the option File -> Options -> Suplements -> Excel suplements.
These techniques requires numerical responses. If you have categorical responses other kind of statistics will be necessary, so we return to the Mohammed question.
Hope this helps.
Hello Betty,
I think Mohammed is right: More information is needed to provide good advice about what you should do.
That aside, I think you should consider using analysis of covariance (ANCOVA). That procedure takes into account the status of your groups before the intervention(s) as well as their status after. This is because quite often groups' original status is different despite attempts to match them. A t-test (or series of t-tests, perhaps preceded with ANOVA) will "deal" with only post-intervention differences in the groups. Sometimes researchers use t-tests (or ANOVAs) to determine whether the difference between pre- and post-intervention scores are greater in some groups than in others, but that procedure is often frowned upon because it doesn't really take any original differences into account. As I wrote above, that's what ANCOVA does - and it permits you to take other variables into account, too.
ANCOVA is a sophisticated procedure, requiring that a number of assumptions are met - but if you want to analyse your data effectively, I think it is worth trying to meet those assumptions.
Robt.
Hi
We have 5 different samples in whole size to nano material size i.e
The term "yield" makes me believe in a numerical response variable, so ok to T-test. But I can't see the pareament so, the t-test to two independent samples may be more appropriated. Unless the same sampled was splited in two sub-samples and one treatment was randomly attributed to each sub-sample, but I did not understood this way.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis