Use independent t-test to compare th data and test for Statistical significance.
VJO
take a look on this definition;
Paired samples (also called dependent samples) are samples in which natural or matched couplings occur. This generates a data set in which each data point in one sample is uniquely paired to a data point in the second sample.
Paired is usually when u have to compare pairs... so if u have 2 samples from lake A and u want to compare it with 2 samples from another lake B... then u would use a paired t test... but if u r comparing just in one lake u could consider wet and dry as 2 different variables and compare using unpaired t-test
I think the answers from Fathima Minisha and Valentine Joseph Owan are not correct! As Aminu Suleiman Mohammed pointed out, paired/matched/dependent samples occur, if you measure the same object/subject more than once. It does not strictly have to be two times, it may be even more (like a longitudinal or time series analysis). Therefore, if you want to compare two different time points (dry vs. wet) at several lakes, a paired t-test (if assumptions are met) is suitable (you did not say it, but I assume that you have more than two data point from only one lake?). But you did not explain you design. How many lakes are there? Are lakes maybe nested within areas? Is the difference between lakes of interest? Etc. To give a proper advice for an analysis strategy, we need more information.
Discard my previous answer. I didnt understand the question initially well. Dependent t-test would be fine for the analysis because it's a paired data obtained from the source at different points.
Rainer Duesing only one lake. i took 3 water samples from 3 different part of the lake, in wet season and dry season. then, i want to know if there's a difference in phytoplankton's chlorophyll concentration that might be affected by seasonal change. The phytoplankton in wet and dry season might be consisted by different dominant groups, so I was told that i can use mann-whitney test, is it correct?
So you have 6 (3x2) samples all together? If correct, forget about any inferential statistics at all! Calculate the differences and report them descriptively. 3 paired samples do not make any sense for any inferential statistics (like t-test).
Rainer Duesing Yes, 3 samples for each season. but I forgot to mention that I also observed chlorophyll a, b , and c in every one sample. So I have 9 data per season, and 18 data in total (wet + dry season).
I've done the normality test for my data using Shapiro-Wilk test, and the value is < 0.05; isn't normally distributed. So I dont think i could use parametric test like t-test...
What does "chlorophyll a, b , and c" mean? Are they different kinds of chlorophyll? If yes then these are 3 different dependent variables. With large samples, this could be done with multivariate analyses, but in your case, you essentially only have 3 different measures (your sites) for each factor level of your indendent variable (dry vs. wet) for 3 dependent variables (chl. a, b,c). This does not make anything better (even worse). As I told you, just calculate the differences and present them.
The concept of "normality" for 3 measures doesn't even make any sense! With 3 measures (without ties in best case), you have one middle value and one left and right, respectively. Since in normal distributions, there is more probability density around the middle values, you have no chance to measure it at all, since with 3 measures there CAN'T be more mass in the middle. Ties would it make even worse, since normal distributions are symmetric. So please dont do any statistical test, just because you can click on them in any statistical program. Only because SPSS or which program, calculates something without warning message, does not mean that this makes any sense.
- Badges
- Science topic
- Similar topics
- Microbiology
- Scientific Research