How do you know if there is a significant difference between two or more samples?I wanted to know if there is a significant difference between samples by comparing measured values, normally I had to use the student test or ANOVA right?
AnalyticsData TableRe: How to know if there is a significant difference between two or more samples?Hello @Syrine ,
There are several aspects to consider when choosing a statistical test:
- The assumptions related to the parametric test (Student, Tukey-Kramer, ...) are they respected? (If not, use of non-parametric tests):
- The data measured are independent (otherwise it is necessary to carry out tests for paired samples),
- The measured data are normally distributed (otherwise choice of non-parametric tests),
- The groups compared have a similar variance (otherwise choice of tests taking into account a difference in variance between groups).
- How many samples or groups are compared? : If the number of groups is greater than 2, we will prefer tests whose alpha risk is adjusted in relation to the number of comparisons to be made, which makes it possible to reduce type I errors (corresponding here to "finding a significant difference between samples when there are none"). If there are only two groups/samples, the Student test as a parametric test or the Mann-Whitney/Wilcoxon test as a non-parametric test should be suitable.
Overall, it gives this decision flowchart:
Note, the French-speaking JMP steering committee has conducted a webinar on this subject in the "Tips and Tricks" series: Video JMP Addict: Workshop tips and tricks Special session Workflow analysis of ... - JMP User Community
Finally, if you have an anonymized dataset or an example on which to base your case, more precise and adapted answers can most certainly be provided to you.
Hope this first answer helps you,
Re: How to know if there is a significant difference between two or more samples?Yes, thank you very much, it's clear.
Actually I have a question very far from the subject on which we are talking about it, I wanted to make a mixing plan, I searched on your platform but I did not find a tutorial. I do not know how I should fix the constraints of each variable if you can inform me on this subject
And thank youuuuuRe: How do I know if there is a significant difference between two or more samples?Good evening @Syrine ,
Regarding the experimental design you want to carry out, I think it would be better to open a new topic (or continue the discussion on the topic you had already opened) to explain precisely what you want to do, your factors, your experimental constraints on the factors (number of experiments, some factors have a fixed total quantity, ...) and your needs: what type of model do you want? Screening, or a little more detail with interactions, quadratic effects, or models adapted to mixing designs like reduced cubic Scheffé models?
- This would allow other community members to help you and better readability/clarity with a single post for each topic.
You had also opened a topic for an experimental plan that you wanted to carry out and on which I had answered you and given some sources to better understand the experimental plans with JMP: Re: What is the experimental plan that I should choose? - JMP User Community
- Did the initial answers and links provided help you? If you need more information on a particular aspect, please feel free to reply to your dedicated Experiment Design topic to provide more details on what you want or explain where you are having difficulty/stuck.
Glad these answers helped you. To help other JMP users find answers to their questions more easily, you can "accept it as a solution" below each answer to your post (see screenshot below): this allows you to highlight it/them just below the original post with the question, so that other users can more easily spot the solution(s) among all the answers provided.
Feel free to join the French-speaking JMP user group on this community; the various topics presented there may inspire you: https://community.jmp.com/t5/Groupe-francophone-des/gh-p/france-jug
Good evening,
Re: How do I know if there is a significant difference between two or more samples?Good morning,
Okay, if I need to, I'll open another chat session on the mixing plan.
Let's go back to this same subject of difference testing, I have a question about R^2 if the value is low so the model does not allow to explain the variability well. Does this pose a problem or can we make improvements?
Re: How to know if there is a significant difference between two or more samples?Hello @Syrine ,
What dataset/analysis/test, etc... are you referring to?
Do you have a practical case study as an example?
Re: How to know if there is a significant difference between two or more samples?Good morning,
I am sending you the data table as well as the results of the difference test (Tukey) that I performed.
but I wanted to know if in a case of low R2, we cannot disregard these results.
