When dealing with nonparametric continuous data that does not follow a normal distribution, you have a few options for conducting correlation analysis. The choice of method depends on the specific characteristics of your data and the goals of your analysis. Here are some common methods for correlation analysis with nonparametric data:

1. Spearman's Rank Correlation:

When to use: Spearman's correlation is a nonparametric measure of association that assesses the monotonic relationship between two continuous variables. It does not assume linearity and is suitable when the relationship between variables can be described by a monotonic function (e.g., increasing or decreasing).

· Advantages: Robust to outliers and does not require normally distributed data.

· Considerations: It only measures the strength and direction of monotonic relationships and may not detect other types of associations.

2. Kendall's Tau Rank Correlation:

When to use: Kendall's Tau is another nonparametric correlation method that assesses the strength and direction of association between two variables. It is suitable for detecting monotonic relationships.

Advantages: Like Spearman, it is robust to outliers and does not assume normality.

Considerations: It is less affected by tied ranks but can be computationally more intensive than Spearman's.

3. Distance-based Correlations:

When to use: Methods such as the Distance Correlation or the Maximal Information Coefficient (MIC) can be used for nonparametric data when you want to capture more complex relationships.

Advantages: They can capture nonlinear associations and are not limited to monotonic relationships.

Considerations: They may require larger sample sizes and can be more computationally intensive.

4. Bootstrapping or Permutation Tests:

When to use: If you have nonparametric data and want to assess the significance of the correlation, you can use bootstrapping or permutation tests in combination with Pearson's or Spearman's correlation coefficient· Advantages: Provides a way to obtain p-values for your correlation estimates without making strong distributional assumptions.

Considerations: Requires resampling your data and can be computationally intensive.

5. Visual Exploration:

When to use: Sometimes, a visual inspection of scatterplots or other graphical representations of the data can provide valuable insights into the relationship between variables, regardless of their distribution.

There ist nothing like "non-parametric data".

Fraud alert : Shital Choudhar 's answer is simply copied and pasted from ChatGPT.

This means that Shital has no idea whether the information and advice in the answer is right or wrong, helpful or misleading. And they don't care.

I'm not sure that a correlation analysis is a very helpful way to look at your data. A correlation does not make any assumption about the direction of the relationship. Most researchers have some idea about how the variables interact – at the very least, they will know which is the outcome of interest and which variables potentially influence it.

Even with families of variables such as scores on items on a scale, correlations are not a great way of exploring the scale structure.

Tell us a little more about your research question and your data. This might help to home in on more specific recommendations.

You will notice that ChatGPT can't think critically. It simply told you how to do something without knowing why you were doing it. And, of course, Shital doesn't care.

I looked at some of her (Shital's) previous replies and cross-referenced the question to ChatGPT, she does seem to have a history of ChatGPT answers but not 100%. Some sentences are 100% identical sometimes but she seems to re-arrange & recombine the info from ChatGPT. A curious case.

Beyond Shital Choudhar pasting chat bot errant answers (as noted by Jochen Wilhelm the notion of non-parametric data doesn't make sense), this is also a near duplicate question by Serkan Özdemir (https://www.researchgate.net/post/Which_correlation_analysis_should_I_use_for_non_parametric_continuous_data/1). Perhaps some broke into both Shital Choudhar and Serkan Özdemir accounts to make them look bad.

Daniel Wright it is for this reason that I am not very optimistic about the future of humanity. Every time we develop something that might benefit humanity – and goodness knows, we are in a pretty dire situation right now – many people's first reaction is to use it to exploit others for personal gain. For a while – quite a while – ResearchGate seemed like a rare instance of the idealism with which the internet was started. But, as chairman Mao observed, patriots come first, then profiteers.

Unless RG acts to preserve the integrity of the community, parasites like Shital Choudhar will drain any honesty and integrity from it. Maybe we should walk away now.

Don't let the few bad apples ruin the goodness coming from the (collective) productive efforts of rest of us. I think your contributions are worthwhile effort Ronan regardless of these "NPC" people. Perhaps they will give up the game sometime soon after peddling with this new tool?

I wish I believed Can Kiessling more than Ronán Michael Conroy , but unless RG does something (like return to allow down voting) a few bad apples does make the whole things less useful.

Being able to down-vote and perhaps in the "Report" drop down button (next to "Share" button) being able to mark posts as ChatGPT-influenced answers could be a good way to start filtering these posts and build up a demerit system. I wonder if anybody even asked ResearchGate.

Can Kiessling I've reported users to RG. I always get an automated acknowledgement, but very few of them have been subsequently deleted. RG is asleep at the wheel while ChatGPG is eating their business model.