MKDA Chi Squared tutorial, December blog post, and How to run instructions #10
Conversation
|
TODO: Rename NiMARE outputs maps |
|
|
||
| On Neurosynth.org, we renamed the pre-generated forward and reverse inference maps; they're now referred to as the "uniformity test" and "association test" maps, performed by the MKDA Chi-Squared algorithm. | ||
|
|
||
| Although the method we used hasn't change,d the latter names more accurately capture what these maps actually mean. It was a mistake on our part to have used the forward and reverse inference labels; those labels should properly be reserved for probabilistic maps generated via a Bayesian estimation analysis, rather than for z-scores resulting from frequentist inferential tests. These maps are more difficult to interpret and use correctly, which is why we don't currently support this approach. |
There was a problem hiding this comment.
NiMARE still creates posterior probability maps, so (carrying on from Slack) we should instead provide guidance on how to interpret them. And maybe include a caveat here that they're distinct from statistical significance?
There was a problem hiding this comment.
Will disable by default. We may want to add more info on how to interpret them to NiMARE
|
|
||
| Next, select the annotation inclusion column you want to use, as before (by default, the "included" column will be used). | ||
|
|
||
| Now, select a reference dataset from the dropdown list below. The Neurosynth dataset represents the latest release of the legacy *Neurosynth* dataset (version 7), released July, 2018. The *Neurostore* dataset represents the latest update of our continuously updating "live" dataset, spanning over 20,000 neuroimaging studies. |
There was a problem hiding this comment.
It's probably worth noting that there are issues with comparing a manually-curated sample against an automated reference dataset, such as sampling biases in either and coordinate extraction. I don't think that merits more than a small admonition box.
Another concern is that you'll want to exclude any relevant studies from the reference dataset, so users should probably make at least some effort to exclude studies investigating the constructs or tasks they are studying with their meta-analysis.
There was a problem hiding this comment.
Also, as you mentioned on Slack, you could add in a small admonition saying that users could employ this estimator to do a two-sample comparison like the ALE Contrast analysis.
There was a problem hiding this comment.
This is a good point but I'm not what the answer is.
The workflow will exclude studies from the reference sample that you have in your meta-analysis.
I think the main issue is if you don't include all relevant studies are just a subset, right?
Trivial and content edits to MKDA association
Co-authored-by: Taylor Salo <[email protected]>
Co-authored-by: Taylor Salo <[email protected]>
Co-authored-by: Taylor Salo <[email protected]>
blog/2023-12-21-december.md
Outdated
|
|
||
| We're excited to share our December updates with you! It’s been a great month, having announced the release of Neurosynth Compose on social media. In the last month we’ve seen **over 1,200 new user visits**, with **200 users signing up for an account**! 🚀 | ||
|
|
||
| Help us keep this growth going by [sharing our announcement](http://localhost:3000/compose-docs/blog/announcing-ns-compose) with your colleagues. 🧑🔬 |
There was a problem hiding this comment.
should this be a relative link (e.g., ./2023-12-21-december.md) instead of a URL?
adelavega
changed the title
TODO: