# Handling Negative wPLI² Debiased Values

**URL:** <https://mne.discourse.group/t/handling-negative-wpli-debiased-values/11668>\
**Category:** Support & Discussions\
**Created:** [January 21, 2026, 12:34pm UTC](https://mne.discourse.group/t/handling-negative-wpli-debiased-values/11668 "2026-01-21T12:34:23Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![SilvanaHuertas](https://yyz2.discourse-cdn.com/free1/user_avatar/mne.discourse.group/silvanahuertas/32/3030_2.png) [@SilvanaHuertas](https://mne.discourse.group/u/SilvanaHuertas)\
**Post date:** [January 21, 2026, 12:34pm UTC](https://mne.discourse.group/t/handling-negative-wpli-debiased-values/11668/1 "2026-01-21T12:34:23Z")

</div>

Hi everyone,

I’ve been working with wPLI² debiased and, as expected (Vinck et al., 2011), I sometimes obtain negative values. From what I understand, these negative values don’t indicate “negative connectivity” per se, but rather reflect issues related to small sample sizes.

I wanted to ask how you typically handle these small negative values in statistical analyses or subsequent processing. Do you clip them to zero, use them as-is, or apply another approach?

Thanks,

---

<div class="post-metadata">

**Author:** ![tsbinns](https://yyz2.discourse-cdn.com/free1/user_avatar/mne.discourse.group/tsbinns/32/4543_2.png) [@tsbinns](https://mne.discourse.group/u/tsbinns)\
**Post date:** [January 21, 2026, 5:17pm UTC](https://mne.discourse.group/t/handling-negative-wpli-debiased-values/11668/2 "2026-01-21T17:17:46Z")

</div>

Hi,

I haven’t worked with debaised wPLI before, but I did come across these old forum posts with a similar question and some suggestions: [[FieldTrip] Negative values of debiased wPLI](https://mailman.science.ru.nl/pipermail/fieldtrip/2015-October/009712.html) and [[FieldTrip] Clarification on plotting WPLI](https://web.archive.org/web/20240812142800/https://mailman.science.ru.nl/pipermail/fieldtrip/2016-February/022995.html).

Skimming through some papers using the measure, I didn’t see this mentioned in their methods sections. Maybe if you dig further you will find an example of handling this, or perhaps they just left the negative values as-is.

Sorry I can’t give a definitive answer. Maybe @wmvanvliet has experience with this measure?

---

<div class="post-metadata">

**Author:** ![SilvanaHuertas](https://yyz2.discourse-cdn.com/free1/user_avatar/mne.discourse.group/silvanahuertas/32/3030_2.png) [@SilvanaHuertas](https://mne.discourse.group/u/SilvanaHuertas)\
**Post date:** [January 22, 2026, 2:23pm UTC](https://mne.discourse.group/t/handling-negative-wpli-debiased-values/11668/3 "2026-01-22T14:23:02Z")

</div>

Thanks for your reply. I have been reading papers that use this metric, but it is not clear what they do or even if they get negative values. That is why I posted here, in case someone knows. Thanks anyway, I will continue my search.

---

<div class="post-metadata">

**Author:** ![wmvanvliet](https://yyz2.discourse-cdn.com/free1/user_avatar/mne.discourse.group/wmvanvliet/32/100_2.png) [@wmvanvliet](https://mne.discourse.group/u/wmvanvliet)\
**Post date:** [February 2, 2026, 8:59am UTC](https://mne.discourse.group/t/handling-negative-wpli-debiased-values/11668/4 "2026-02-02T08:59:55Z")

</div>

> [@tsbinns](#):
>
> Maybe @wmvanvliet has experience with this measure?

Sorry, I don’t ☹