[FieldTrip] Correct wPLI analysis

[Eelke Spaak]

Dear Daniel,

Even though in principle wPLI should be unbiased w.r.t. trial count, I
would not expect to get very informative results from a contrast
between a condition with 1000 and one with 5 trials. Specifically, the
estimate (wPLI or otherwise) for the condition with only 5 trials will
be rather poor. You could always resort to downsampling the condition
with 1000 trials (i.e., randomly pick 5 trials from that one) to get a
feeling for the quality of the data you are dealing with.

Best,
Eelke

On Fri, 22 Nov 2019 at 09:32, <daniel.strahnen at uni-ulm.de> wrote:
>
> Dear Fieldtrip-Community,
>
>
>
> I have a question regarding the correct interpretation of my wPLI-results.
>
> I am comparing the wPLI obtained from 1.4 seconds trials from a behavioural task in mice where they can either do a correct or a wrong choice. The
>
> sampling rate is 1kHz.
>
> Is it then safe to statistically compare results (t-test, ANOVA) from the correct choice with the wrong choices even if the number of trials differs between these
>
> two states, e.g. 100 trials vs. 5 trials. Or is the low-trial-number wPLI too biased at all?
>
>
>
> Thank you very much and best regards
>
>
>
> Daniel
>
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://doi.org/10.1371/journal.pcbi.1002202