Editing Talk:Is it chance? Use a T-Test to identify how likely an intervention worked

== potential page merging == 

Hi! This is a lot like my topic https://wiki.openhumans.org/wiki/Finding_relations_between_variables_in_time_series. Combine somehow? - [[User:DG]]

: Given that standard t-tests like this aren't really used for time series (or at least not in the way described here) I'd keep it separate for now, but we could think about a larger 'meta'-topic about statistical testing? - [[User:Gedankenstuecke|Gedankenstuecke]] ([[User talk:Gedankenstuecke|talk]]) 09:02, 4 March 2022 (UTC)

:: Start with an explanation of T-tests then what apps use them then why they do not work on time series then solutions. As the wiki gets filled the statistical issue page will be split into multiple and some of the problems that arise from QS are not related to time series. [[User:DG|DG]] ([[User talk:DG|talk]])

::: I think I'll just go ahead and make a [[Statistical testing]] page as a general introduction to the topic and link to the time series & t-testing pages. - [[User:Gedankenstuecke|Gedankenstuecke]] ([[User talk:Gedankenstuecke|talk]]) 10:09, 7 March 2022 (UTC)

== Excel parameters == 

Currently the page says that the "2" parameter is for the type and is chosen because the samples are not of same length. Is this really the meaning of the parameter? 'type' sounds to me whether it's 'paired' or 'unpaired' data, which would go a bit farther than just having the same amount of observations in both categories: A 'paired' test would be comparing a series of 'before' and 'after' measurements, so each data point has a corresponding 'partner' :) - [[User:Gedankenstuecke|Gedankenstuecke]] ([[User talk:Gedankenstuecke|talk]]) 09:02, 4 March 2022 (UTC)

: I just read up on it and expanded the article with some background on choosing the parameters etc! I hope it didn't turn too technical - [[User:Gedankenstuecke|Gedankenstuecke]] ([[User talk:Gedankenstuecke|talk]]) 13:34, 4 March 2022 (UTC)

== Independence ==

"Depending on what you are measuring this might not be the case, in particular for time-series data where you take very frequent samples (e.g. two heart rate measurements taken a minute apart will be highly linked to each other). "