@hakwanlau More on the small side https://t.co/7ew8ePBUsj
@hakwanlau This one is a classic and takes motivation from Fechner/Ebbinghaus: https://t.co/JYraAX4YL8 https://t.co/dTiWq1gtxY
@JamesSteeleII @SolomonKurz @Physical_Prep Like a small-N design? https://t.co/YDuzanOFww
@kohinoordarda @pgmm03 @FrankPollick @RuudHortensius @HandandBrain_BU @helenahhartmann @AnnaHenschel @brain_on_dance @RichRamseyPhD Quite a few papers out there now on this topic, but this one comes to mind: https://t.co/inZl8Ls0lr .
@Leesplez https://t.co/MFBSj8JVnM This paper deals with precisely that issue. If you are looking for more examples, have a look through papers that have cited it (self-interest disclosure: I have one in there!)
@JeanLaurensLab I'm a big fan of single-subjects / small-N designs (see for example https://t.co/eTBg4i7fEb, for recent related ideas in psychology). What your paper seems to sidestep is the issue of uncertainty in classifying animals as "outliers".
day 9 of 1 paper/day more on “better ways to do science”: advantages of small-N studies with many trials, as often used in (visual-)perceptual research. I guess the call for better theory (which will lead to better within-subject predictions) is warranted
@BinneyRJ For example: https://t.co/4IOHIIZbcF
@AlexKale17 @elibryan I mostly agree with @AlexKale17 except for power analyses, which are pretty difficult to compute/interpret for within-subjects designs with many repeated trials. Methodological vs statistical tradeoffs is where a lot of debate happens
Great thread. Similar points about individuals as replication units made here: https://t.co/X0JKjVq5Sy
RT @tsawallis: @NeuroPolarbear Similar discussion going on over here in psychology; a bunch of us are pushing back in defence of small N. e…
@NeuroPolarbear Similar discussion going on over here in psychology; a bunch of us are pushing back in defence of small N. e.g. https://t.co/eTBg4i7fEb) Smith and @danielrlittle; there was a "small N but high power" workshop at @improvingpsych this year, o
@zerdeve @dan_p_simpson And Smith and Little make a great nuanced argument about how between person heterogeneity can help to justify small n studies (https://t.co/eMHeNwe4Eq)
RT @bg_farrar: @js_simons This is perhaps a useful paper - I think the argument comes down to what question you're trying to answer and som…
RT @bg_farrar: @js_simons This is perhaps a useful paper - I think the argument comes down to what question you're trying to answer and som…
❤️ this evergreen citation
for those who need to see this 🧠
@esdalmaijer Yes, that's a good point. The distinction between individual-level effects and population-wide estimates is made nicely in the Smith & Little paper https://t.co/dIP3NxbQv0
RT @js_simons: Many thanks, everyone. I think the general consensus seems to align with my view that small-N can be fine with enough measur…
@brkicdiandra
Many thanks, everyone. I think the general consensus seems to align with my view that small-N can be fine with enough measurements in each subject (see eg https://t.co/dIP3NxbQv0), but it needs greater justification than merely "this is what the field does
@js_simons This is perhaps a useful paper - I think the argument comes down to what question you're trying to answer and some a priori reasoning about the probably generalizability of results - with greater mechanistic knowledge the need for large samples
@manes @js_simons @hugospiers @willjharrison @BaysLab https://t.co/wb2haIz5Kv And a lot who cite this...
@hugospiers @js_simons Not a vision scientist myself but the debate is familiar. I guess it depends on what is the unit of observation, this paper makes that point very clear https://t.co/jEOmFffU7s with examples from psychophysics where a few individuals
@HirstRj Yes, this exactly the tradition I am referring to (you may take a look at this article, rightfully praising this approach, if you are interested: https://t.co/O0Le3ZjEX5).
RT @n_ramnani: Learned about this thought provoking paper from @rikhens earlier today at #FENS2020 @BritishNeuro Credibility in Neuroscienc…
RT @n_ramnani: Learned about this thought provoking paper from @rikhens earlier today at #FENS2020 @BritishNeuro Credibility in Neuroscienc…
Thanks for sharing @n_ramnani !! I wanted to get hold of it after this mornings excellent #credibility session!!
Learned about this thought provoking paper from @rikhens earlier today at #FENS2020 @BritishNeuro Credibility in Neuroscience session. Reproducibility: Small-N designs can be useful if used in the right way. https://t.co/DK3ifkbvlc
RT @nschawor: I like working with large datasets. But I find it preferable to really understand what is happening for an individual subject…
@vineettiruvadi for me it is much more satisfying to see robust effects within one single participant. but the search for universal laws is pervasive 🙂 https://t.co/wg99fcj6Uf
... pointing to this article https://t.co/O09Kc6NaM8
@BRSLWP @valentinwyart @NatureComms I really like this paper about small n designs :) https://t.co/EQLCCUr2bK
@SarahWieten I don't think that was me, but I did recently find out about this paper on analytical methods for small-N designs in psych. Haven't had a chance to read yet, but I'm intrigued. https://t.co/l39nRkleCt
@bjm262run @RobinMazumder There are no rules about N that apply to all research. Statistical tests were often designed explicitly for small N. The 'degrees of freedom' takes N into account. We can learn an awful lot from single case studies, too! https
RT @nschawor: I like working with large datasets. But I find it preferable to really understand what is happening for an individual subject…
Cute! Small from Little
RT @nschawor: I like working with large datasets. But I find it preferable to really understand what is happening for an individual subject…
I like working with large datasets. But I find it preferable to really understand what is happening for an individual subject, to arrive at a robust prediction for this single person. Interesting paper in this realm: https://t.co/rC5DXdSJI8 https://t.co/Vz
Quote from "Small is beautiful." The only problem is that this is false. The quote is not in the cited article. https://t.co/ClHr5eNnk2 https://t.co/2aPgcP5OQE
😉
@DrDylanThompson Small n's are fine for certain questions https://t.co/ZsqB7XjPK7
@sophiescott @PsychScientists @mrccbu Ah, OK! It's quite common in some fields - monkey neurophysiology, psychophysics. Often fewer (trained, motivated) Ps is better! I'm running several N<6 studies now, up to ~30 hours testing per P. This paper is my n
RT @Naomi_D_Harvey: Small-N designs, with repeated measures, can "concentrate their experimental power at the individual participant level…
RT @Naomi_D_Harvey: Small-N designs, with repeated measures, can "concentrate their experimental power at the individual participant level…
Small-N designs, with repeated measures, can "concentrate their experimental power at the individual participant level and provide high-powered tests of effects at that level. As a result, they are in a sense automatically “self-replicating"" https://t.co/
@Naomi_D_Harvey @LjerkaOstojic Yep, I completely agree - there's a good outline of an approach like this here https://t.co/cOetdPBAee One issue for individual-level analyses in animal cognition is to find out where they sit on the case study vs n-of-1 vs s
@westwoodsam1 @thinkhumm @vinwalsh Thanks, I'm not a 'TES' person, those *are* small... I appreciate the move for 'larger N', especially where 'products' will be sold, but larger N does not guarantee better data or even better power. Better experiments do.
RT @eglerean: @CyrilRPernet @russpoldrack @club_scan Ok but here N can be seen as number of studies and T (the times an individual is scann…
@CyrilRPernet @russpoldrack @club_scan Ok but here N can be seen as number of studies and T (the times an individual is scanned) as the number of "subjects". 1 subject measured 100 times can be more informative than 100 subjects measured once https://t.co/
RT @mspitschan: This is worth a read: https://t.co/p7Lvn6cD0P https://t.co/01bcKyKtSg
This is worth a read: https://t.co/p7Lvn6cD0P
@SandersanOnie Agreed! Although if the theory and experimental control etc are strong, there can be instances where small N can have high power and inferential validity: https://t.co/YiWlPhFgVu
RT @ykamit: “It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/m…
RT @rw_hass: Love this whole section from https://t.co/PfVlb47wUS It explains why cognitive scientists don't always see eye to eye with oth…
RT @ykamit: “It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/m…
RT @ykamit: “It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/m…
RT @ykamit: “It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/m…
@rogierK @beckyneuro higher power does not necessarily come from higher N: effect-size & power can be larger with smaller, more motivated, better-selected, more representative, better-trained samples. Smaller can be beautifuler https://t.co/TOsZfgh2SJ
RT @ykamit: “It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/m…
RT @ykamit: “It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/m…
RT @ykamit: “It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/m…
I retweet whenever I see this paper.
RT @ykamit: “It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/m…
RT @ykamit: “It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/m…
RT @ykamit: “It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/m…
RT @ykamit: “It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/m…
RT @ykamit: “It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/m…
RT @ykamit: “It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/m…
“It is more useful to study one animal for 1000 hours than to study 1000 animals for one hour” — B. F. Skinner https://t.co/mO64rwKsXW
wrote to my collaborator: Four subjects can be enough for psychophysics (see https://t.co/adH1HkbrWN) although not if you want to use statistics to generalize to the larger population (https://t.co/yb9h0daFuz). Does that sound about right, @danielrlittle @
Love this whole section from https://t.co/PfVlb47wUS It explains why cognitive scientists don't always see eye to eye with other psychologists on certain "new" practices that will remain nameless https://t.co/HgvneNJbFT
RT @tom_hartley: And I have been re-reading this article Smith & Little article: https://t.co/yWdoeATPvJ Recommended. Framed as an article…
RT @tom_hartley: And I have been re-reading this article Smith & Little article: https://t.co/yWdoeATPvJ Recommended. Framed as an article…
RT @tom_hartley: And I have been re-reading this article Smith & Little article: https://t.co/yWdoeATPvJ Recommended. Framed as an article…
RT @tom_hartley: And I have been re-reading this article Smith & Little article: https://t.co/yWdoeATPvJ Recommended. Framed as an article…
And I have been re-reading this article Smith & Little article: https://t.co/yWdoeATPvJ Recommended. Framed as an article in favour of small-N designs, but going much further, and beautifully clear/crisp writing, IMO.
RT @tom_hartley: Small is beautiful. In defence of the small-N design. https://t.co/yWdoeATPvJ Smith & Little.
”We argue that, if psychology is to be a mature quantitative science, its primary theoretical aim should be to investigate systematic, functional relationships as they are manifested at the individual participant level.”
RT @tom_hartley: Small is beautiful. In defence of the small-N design. https://t.co/yWdoeATPvJ Smith & Little.
RT @tom_hartley: Small is beautiful. In defence of the small-N design. https://t.co/yWdoeATPvJ Smith & Little.
RT @tom_hartley: Small is beautiful. In defence of the small-N design. https://t.co/yWdoeATPvJ Smith & Little.
RT @tom_hartley: Small is beautiful. In defence of the small-N design. https://t.co/yWdoeATPvJ Smith & Little.
RT @kaz_yos: “We argue that this recommendation misses a critical point, which is that increasing sample size will not remedy psychology’s…
RT @tom_hartley: Small is beautiful. In defence of the small-N design. https://t.co/yWdoeATPvJ Smith & Little.
RT @tom_hartley: Small is beautiful. In defence of the small-N design. https://t.co/yWdoeATPvJ Smith & Little.
Thank you twitter lovelies for the fantastic paper 'in defense of small N design' https://t.co/rKNL1JqIEJ - it's a true masterpiece and was super useful. However not quite the "let me just not bother with small effects' msg I was after. Any ref you can sug
RT @tom_hartley: Small is beautiful. In defence of the small-N design. https://t.co/yWdoeATPvJ Smith & Little.
RT @tom_hartley: Small is beautiful. In defence of the small-N design. https://t.co/yWdoeATPvJ Smith & Little.
RT @tom_hartley: Small is beautiful. In defence of the small-N design. https://t.co/yWdoeATPvJ Smith & Little.
RT @tom_hartley: Small is beautiful. In defence of the small-N design. https://t.co/yWdoeATPvJ Smith & Little.
RT @tom_hartley: Small is beautiful. In defence of the small-N design. https://t.co/yWdoeATPvJ Smith & Little.
@apdemetriou small is beautiful! https://t.co/TOsZfgh2SJ @SanabriaLucenaD
My first lab experience was in a visual psychophysics lab where I was a paid observer to test visual threshold detection. The experimenter and a fellow grad student were the other observers. Perfectly replicated frog retina recordings. Had a profound effec
RT @tom_hartley: Small is beautiful. In defence of the small-N design. https://t.co/yWdoeATPvJ Smith & Little.