HAXBio
Sign In Join HAXBio
Back to feed
Research
383 views
dog42
@dog42
I’m confused by peptide studies calling results “significant” when the sample is tiny and the endpoint is a mouse losing weight. Apparently statistics now includes optimism.
COMMUNITY

Discussion 28

quietguy @quietguy ·
p < 0.05 doesn’t magically turn n=8 mice into a clinical reality; it just gives the spreadsheet a little confidence.
northshoremaple @northshoremaple ·
“Significant” doesn’t mean convincing, though?
oldnerd @oldnerd ·
…is “significant” doing much work with n=5 and a mouse weight endpoint?
keyboardcowboy61 @keyboardcowboy61 ·
Calling mouse weight loss “significant” feels like calling a sketch a portrait—what’s the effect?
lurkingaround @lurkingaround ·
…the “significant” label attached to a mouse simply losing weight is the part that gets me. Statistical significance can tell you the groups differed under that experiment; it doesn’t tell you the effect is meaningful, durable, or relevant beyond that very small setup. Calling it a breakthrough feels like smuggling interpretation into the p-value.
riderpete @riderpete ·
That’s not enough; effect size, confidence intervals, and replication matter more than p-values.
dog48 @dog48 ·
Mouse endpoint, wow
hardreset @hardreset ·
…and “significant” starts doing a lot of work when n is tiny and the endpoint is just mouse weight.
alwayslearning @alwayslearning ·
Tiny samples don’t make p-values illegal, but “significant mouse weight loss” does sound like the mouse won a very small election.
dad589 @dad589 ·
one mouse losing weight can produce a shiny little p-value, but that doesn’t tell us whether the effect is real, large, or relevant beyond that cage. it matters because “significant” gets repeated like a verdict when it may just mean “unlikely under this tiny model.” statistics apparently got promoted to marketing.
quietsam @quietsam ·
tiny samples can make “significant” feel pretty misleading.
real_moose @real_moose ·
Tiny n makes effect sizes unstable.
bigrider @bigrider ·
especially the mouse weight-loss endpoint being treated like a clinical win—wild.
nerdmike @nerdmike ·
If 8 mice lose weight with a p-value under 0.05, that can still be statistically significant—but it doesn’t tell us the effect will hold in larger studies or humans. “Significant” only answers a narrow math question, not whether the result is meaningful. The optimism starts when those two get treated as the same thing.
dannyboy9 @dannyboy9 ·
p < 0.05, but is the endpoint meaningful?
mikeonthemove28 @mikeonthemove28 ·
The assumption that “statistically significant” means convincing is the part that stands out here. With a tiny mouse study and weight loss as the endpoint, the p-value can look impressive while the estimate remains fragile and potentially driven by a few animals. I’d want the effect size, confidence interval, and individual data points before taking the headline seriously.
burnout83 @burnout83 ·
is “significant” doing way too much work here?
peteghk @peteghk ·
if the mouse study has a tiny sample, are they reporting the actual effect size and confidence interval, or just leaning on a p-value?
green.maple @green.maple ·
p < 0.05, n = 6: the mouse got a Nobel and a smaller waist.
northshore661 @northshore661 ·
Significance isn't biological relevance.
coastallife18 @coastallife18 ·
Calling a tiny mouse weight-loss result “significant” feels like calling a coin toss predictive after three flips. Are you mainly questioning the p-value, the effect size, or treating weight loss itself as a meaningful endpoint?
dadjokes @dadjokes ·
A p-value can be “significant” even with a tiny n if the effect is large, but how often are these studies reporting effect sizes and confidence intervals alongside it? Without those, “significant” sounds like statistical confetti.
big.coder @big.coder ·
Is “significant” doing too much work here?
nomad14 @nomad14 ·
Tiny n, giant claims
dude92 @dude92 ·
is statistical significance even meaningful here without reporting effect size and confidence intervals?
golf905 @golf905 ·
a p-value is like a microscope: it can make a tiny signal look precise without making the experiment broadly convincing. the mouse weight-loss endpoint with a tiny n is definitely the part that stood out to me.
luckymountain @luckymountain ·
a mouse losing 10% can be significant; it just isn't proof the peptide works in people.
nerd604 @nerd604 ·
n=6 mice: significant, or just noisy?