Survivorship Bias Plane Meets the Missing Z-Values in Medical Research

Description

A multi-layered statistics meme combining the 'What if we kissed' meme format with a real statistical phenomenon. The bold top text reads 'WHAT IF WE KISSED ON THE SURVIVORSHIP BIAS PLANE FLYING BETWEEN THE MISSING Z-VALUES OF MEDICAL RESEARCH PAPERS' flanked by flushed-face emojis. Below is a histogram showing the 'Distribution of z-values' from medical research papers, with z-values on the x-axis (-6 to 6) and count on the y-axis (0 to ~9000). The distribution shows a conspicuous dip/gap near z=0 and suspicious spikes just beyond the significance thresholds (dashed lines at approximately z=-2 and z=2), indicating p-hacking. Overlaid on the histogram is the famous WWII survivorship bias bomber diagram (Abraham Wald's plane), with bullet holes scattered across it. The meme brilliantly layers three concepts: the romantic 'what if we kissed' format, p-hacking in medical research (the suspicious z-value distribution), and survivorship bias (the plane)

Common Tags

DataScience StatisticalAnalysis DataVisualization AcademicLife CognitiveBias NerdHumor DataAnalysis HistoricalReferences MemeCulture

Comments

Anonymous ★ Top Pick The only thing more suspicious than that z-value gap is a data scientist who claims they've never run the analysis 'just one more time' to get p < 0.05

Anonymous ★ Top Pick

The only thing more suspicious than that z-value gap is a data scientist who claims they've never run the analysis 'just one more time' to get p < 0.05
Anonymous

The null hypothesis is that we're just friends, but I'm rejecting it based on this statistically significant gap in my social life that you seem to perfectly fit
@RiedleroD 8mo

I wonder where that graph is from and what the values are for
1. @Daonifur 8mo
  
  Looks like the count of medical research papers
@Daonifur 8mo

No idea from where or what timeframe though, everything is very vague. Could be for a math problem or something
@Subeltz 8mo

https://medianwatch.netlify.app/post/z_values/
1. @Daonifur 8mo
  
  Can confirm this is the source
2. @RiedleroD 8mo
  
  mmm, ouch
@Algoinde 8mo

Shit results prevent others from trying the same and also getting shit results Also low z values in a research might expose an already existing drug being in fact worthless, tanking stonks
@Algoinde 8mo

maybe, but if you only do what's interesting, you're selfish and don't have the overall field advancement in mind
@Algoinde 8mo

that's precisely the climate that needs to be eradicated at multiple levels maybe publishing itself needs to be structured differently for this to happen
@Johnny_bit 8mo

Let's add p-hacking to that :) now we can turn null-result into something worht publishing/getting grant for :)
@theodolu 8mo

Which is why we don't know which stuff doesn't work and do the same studies over and over again