about · email me · subscribe

spurious scholar

Because if p < 0.05, why not publish?

Step 1: Gather a bunch of data. There are 25,156 variables in my database. The data ranges from the mundane (air pollution in Chicago) to the weird (Hotdogs consumed by Nathan's Hot Dog Eating Competition Champion) to the super-niche (How clickbait-y Numberphile YouTube video titles are, as rated by an AI).
Step 2: Dredge that data to find random correlations between variables. "Dredging data" means taking one variable and correlating it against every other variable just to see what sticks. It's a dangerous way to go about analysis, because any sufficiently large dataset will yield strong correlations completely at random.

Fun fact: the chart used on the wikipedia page to demonstrate data dredging is also from me. I've been being naughty with data since 2014.

Step 3: Calculate the correlation coefficient, confidence interval, and p-value to see if the connection is statistically significant. "Statistically significant" is a misleading term. It sounds like it means "statistically significant" because, you know, those are the same two words. Unfortunately statistical significance is a technical term that means mumble mumble at least as extreme mumble mumble null hypothesis mumble mumble probability mumble mumble p-values.

You know what? Forget the technical definition. "Statistically significant" just means "someone did some fancy math."

I really did the fancy math below and you can check it by clicking on the "view detailed data" link under each paper. And these really do qualify as "statistically significant" in the technical sense. It's just that "statistically significant" does not mean the results are "significant."

Step 4: If it is, have a large language model draft a research paper.
Step 5: Remind everyone that these papers are AI-generated and are not real. Seriously, just pick one and read the lit review section. The silliness of the papers is an artifact of me (1) having fun and (2) acknowledging that realistic-looking AI-generated noise is a real concern for academic research (peer reviews in particular).

The papers could sound more realistic than they do, but I intentionally prompted the model to write papers that look real but sound silly.

Also: every page says "This paper is AI-generated" at the bottom and the first letters of the names of the authors always spell out C-H-A-T-G-P-T.

Step 6: ...publish:

Sun-Kissed Suds: The Brew-tiful Relationship Between the Number of Breweries in the United States and Solar Power Generated in the Cook Islands
Journal of Solar-Powered Suds
r=0.942 · 95% conf. int. [0.880,0.972] · r2=0.887 · p < 0.01
Generated Jan 2024 · View data details

Kernels of Truth: The GMO-Cant Connection in Corn Crop Culture
The Journal of Agricultural Genetics and Cultural Studies
r=0.914 · 95% conf. int. [0.791,0.966] · r2=0.835 · p < 0.01
Generated Jan 2024 · View data details

From Juris Doctor to Fjords: A Correlation Between Legal Education and Aspiring Norwegians
The Journal of Legal and Nordic Studies
r=0.919 · 95% conf. int. [0.686,0.981] · r2=0.844 · p < 0.01
Generated Jan 2024 · View data details

From Shakespeare's Sonnets to Discounted Trinkets: The Curious Link between Associates Degrees in English and Google Searches for 'Dollar Store Near Me'
The Journal of Linguistic Curiosities
r=0.988 · 95% conf. int. [0.952,0.997] · r2=0.976 · p < 0.01
Generated Jan 2024 · View data details

The Page Count Connection: An Examination of the Correlation Between First-Graders and Bookworms in the UK
Journal of Childhood Literacy and Development
r=0.937 · 95% conf. int. [0.784,0.982] · r2=0.877 · p < 0.01
Generated Jan 2024 · View data details

Frankfurter Fuel: Fossil Fumes' Festive Fluctuations and Franks' Famished Fanfare
Journal of Gastronomic Geology
r=0.830 · 95% conf. int. [0.704,0.906] · r2=0.690 · p < 0.01
Generated Jan 2024 · View data details

Blown Away: Uncovering the Zephyrous Link Between Mauritanian Wind Power and Idaho's Accounting Population
The Journal of Transcontinental Aerodynamic Connections
r=0.989 · 95% conf. int. [0.953,0.998] · r2=0.978 · p < 0.01
Generated Jan 2024 · View data details

Lucas at the Top: A Hydrodynamic Popularity Flop
The Journal of Aquatic Social Dynamics
r=0.966 · 95% conf. int. [0.938,0.982] · r2=0.934 · p < 0.01
Generated Jan 2024 · View data details

The Power of Business Education: A Current Flow Analysis of Associates Degrees in Business and Management on Hydroelectric Energy Production in Algeria
Journal of Renewable Energy Economics and Management
r=0.946 · 95% conf. int. [0.800,0.986] · r2=0.895 · p < 0.01
Generated Jan 2024 · View data details

Bo Knows Stocks: A Cautionary Tale of Zimmer Biomet Holdings' Stock Price and the Popularity of the First Name Bo
Journal of Quirky Economics
r=0.918 · 95% conf. int. [0.806,0.967] · r2=0.843 · p < 0.01
Generated Jan 2024 · View data details

Rice-ing to the Top: Exploring the Link Between Global Rice Consumption and Capital One Financial's Stock Price
Journal of Culinary Economics
r=0.954 · 95% conf. int. [0.858,0.986] · r2=0.910 · p < 0.01
Generated Jan 2024 · View data details

Dairy Dollars and Bankin' Bucks: Exploring the Financial Moovement between Annual US Household Spending on Dairy Products and ICICI Bank's Stock Price
Journal of Financial Bovinomics
r=0.896 · 95% conf. int. [0.758,0.958] · r2=0.803 · p < 0.01
Generated Jan 2024 · View data details

Brew-nomics and Stock Spirits: A Sudsy Look at the Relationship Between U.S. Brewery Count and Humana's Stock Price
The Journal of Fermented Finance
r=0.971 · 95% conf. int. [0.929,0.988] · r2=0.943 · p < 0.01
Generated Jan 2024 · View data details

Stocked Up on Greens: Unearthing the Correlation Between Annual US Household Spending on Fresh Vegetables and Cummins' Stock Price
Journal of Vegetables & Finance
r=0.948 · 95% conf. int. [0.873,0.979] · r2=0.898 · p < 0.01
Generated Jan 2024 · View data details

The Niko-nomical Impact: Analyzing the Connection between Niko's Popularity and Mastercard's Stock Price
Journal of Financial Psychology
r=0.988 · 95% conf. int. [0.963,0.996] · r2=0.975 · p < 0.01
Generated Jan 2024 · View data details

Jamesha's Fame and MFG's Claim: A Rhyming Game
The Journal of Quirky Linguistics
r=0.900 · 95% conf. int. [0.652,0.974] · r2=0.810 · p < 0.01
Generated Jan 2024 · View data details

Pawsitive Impact: Exploring the Cat Memes Craze on Delta Airlines' Stock Price
The Journal of Feline Finance
r=0.888 · 95% conf. int. [0.700,0.961] · r2=0.788 · p < 0.01
Generated Jan 2024 · View data details

Cheers to Finance: Analyzing the Ale-ment of Breweries on Ameriprise Financial's Stock Price
The Journal of Financial Foam and Fermentation
r=0.870 · 95% conf. int. [0.670,0.953] · r2=0.757 · p < 0.01
Generated Jan 2024 · View data details

Powering Up: The Electrifying Connection Between Theodore's Popularity and DTE Stock Price
The Journal of Meme Economics
r=0.988 · 95% conf. int. [0.971,0.995] · r2=0.977 · p < 0.01
Generated Jan 2024 · View data details

The Final Frontier on Wall Street: To the Moon or to the Burrito? A Case Study of Google Searches for 'How to Go to Space' and Chipotle Mexican Grill's Stock Price
The Journal of Interstellar Economics and Gastronomy
r=0.961 · 95% conf. int. [0.892,0.986] · r2=0.923 · p < 0.01
Generated Jan 2024 · View data details

Support Staff: A Secret Ingredient in the Stock Market Brew?
The Journal of Financial Quirks and Curiosities
r=0.980 · 95% conf. int. [0.933,0.994] · r2=0.960 · p < 0.01
Generated Jan 2024 · View data details

Planetary Alignment: Out of This World Predictors of Stock Prices
The Interstellar Journal of Financial Astrology
r=0.653 · 95% conf. int. [0.320,0.843] · r2=0.427 · p < 0.01
Generated Jan 2024 · View data details

Military Marvels and Market Movements: The Link between Bachelor's degrees in Military Technologies and Applied Sciences and Global Payments' stock price
The Journal of Military Techno-Economics
r=0.993 · 95% conf. int. [0.971,0.999] · r2=0.987 · p < 0.01
Generated Jan 2024 · View data details

Popularity of the Name Skylar Drives Stock Price: A Cardinal Connection Analysis
Journal of Quirky Social Influences
r=0.859 · 95% conf. int. [0.679,0.941] · r2=0.737 · p < 0.01
Generated Jan 2024 · View data details

Stevie's Significance on Stock: A Statistical Study of Analog Devices' ADI
Journal of Quantitative Analysis in Technology and Finance
r=0.977 · 95% conf. int. [0.943,0.991] · r2=0.954 · p < 0.01
Generated Jan 2024 · View data details


Currently viewing 25 of 4,731 spurious research papers

Page
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190



Why this works

  1. Data dredging: I have 25,156 variables in my database. I compare all these variables against each other to find ones that randomly match up. That's 632,824,336 correlation calculations! This is called “data dredging.” Fun fact: the chart used on the wikipedia page to demonstrate data dredging is also from me. I've been being naughty with data since 2014.
    Instead of starting with a hypothesis and testing it, I isntead tossed a bunch of data in a blender to see what correlations would shake out. It’s a dangerous way to go about analysis, because any sufficiently large dataset will yield strong correlations completely at random.
  2. Lack of causal connection: There is probably no direct connection between these variables, despite what the AI says above. Because these pages are automatically generated, it's possible that the two variables you are viewing are in fact causually related. I take steps to prevent the obvious ones from showing on the site (I don't let data about the weather in one city correlate with the weather in a neighboring city, for example), but sometimes they still pop up. If they are related, cool! You found a loophole.
    This is exacerbated by the fact that I used "Years" as the base variable. Lots of things happen in a year that are not related to each other! Most studies would use something like "one person" in stead of "one year" to be the "thing" studied.
  3. Observations not independent: For many variables, sequential years are not independent of each other. You will often see trend-lines form. If a population of people is continuously doing something every day, there is no reason to think they would suddenly change how they are doing that thing on January 1. A naive p-value calculation does not take this into account. You will calculate a lower chance of "randomly" achieving the result than represents reality.

    To be more specific: p-value tests are probability values, where you are calculating the probability of achieving a result at least as extreme as you found completely by chance. When calculating a p-value, you need to assert how many "degrees of freedom" your variable has. I count each year (minus one) as a "degree of freedom," but this is misleading for continuous variables.

    This kind of thing can creep up on you pretty easily when using p-values, which is why it's best to take it as "one of many" inputs that help you assess the results of your analysis.
  4. Outliers: Some datasets here have outliers which drag up the correlation. In concept, "outlier" just means "way different than the rest of your dataset." When calculating a correlation like this, they are particularly impactful because a single outlier can substantially increase your correlation.

    Because this page is automatically generated, I don't know whether any of the charts displayed on it have outliers. I'm just a footnote. ¯\_(ツ)_/¯
    I intentionally mishandeled outliers, which makes the correlation look extra strong.



Spurious Scholar was launched January 27, 2024. If you have feedback on it, I'd love to hear from you! Shoot me a note: feedback@tylervigen.com.


Project by Tyler Vigen
emailme@tylervigen.com · about · subscribe


CC BY 4.0