Do you think you would you have discovered a gorilla hiding in plain sight in your data? We found that you are 5 times less likely to do so if you have a specific hypothesis in mind. *A hypothesis is a liability*
Cool new version of the famous invisible-gorilla study: Students were less likely to find a gorilla hidden in a dataset if they'd been asked to test a specific hypothesis than if they'd just been asked to explore the data. HT
V Interesting study - to hypothesise or not? Reminded me of discussions w and et al about opportunistic exploration vs hypothesis-driven science.
Love this paper ( 🦍 ) and couldn't agree more: "A hypothesis is a liability" by and and
Exploring your data is not a sign of a lack of scientific rigor. That robustness comes later with hypothesis testing.
In their latest installment of their night science/day science series, and discuss how having a hypothesis can lead you to miss patterns in data. Analyzing data with no preconceptions can lead to new hypotheses to explore
Authors argue that if we test specific a priori hypotheses, we may miss obvious insights in the data. They showcase this with a cute experiment. Bears on the testing vs discovery discussion I've been reading up on for my psych inquiry piece.
Researchers are less likely to find a gorilla hiding in the data if they already have a hypothesis in mind. This shows how researchers can sometimes fall prey to confirmation bias or tunnel vision, and how having a hypothesis may be a liability. Source:
Brilliant - on the complementarity (rather than opposition) of hypothesis-driven vs non-hypothesis exploratory research.
Students who were directed to test several specific hypotheses where 5 times less likely to discover a gorilla hidden in the data 😲 "A hypothesis is a liability"
Beautiful paper! Australia is especially conservative about this, but we must support researchers and students doing hypothesis-free work. "Not all those who wander are lost."
A hypothesis is a liability. "Students without a specific hypothesis were almost five times more likely to discover the gorilla when analyzing this dataset" h/t
Hypotheses are important, but also important to recognize there are other ways to approach a problem.
A hypothesis is a liability - interesting editorial making the case for data exploration #hypothesis #philosophy #science #scichat #sciencetwitter HT
"Would the mental focus on a specific hypothesis prevent us from making a discovery?" Unique extension of the famous invisible gorilla experiment by