Single character coding error changes conclusion and leads to retraction of medical paper https://www.medscape.com/viewarticle/wildfire-smoke-dementia-risk-paper-retracted-2025a1000hpw?form=fpf
-
Single character coding error changes conclusion and leads to retraction of medical paper https://www.medscape.com/viewarticle/wildfire-smoke-dementia-risk-paper-retracted-2025a1000hpw?form=fpf
No shame on the researchers for missing this; mistakes are normal and credit to them for correcting the record. What I'm curious about is specifically what language or system the coding error was in (R, Excel, SQL?) to understand how easy it was to make the mistake and if there are other characteristics of the system that make it easier for mistakes to slip through. We need better deterministic tools tuned for scientific and engineering use and need help defending against these sorts of common simple errors.
-
Single character coding error changes conclusion and leads to retraction of medical paper https://www.medscape.com/viewarticle/wildfire-smoke-dementia-risk-paper-retracted-2025a1000hpw?form=fpf
No shame on the researchers for missing this; mistakes are normal and credit to them for correcting the record. What I'm curious about is specifically what language or system the coding error was in (R, Excel, SQL?) to understand how easy it was to make the mistake and if there are other characteristics of the system that make it easier for mistakes to slip through. We need better deterministic tools tuned for scientific and engineering use and need help defending against these sorts of common simple errors.
@arclight "Where there should have been an ‘and’ symbol in our code, there was an ‘and/or’ symbol,”
-
@arclight "Where there should have been an ‘and’ symbol in our code, there was an ‘and/or’ symbol,”
@autolycos @arclight That definitely sounds like they're describing an inclusive or. There aren't many languages that use only a single character for the *logical* boolean operations -- off the top of my head, MATLAB and R -- and no sensible person would use bitwise boolean operations (it's just not the way it's taught). If we consider compound assignments, then `&=` and `|=` differ by only one character, and now you're looking at pretty much any language that inherited syntax from C.
-
@autolycos @arclight That definitely sounds like they're describing an inclusive or. There aren't many languages that use only a single character for the *logical* boolean operations -- off the top of my head, MATLAB and R -- and no sensible person would use bitwise boolean operations (it's just not the way it's taught). If we consider compound assignments, then `&=` and `|=` differ by only one character, and now you're looking at pretty much any language that inherited syntax from C.
@autolycos @arclight it'd be pretty hard to make a typographical error like this in FORTRAN, Ada, Python, and other languages where the boolean operators are `and`, `or`, etc (or `.AND.`, `.OR.`, etc). You could still make a design error, but you wouldn't be able to blame it on a typo.
-
@autolycos @arclight it'd be pretty hard to make a typographical error like this in FORTRAN, Ada, Python, and other languages where the boolean operators are `and`, `or`, etc (or `.AND.`, `.OR.`, etc). You could still make a design error, but you wouldn't be able to blame it on a typo.
-
-
@autolycos @arclight That definitely sounds like they're describing an inclusive or. There aren't many languages that use only a single character for the *logical* boolean operations -- off the top of my head, MATLAB and R -- and no sensible person would use bitwise boolean operations (it's just not the way it's taught). If we consider compound assignments, then `&=` and `|=` differ by only one character, and now you're looking at pretty much any language that inherited syntax from C.
@DocBohn @autolycos @arclight According to the original paper the analysis was carried out in R:
https://jamanetwork.com/journals/jamaneurology/fullarticle/2827124
-
R relay@relay.mycrowd.ca shared this topic
