| — |
me after rewriting my 1500 line analysis software for the 4th time now, i have caught countless errors that affect the overall output of my data, and inevitably my findings. i have realized that your analyzed data is only as strong as the weakest portion of your software. thats why it is absolutely critical to rigorously validate your logic, and i think part of this process should include peer-review. thats right, journals need to hire in-house programmers (or find separate reviewers) to review code to help scientists review claims! this is the only way to increase quality and trust in research. furthermore it will force analysis to be even more open source so that groups can build off each others work in both domains, the findings AND the analysis!! no more voodoo, no more lying with statistics. i can only wonder how many researchers have had to reject perfectly valid hypothesis because of some flaw in their analysis pathway. the reverse case is even worse. this needs to end. |
scientific research need more transparency. the actual code used in analysis should be submitted for peer-review. how can we accept a claim if we don’t know EXACTLY how the data has been analyzed??
”
