Twelve points important for not being deceived by science articles
We are publishing infography on chemistryCompound InterestHowever, in order not to be fooled through the "Tondemo" science article flooded on the Internet, 12 points to be seen are summarized in infographics.
A Rough Guide to Spotting Bad Science | Compound Interest
◆01: Overdoing title
The title of the article is an important element that decides whether or not to be clicked whether it is noticed or not. A good title is a simple thing. Too much title and misleading title are attention.
When quoting scientific papers, mistranslation is inherent, sometimes even to decorate the story to make it interesting. It is important that the source of information should be as original as possible.
◆03: Reliability of information
Many companies hire researchers themselves to write scientific articles. In other words, you should keep in mind that 'the researchers who are hired also have life.'
◆04: Impossible causal relationship
Let's pay attention to correlations and causal relationships. Are not you short-termly determining the factors that will bring results? For example, pirates have been decreasing since 1800 and global warming has occurred since 1800, but it is clear that the cause of the decrease in piracy is not global warming.
◆05: speculative expression
Guesses such as "maybe", "may have a possibility", "seems like" "is not preferable for scientific articles. Evidence and proof are required by science. It is only the objective fact that the conclusion can be drawn.
◆06: Too few samples
Attention should be paid to experimental results where the number of samples is too small. In some experiments it may be difficult to secure the number of samples, but keep in mind that the number of samples is proportional to the reliability of the study. If there are not too many samples even though there is no trouble in experimenting, you should turn doubtful attention.
◆07: Special conditions
Special factors should be excluded as much as possible as experimental conditions. The fact that special conditions are included is because it may produce universal results. In the case of too special conditions you should doubt the universality of the result.
◆08: No control experiment
In clinical trialsPlacebo effectTo eliminateControl experimentWhat to do. If there is no control experiment, it is cautionful.
◆09: No double blind test
In consideration of not only avoiding the placebo drug effect due to the subject's belief but also possibly influencing the subject from the behavior of the practitioner, do so without notifying the practitioner of the truth of the examinationDouble blind testIs valid. Attention should also be paid to experiments that have not done double blind test.
◆10: Adopt data conveniently
Articles written in favor of favorable data to guide "the desired conclusion" are insufficient for trust.
◆11: No reproducibility
Experiment is not completed at the stage of publishing the results in the paper, but it is recognized for the first time by having other researchers verify it. In the verification, we test the generality and universality of the experiment result by expanding the condition more widely. In other words, the experimental result is recognized as a result only by reproducibility, and the result which can not be reproduced is meaningless. It should be noted if the experiment is reproducible.
◆12: Do not blind the authority
NatureYaSciencePapers published in prestigious academic journals such as those that are inevitably authoritative. However, prestigious papers are not absolutely perfect. We should avoid blindly accepting research results as correct. Also, a number of quoted papers are equally authoritative, but the same is true that no confidence is prohibited.