There is a super interesting study on citation quality in scientific studies.
In this study, we introduced a computational pipeline to analyze citation fidelity on a large scale and developed an automated measure to evaluate how authors report prior findings. By applying this method to 13 million citation sentence pairs, we uncovered several key insights into the dynamics of scholarly communication and show citation fidelity is higher with closer proximity. Through a quasi-causal experiment, we further establish that the “telephone effect” contributes to information loss: when citing papers exhibit low fidelity to the original claim, future papers that cite both the citing paper and the original show even lower fidelity to the original.
Clearly incorrect or misleading information couldn’t be identified in this study although the definition a cut-off should be possible.
And well, I would be also very interested if there are differences by scientific field.