3 ms·
> 'Boys score almost half a grade level ahead' -- ahh so much imprecise terminology that we can't know what it even means, let alone if this example was represe
by mrep 8y ago
> 'Boys score almost half a grade level ahead' -- ahh so much imprecise terminology that we can't know what it even means, let alone if this example was representative or cherry-picked.
Well the paper is linked in the third sentence if your curios and want to get all the details [0]. Newspapers generally try to keep it simple for the average person. From the paper: "We estimate the mean math and ELA test scores for male and female students for each of roughly ten thousand U.S. school districts in grades three through eight from the 2008-09 to 2014-15 school years. These data enable us to estimate male-female testscores gaps, as well as changes in the gaps over grades and cohorts within districts, providing adescription of gender differences in academic performance at an unprecedented level of detail.
[0]: https://cepa.stanford.edu/sites/default/files/wp18-13-v201806_0.pdf https://cepa.stanford.edu/sites/default/files/wp18-13-v20180...
- mlthoughts2018 8y agoWhy does the paper being linked have a bearing on this? The point is that even in a popular article or cursory summary, you at least need the context of an effect size with interpretable units. If you have to follow a link to the paper even for that then the article itself would have to be worded in an entirely neutral way (as in, not indicating that the data supports any particular conclusion). Any non-neutral presentation (like saying the data supports any type of conclusion about boys’ math performance in rich, white suburbs) will create subjective impressions about what the result means (which can be fine, so long as an effect size and interpretable units are attached). People seem to disagree with my comment because the study (among others) was hyperlinked in the Times article. This strikes me as entirely missing the point of my comment.
- mrep 8y agoBecause this is a newspaper, not a scientific journal or peer review and newspapers are generally written for the average reader who probably does not want nor even understand all of those details.
- mlthoughts2018 8y agoBut a basic notion of effect size and units are often included in popular news coverage of statistics. That's a basic expectation, involving zero nuance or statistical rigor. "This is a newspaper" is a reason why they might leave off details about methodology, metrics of uncertainty, etc. Not a reason why they would make an entirely unqualified claim attached to no notion of the effect size. Essentially you're saying we should hold the New York Times to the same reporting standards as some clickbait site that fuels confirmation bias with sensational headlines, and not expect large world-spanning newspapers to have even the slightest of better practices than that.