Hacker News Reader

Comment by droithomme | Hacker News Reader

Wow, the article's presentation of statistics is abysmal and certainly intentionally misleading. The chart produced, by including each bar as all persons that have greater than the level of debt, rather than into bins, means everyone in bars to the right is also included in the count of bars to the left. Sure, she explains this, but it is completely different than normal convention, isn't going to be expected or realized but any but careful readers, and the result is to support her assertion that debt levels are not that high by massively misrepresenting the left side of the chart.

Then there is the issue that she is not even including anyone who graduated in less than 6 years! It's hard to imagine that if we saw those numbers they wouldn't skew the curve as well. The whole presentation reeks of cherry picked data classification.

View on HN

> by including each bar as all persons that have greater than the level of debt, rather than into bins, means everyone in bars to the right is also included in the count of bars to the left. Sure, she explains this, but it is completely different than normal convention

It's called "cumulative", and it is quite common, and not misleading. For this kind of presentation it is what you want because most people are interested in picking some particular debt level and saying "X% are above this debt level". If you don't do a cumulative chart, then you have to take all the bins above X% and add them up to get the number you want.

> Then there is the issue that she is not even including anyone who graduated in less than 6 years! It's hard to imagine that if we saw those numbers they wouldn't skew the curve as well. The whole presentation reeks of cherry picked data classification

Yes--it skews the numbers toward more debt and away from the point she is making it. This strengthens her point, not weakens it.

"Yes--it skews the numbers toward more debt and away from the point she is making it. This strengthens her point, not weakens it."

No it doesn't. I believe GP was pointing out how it further contrasts the lower debt loads against the higher debt loads, which I agree with. This layout definitely favors her point in a misleading way.

I'd appreciate citations to use of this form of graph in reliable related research that isn't clear propaganda. The graph clearly shows at a glance that the vast majority of debtors have a very small level of debt. The large font title of the chart "Percentage of Students with Different Levels of Debt" is misleading. The term cumulative appears in much smaller print at the bottom. Only careful thought reveals to most readers that the left most category is not that level of debt 0-$10,000 that its measure of 64% of students have, but it is that level of debt and all levels above it, (including $100,000, $1 million, etc) summed. The actual percentage of degree holders with 0-$10,000 debt, 64%-51% = 13%, is not shown directly at all. 13% is an extremely different number from 64% and can only be extrapolated from the chart by readers familiar with set theory.

I think you may have misread that part, she includes people who graduated in less than 6 years.

> The chart below shows the percentage of beginning undergraduate students who, six years later, had accumulated more than the indicated levels of debt.

That's actually a pretty standard way to show that kind of data. She both describes it in text and properly labeled the graph itself.

First thing I noticed. Thought about it for a second, then also recognized that the charts were made in a 5 year old excel version and promptly left the site.

Reply on news.ycombinator.com