Hacker News Reader

hmm, a lot of these stats are worded in a very obtuse way.

Calculating percentage difference of two percentages overstates their differences in regards to the underlying statistic.

For example... 'be 12% more likely to be unemployed' is used to state the difference between an unemployment rate of 5.6% and 5%

Unfortunately that removes a lot of usefulness from the site.

View on HN

Why? That is the far more correct way of presenting data.

Yes and no. 5 is 12% away 5.6, but so is 50 and 56. I would argue that the latter difference is much bigger and more likely to be noticeable by average Joe going about his daily life than the former.

Or compare 1% and 2% to 50% and 100%

It's not though. If you take a random sample of people the odds that one is unemployed is .6% higher in one set than the other, not 12. The 12 only makes sense if you are only sampling from the unemployed, which is not how anyone on the street is going to parse that statement.

The difference between 5% and 5.6% is approx 11%. But I agree it’s more useful to look at percentage points here

Seems like adding a hover/click to see the raw numbers feature would be nice

That feature already exists, also includes the source of the data.

Reply on news.ycombinator.com