Hacker News Reader

Comment by pge | Hacker News Reader

The plot is showing that there is a negative selection bias in the reporting of stories about minimum wage. That is, that stories suggesting that increasing minimum wage has negative effects (increases unemployment, etc) are overreported relative to stories showing a positive effect from an increase in minimum wage.

I am only repeating what the paper says; I don't claim to have validated their methodology or sources.

View on HN

That is an incomplete summary. Their conclusion is that despite this larger than normal selection bias, the minimum wage appears to have no impact on unemployment. The paper's abstract, discussion, and summary all mention this as part of the central thesis - why leave it out?

It's not an HN commenter's obligation to post everything a reader might want to see, certainly not highly technical point about the distinction between "this paper claims that evidence for an employment effect is over-reported" and "this paper claims that employment effect is approximately 0". They linked to the whole paper for the curious, and answered a particular question about it.

Also, the use of "despite" does not logically fit the contents of (my) grandparent poster and parent poster, as grandparent poster didn't claim mention employment as a possible positive effect of minimum wage, and even within context of my parent post, "beyond" is a better word choice that n "despite"

I think summaries (by their nature) must leave something out. Considering the abstract mentions what you said (and that it's reasonable for me to have read the abstract), I think it was reasonable for the responder to focus their attention on the plot alone (which was the focus of my question).

It is an incomplete summary of the paper, however it seems to be a pretty solid summary of what the plot is telling us.

The obvious starting position is that raising minimum wages has a negative effect on the number of low wage employees hired; because the money from higher wages has to come from somewhere and might not be available - leading to reduced business activity. This is also the classic economics position. With that in mind, there is going to be a bias towards publishing the 'surprising' results that there is no impact/a positive impact from minimum wages. The bias that this paper is identifying is surely a real thing, but it might potentially be a bais against, eg, poorly run studies that pick up spurious effects.

So the conclusions the paper draws w.r.t. elasticity aren't that interesting; we'd need to know why the bias exists before it is particularly useful. I'm a bit suspicious of using 'meta-analysis' at all because that just means that the faction that repeats themselves the most with low-quality research gets to pick what the elasticity is.

In theory we shouldn't need 200 studies to make a claim, we should need 1 or maybe a small suite (like, 5) of very high quality studies that intellectually honest economists struggle to rebut.

>why leave it out?

Because he summarized the plot, not the paper.

Much appreciated.

And thank you to killjoywashere for selecting out the part of the paper that answers my question; gowld for reformatting appropriate to my mobile device.

Apologies for not replying directly. I have a commenting rate-limiter applied to my account.

I, too, have a commenting rate-limiter applied to my account. I can only post about 5 comments in a day before HN tells me to "Slow Down, You're Posting Too Fast".

Reply on news.ycombinator.com