How people judge your intelligence and social skills based on your looks: Take 2 | Hacker News Reader

How people judge your intelligence and social skills based on your looks: Take 2

judg.me

47 pointsbvi28 comments

From personal experience:

I've lost a bit more than a hundred (yes, 100) pounds during the last year. My social skills have improved due to an increase in my confidence.

But here is the interesting part.

People do treat me much better. I get more respect. More smiles. They think I am more intelligent, and show this by listening to what I have to say.

And even though I am an introvert, and somewhat shy, people tell me how "social" I am.

I have not changed the way I dress (jeans, t-shirt, sneakers). Though when I'm wearing exercise clothing (running shorts, a sleeveless shirt, running shoes), people are more likely to approach me.

So, this person does have a good point. We are indeed judged by our appearances. In fact, our social standing is calculated on the fly by others on how we look.

Edited to fix spelling disaster.

Those graphs are absolutely vile.

I kept wondering how anyone who chooses hard-to-read shonky 3d bar charts like that could possibly perform a decent statistical analysis.

Agreed, not the prettiest graphics, but you shouldn't dismiss the data & findings because of it.

Unless of course you judge the validity of a study based on the appearance of its graphs?

Well, the title of the blog post is "First impressions count".

Also, the irony of that title, and the way this is a second go at this blog post...

To me, authors decision to use terrible 3D graphs and non zero Y axis makes makes me suspicious of his other decisions and comments.

Agree that 3D graphs are terrible, but it's a canard that y-axes always need to start at 0 and that doing anything else is somehow misleading.

On this point, see http://krugman.blogs.nytimes.com/2011/09/14/axes-of-evil/

Y-axis need to start at zero when the data being presented is a bar graph and the bar is filled in. If it is a scatter plot with points being plotted, then it's generally not misleading to shift the axis.

I don't disagree with that - a bar chart that doesn't start at zero can be misleading. But for what he was trying to do---compare means---it would have been a perfectly fine choice to do a lineplot/very thin bar plot and start the y-axes somewhere other than zero if he included error bars and clearly labeled things.

http://www.edwardtufte.com/bboard/q-and-a-fetch-msg?msg_id=0...

Annoyed me so much I quit reading half way through.

Not just that it is ugly, but also hard to read. ugh.

Graphs updated, as noted above.

Last time something from judg.me was posted, there was a lot of criticism about your statistical methods of analysis. You've clearly put a lot of effort into attempting to do a better analysis of the data, and this time there's certainly an improvement in terms of actual analytical methods used.

However, the graphs (as others have said) are terrible. Simply having a graph type available to you doesn't mean you have to use it! For the sake of your readers, please, please, only use one type of graph. Stick with a plain bar chart. Nobody minds that it's not as fancy looking as a pseudo-3D bar chart or one with pointed cones. People want to see the data, not the fancy chart.

An additional point to this is labelling your axes. Please do it! If I can't see clearly what an axis represents, I'm inclined to ignore your data. Regardless of how interesting your data is, I'm going to stop reading if it's presented terribly.

These are the sorts of things that get hammered into you in scientific subjects at school/college. You don't need to do that type of class to know them though. There's plenty of information online about this stuff.[1] Worth noting is section 2.6, "Do not misuse the tools". Bad graphs are to statistics what the <blink> tag is to HTML.

[1, The UNs guide to presenting statistics]: http://www.unece.org/fileadmin/DAM/stats/documents/writing/M...

Agreed - and I've updated the graphs. Thanks for the comments!

Please consider showing more of the variation and including more information about it in your results.

You conclude, for example, that "In general, it can be concluded that people can and do make snap judgments about personality and abilities based on superficial physical characteristics."

However, a cursory stroll through ranking people on your site reveals that actually, peoples' "snap judgments" show very high subjective variability, and you see these graphs which just have data points all over the place, looking more like a uniform distribution than a clear "everyone things you look smart and extroverted."

Given this, can you profile a burst of ratings? Can you say that there are a coherent group who rated Alice as smart extrovert and Bob as smart introvert and Carol as dumb extrovert -- and can you profile their preferences? Or do most people just browse through your site clicking randomly on the axes without any real reflection on the people they're rating?

We could see this if we could see error bars, but aside from your very first image where you try to show that there is a negative correlation between smartness and sociability, info on variability seems quite nonexistent. (And even there, it is clear that the variability dwarfs the trendline.) Also, you might want to see if there are systematic biases -- what happens if you flip the introvert/extrovert axis for some IP addresses?

Agreed. All plots should be scatter plots with line-of-best-fit drawn over it.

You claim to have proven that people make snap judgments based on superficial physical appearance. But it's based on a test in which you explicitly demanded that the test subjects do so. Failure to do so meant non-participation in your test. How can you claim to have tested the hypothesis?

Those graphs are truely horrible, especially the 3D cone/tower chart for the race vs. intelligence data

When showing similar types of data try to use the same type of graph. Once the reader has learnt to interpret results from one graph they can easily apply it to the next. It looks like you've cycle through every graph type Excel has to offer.

Other than that, quite interesting!

Used to have a terrible posture until it gave me back problems. Spent a week correcting it only to be shocked at the difference it made to my attractiveness to women - wish I'd known earlier '-(

Standing tall, head and shoulders back, making eye contact - almost everyone will treat you differently than when slouched or submissive-looking.

The research is a good idea, an investigation of the halo effect. The population of graders wasn't mentioned, that can have a huge effect on the result. Also, was 1000 people enough to analyze 12 variables?

However, I have to say, the graphs are ugly. Eg, spikes aren't good, you could have gone with with a bar chart. Also, 3d graphs are unneeded, and in general are very hard to do right. They can't convey much information. With just 2x2 variables, there are 4 datapoints. That would be fine on a 2d picture.

Otherwise keep on with the research...

Great idea but the size and scope is too small to be meaningful.

Also, asking people a direct question will often get you a dishonest answer. Maybe there's a way to ask the same question in an indirect way?

Please read Tufte's advice on graphs. 3D graphs are especially irritating.

How people judge your intelligence based on the looks of your graphs: Take 1.

Haha, fair point. I've updated the graphs!

Wow, this made an even bigger difference than I would have expected. Nicely done!

"Click here" in the beginning has no link.

I'm astonished at the finding showing baldness having no significance for men.

This would be an awesome resume piece to get a job at OKCupid.

Reply on news.ycombinator.com