glaugh

24 ms·

181.
by glaugh 13y ago
Agreed. Sort of a satire of how articles will take some research and draw wild conclusions about it for the headline. We also considered "Psychopaths start companies" and "Some third variable relates to both psychopathy and s
182.
Starting a Company Can Turn You into a Psychopath(blog.statwing.com)
32 points by glaugh 13y ago|21 comments
183.
by glaugh 13y ago
It'd be interesting to get data on this. Thoughts on how that would work? Something like "Take a paragraph, A|B test it on CrowdFlower/Turk with both types of spacing, with the outcome being [?]" Speed? Being able to ans
184.
by glaugh 13y ago
I'm sympathetic to that argument. It's tough to write one comment that addresses two audiences: (1) person who wants some simple, clear guidance (2) you and me who are debating at a slightly different level what advice we should g
185.
by glaugh 13y ago
Here's my favorite summary of the science on salt: http://ije.oxfordjournals.org/content/31/2/311.full
186.
by glaugh 13y ago
I think this thinking is tempting and reasonable, but it misses the role of hormones. As an example, women gain weight during menopause because of hormones, not because they eat more.[1] So thinking of the body and losing weight as being pu
187.
by glaugh 13y ago
We put together some fun histograms from the raw data in Statwing. You can play around a bit more yourself. https://www.statwing.com/open/datasets/cb2e735b5e8428f521828... Thanks OP for making the raw data availab
188.
by glaugh 13y ago
Personally I prefer histograms (assuming you have all the data, not just mean and std dev). For example, here's a histogram of ages of passengers of the Titanic: https://www.statwing.com/demos/titanic#workspaces&#x
189.
by glaugh 13y ago
I received that kind of anti-pitch before I joined CrowdFlower back when, and I found it very compelling. It did in fact turn out to be an accurate representation of the cofounders' culture of transparency, too.
190.
by glaugh 13y ago
I just played around with this for similar reasons: http://www.wideanglesoftware.com/touchcopy Didn't finish up the export because I didn't feel like paying just yet, but getting to that point was pretty easy. $30
191.
by glaugh 13y ago
There's been some rejoicing in my nonprofit nerd circles about a recent acknowledgement by a few important nonprofit rating orgs that overhead is a bad measure of effectiveness: http://www.guidestar.org/rxa/news&#x
192.
by glaugh 13y ago
Hey Rick, I'd humbly suggest that you load the dataset into Statwing, and you'll get visualizations (histograms, scatterplots, etc.) and stats out of the box, as well as a link to share so others can play with the data. Or ping me
193.
by glaugh 13y ago
Skipping to the intense part of the second video in the article, the actual footage from the drone: http://vimeo.com/68229603#t=5m23s Really amazing to see watch a movie-like, birds-eye view of a conflict (water cannons, te
194.
by glaugh 13y ago
Wow, that's an amazing resource. Thanks for pointing out, and I stand corrected.
195.
by glaugh 13y ago
I think it's impossible to replace Excel en masse, so lots of folks are starting to pick off bits of it to replace. And hopefully in doing so they'll invest in making those bits a lot smarter than they are in Excel. Examples from
196.
by glaugh 13y ago
Agreed that the most important point is that liquid carbohydrates and plain sugars are generally bad, and that that's 90% of what people should take away from discussion like this. But it appears that there's still some particularly negativ
197.
by glaugh 13y ago
I typically recommend against buying Vastrm because now that I have a bunch of Vastrm shirts all my other shirts feel like they don't fit that well in comparison. They can't compete. I'm not a super fashion-conscious person but damn these s
198.
by glaugh 13y ago
Yeah, my point isn't that the statistical significance is erroneous or that the difference shouldn't have been significant. In my totally made up example, there's really a .1mm difference. And maybe to someone that's meaningful, so it's wor
199.
by glaugh 13y ago
My biggest pet peeve is the interpretation of p-values absent the context of effect sizes. If you have a huge sample size you're quite frequently going to find statistically significant differences between groups, but often those difference
200.
by glaugh 13y ago
Here's a quick and dirty visualization of the distributions for each kind of care across lots of different hospitals: https://www.statwing.com/open/datasets/c5fc084c00d24cf221c17... Other visualizations today: NYTimes -- http://www.nytim
201.
Are all telephone calls recorded and accessible to the US government?(guardian.co.uk)
1 points by glaugh 13y ago|1 comments
202.
Are all telephone calls recorded and accessible to the US government?(guardian.co.uk)
145 points by glaugh 13y ago|57 comments
203.
by glaugh 13y ago
All the questions that aren't part of that first big bunch are just fun things we can cut the data on later. In retrospect we weren't terribly clear on that.
204.
by glaugh 13y ago
Just want to be really clear that we're not doing any diagnosis, and no matter what you responded we'd never say "That's it, you're a psychopath." Also perhaps 'demographic' is probably the wrong word on our part, point taken :)
205.
by glaugh 13y ago
OP here, totally agree with this. That question is almost "How honest are you with yourself?" Takes a bit of scrolling down the page of results[1], but people are actually pretty honest on that one, the mode response is "Agree somewhat" (th
206.
by glaugh 13y ago
Here's Robert Hare, the expert on this stuff: "We refer to individuals with a very high concentration of these features (perhaps 1% of the general population) as psychopaths" [1] So there's no clear cutoff, it's not a bimodal distribution,
207.
by glaugh 13y ago
First link of that footnote was just to describe the scale, via the Wikipedia article. Second link (on the word 'validated') was to this article: http://www.ncbi.nlm.nih.gov/pubmed/10497804
208.
Are you a psychopath?(blog.statwing.com)
54 points by glaugh 13y ago|59 comments
209.
by glaugh 13y ago
Statwing Front End Engineer Radically improving the UI for data analysis is the core challenge of our company. As our first hire and only front-end specialist, you'll drive the function and feel of our UI, creating entirely new ways of inte
210.
Dataset: U.S. Senate Votes on Gun Control (vs. NRA Donations, Party, etc.)(statwing.com)
13 points by glaugh 13y ago|0 comments

More ›