How Different Groups Use Their Time | Hacker News Reader

How Different Groups Use Their Time

nytimes.com

108 pointskingkawn38 comments

This graph does not show how people use their time. It shows how people claim to use their time. The caption clearly says "The American Time Use Survey asks thousands of American residents to recall every minute of a day." Not only is it just what people claim to be doing, it's what they claim to be doing while later reflecting on it. I don't have the slightest idea what I was doing at 1:00 this afternoon. I would probably guess eating lunch, but that could just as likely be wrong. Also, I'd be much more apt to claim to be eating lunch than wasting time with the TV or by just sitting around doing nothing.

I'm not suggesting the graph is completely unreliable. I'm just saying you need to consider that certain aspects of the graph will be influenced by the human nature to lie and to remember imperfectly.

Assuming you have a statistically sound sample size, then these imperfections should even out overall.

Not really. People have all sorts of biases that will come out when self-reporting. Things which carry a positive connotation, or that a person is proud of, will be over-reported. Things that are viewed negatively will be under-reported. Things which are embarrassing or controversial will be skewed as well.

If if was random, normally distributed error, then the errors would cancel out. However, many questions will be biased in one direction, not randomly around a true median.

Lots of interesting stuff in here. Computer usage seems way too low though. Who are these people?

The ATUS categorization doesn't have a top-level category for computer use, and some varieties of computer use get recorded in ways that don't indicate that a computer was involved. So the actual amount of computer use time might be bigger than reported, both because not all computer use is explicitly recorded as such and because whoever made the interactive graphic might have missed some categories.

Which just leads me to say, if you can't report something accurately, don't report it at all. Reporting an average of 8 minutes a day of computer use is just absurd.

Ordinary people.

I assume a lot of people's work time involves computer use, but wouldn't be counted as such on this graph.

Breakfasts are quite faint, unlike lunch and supper. It seems like people with higher education spend more time eating in the morning, they also have lunch later (makes sense).

I wonder if it is because they can afford the time and know the benefits, or if it is a good practice that made you more successful in school.

The quiet presence of sleeping is amazing. Also I'd be interested if there were regional comparisons.

Most depressing finding: People age 65+ watch more TV than any other demographic. If that's what I'm looking forward to, then I am not retiring.

Though it does explain why all TV ads are for drugs...

Conclusion: think about how your startup helps people with their work.

Looking purely at the graph, I'd say the conclusion should be "think about how your startup will be more engaging than TV".

TV isn't engaging, but very passive. In fact going for engaging might not be a good thing.

When deriving insights into target use-cases for tech startups, the graph you want to look at is the one for "employed" people. Serving unemployed people is probably a poor choice ;)

Excellent graphic! Would make Tufte proud.

I clicked on the different demographic buttons and also moved the mouse around and read the tool tip. Was able to come up with quite a few of my own hypotheses doing this. Amazing what you can do with an organic app over one that just sits there.

I need to find a way to put this kind of thinking into my own apps. My users would love it.

I'm not sure Tufte would agree. This is data that would probably be better served by a table in many ways.

For example, it's easy to see the large-scale differences between demographics using this animated graph, but the details are difficult to grasp. And since the differences in these graphs tend to be in the details, I'm not sure that the selected visual treatment is particularly useful. Why not an animated table, with significant changes highlighted on transition?

Click on the detail you want to track.

Yes, I saw that. But that just zooms you into a specific type of task, where you lose all context. You also gain no extra insight into the differences between demographic groups.

Really, this is just an attempt to put a lot of data into a stacked bar-chart by making it clickable. It's not particularly judicious in its presentation.

You can see the differences really clearly when switching between groups - so I think the graphic is doing its job.

A table would end up being 24 x ~15 - tough to notice anything in particular as every number is going to change somewhat.

Eh, all I got was a big Flash error button. :\ I would have liked a table compared to having nothing.

That's because you don't have Flash installed. Simply install it and you'll be fine. The visualization's whole purpose is to be able to see (visualize) the change as you change the demographic. You can't do that as a table. If you don't want to have Flash, then this, and many other visualizations just wont be something you'll be able to see.

The thing is that I do have flash installed, it just had some sort of error. If I didn't have flash I would get the little blue box with a ? in it.

Really interesting way to see the data. Tufte would indeed be pleased with this.

TV usage seems to be inversely proportional to higher levels of education and increasing number of children.

There was nothing really very surprising though, except maybe the racial differences about which I'd not even be able to begin to guess as.

One thing I found interesting was that those with no children seemed to work considerably less. Perhaps it's an age/career-centric thing. Also to note, those with children spend quite some time in "Family Care" which to me sounds like maintenance moreso than leisure.

Could also be a statistical blip. People with no children includes children themselves (hopefully ;-)), who probably will not be working. Averaged between them and 20-something yuppies, working hours will be less.

The part that really surprised me is that sleep does not go down much between having no children and one child. It does go down between 1 child and 2+ children, but not by as much as I'd thought.

The ATUS only covers people aged at least 15, so not exactly children. But yes, this is surely an age effect: university students, e.g., will typically have no children and do less (of what the ATUS classifies as) work.

Interesting that across the board, chances are everyone is awake at 7:10 pm versus noon.

Kinda off-topic, but I'm surprised sex wasn't included.

(I don't see why it's off-topic.)

Presumably categorized as "other leisure" or (less likely) "socializing".

... Nope, wrong, it's actually categorized as "personal care". At least, that's the top-level category it goes in in the survey itself, but whoever created the NYT graphic may have organized things differently. (In fact, they definitely did, since they have a "computer use" category and that is certainly not a top-level category in the ATUS.)

It is. See the description for "Personal Care". They say the average for sex is 54 seconds. :-)

Nice example of the distinction between different kinds of average (see the "lying with statistics" thread). This is presumably the mean rather than (say) the median. Perhaps, e.g., 1/3 of the population they sampled is in a reasonably healthy sexual relationship and not grossly hampered by extreme old age, ill healthy, etc., in which case the average for that portion of the population is about 3 minutes per day, which could mean e.g. once a week for 20 minutes.

Then, of course, some respondents might be reluctant to report that they had sex (pushing the average down) or might tend to exaggerate how long they spent at it (pushing the average up) or something.

The one thing the figure almost certainly doesn't mean is that a typical sex act reported in the survey took 54 seconds. Although, depending on how you define your terms, many of them might be very quick...

It's also really interesting the differences in sleep habits between white/hispanic vs. black.

There's an even greater discrepancy between people with High school vs advanced degrees, but I'd be willing to bet that in both cases the difference is accounted for by people doing low-paid night work rather than any inherent preference for staying awake until 3am.

But if we're going off of general population types, then wouldn't that assumption apply to low-paid hispanic immigrants? I think the difference between black and the other designated races is compelling and worth looking in to.

Also, why no Asian differentiation?

It's certainly interesting - there might be more data in the government report.

Awesome visualization.

Proof that our society dictates roles & routines based on age. I think this will change in the future.

Reply on news.ycombinator.com