Whaddya know? This turned out to be simpler than what I thought--for an unexpected reason.
Here's what I did. My first query was to see if I could find any database that had reasonable information about how Americans spend their time day-to-day.
[ Americans spend time ]
This query led me quickly to the American Time Use Survey at the US Bureau of Labor Statistics (BLS). This is their job: collect data and stats on American behavior across many different dimensions.
I drilled down almost immediately to their data sets about time use and found data from their big survey of ~13,200 Americans in 2010. The metadata for this survey is here (which is where they write down all the details of how the surveys were coded, the survey question form (ever want to see the script that surveyers use when asking you questions--this is it), etc.
Eventually, I got to the summary table of "Time spent in primary activities (1) and percent of the civilian population engaging in each activity, averages per day by sex, 2010 annual averages." I exported that data into my spreadsheet, and generated the following chart rather quickly.
Whenever you look at a chart like this, questions and insights immediately spring to mind. For instance, WHY is the average amount of work / day only 4.09 for men and 2.94 for women? (Or 3.5 average between the two.) Or, to answer my original question--"Do I spend a lot of time in email?" The answer is "yes, you do... but you're in Silicon Valley, what did you expect??"
You might prefer to see the data in this format:
(Same data, just different chart.)
Remember that this is a sample of people nation-wide, balanced across demographic categories (age, gender, location, etc.) and included a representative sample of unemployed people as well. Keep in mind, if you work a standard 40 hour work-week for 50 weeks of the year, you're really only working 4.3 hours/day averaged over the entire year.
Contrast that with the number of hours / day if you're engaged in that activity--7.82, which reflects more what you'd expect for number-of-hours worked given that you're working on that day.
We could continue analyzing the data, which is really interesting, but I want to return to the search question for a minute.
Why did I say this was simpler than I thought?
Answer: Because I had a particular solution path in mind when I started out. I would (1) locate the data, (2) extract it, and (3) analyze it.
I never expected that the BLS would have already done this for me! As gasstationswithoutpumps commented,
"I searched for [ hours spent american statistics ] which got me to
http://www.bls.gov/news.release/atus.nr0.htm which has pointers
to tables for 2010 statistics divided in lots of ways. (in particular...)
Table 12 http://www.bls.gov/news.release/atus.t12.htm
seems to be the one you want, with sleeping and watching
TV as the two biggest categories."
I never expected that they would have already done the analysis, with charts and everything! (Note that this chart from BLS is slightly different than mine--they're charting the time use on an average workday for employed persons with children. The chart above is for everyone in the sample set.)
On the other hand I read analyses of how people spend their time as compiled by various other organizations. Neilsen, for instance reports that people spend 4.9 hours/day watching television. That's very different than the amount that the BLS finds in their data. You can find similar differences for other measures as well (e.g., internet use; amount of time spent care-giving; household activities; ....).
The big takeaway is that you have to be VERY careful that you know exactly what data is going into your analysis. It matters how it was collected, what the coding scheme was, and how the sample was chosen. Was it random digit dialing (as in the case of the BLS data), or was it "Neilsen Family" people? The outcome can vary tremendously.
One of the beautiful things about being able to find data of this kind is the ability for us to download and analyze it according to our own interests. The folks at FlowingMedia did a beautiful interactive version of this same data by building on D3, Mike Bostock's visualization system. The image below is just a single frame from their system. You should visit the FlowingMedia interactive visualization (click on the demographic segments in the upper right) to get the full experience.
Bottom line: Searching for a data that you can easily download, understand and do your own analysis of is a great feature of the new Data Age. You should learn how to get to data AND how to understand it. That metadata stuff turns out to be deeply important.
Yours in finding ever more data!