On the mundanity of American family life

The world has been through a lot since 2004. Some things don’t seem to have changed much.

I’ve stared at this pattern for a while, and if there are trends to it — evidence of the recession, the wars, the elections, or the rise of social media, for example — I don’t seem them. You can view or download the underlying data here. Maybe you’ll do better.

I can’t predict the future with certainty, but I bet both searches will peak this year in early June. Is this reassuring, numbing, boring, or inspiring?

2 Comments

Filed under In the news, Me @ work

2 responses to “On the mundanity of American family life

  1. I think I found some patterns! The overall seasonal trend kind of overwhelms other patterns that might be involved, but you can control for that.

    First, I just plotted the data for each month, with a line for each year. What’s especially clear is that in the months of January and February, the volume of searches for “summer camp” has been decreasing.

    http://imgur.com/ZDS5w

    It’s possible to deseasonalize data like this. I am by no means an expert on time series analysis, but what I did was fit a linear regression model with predictor terms Month, Week of the Month, and their interaction. From this model, you can take the residuals, and that gives you a value for how unusual the data for a particular week from a particular month from a particular year is compared to that same week of the month from other years.

    The result was this plot, which has Year-2000 on the x-axis and the residual values on the y-axis, plus smoothing local regression lines.

    http://imgur.com/ePpbH

    I did the same thing with the summer school data, and plotted the whole thing, plus some super smoothing regression lines.

    http://imgur.com/gScCZ

    It looks like googling for “summer camp” has been decreasing, while googling for “summer school” has been relatively stable. You’ll notice some extreme up and down spikes here and there. I haven’t tried to connect those to anything in the world, but they might be

    1) artifacts of my deseasonalization (failures that is)
    2) noise in Google’s data collection
    3) something real.

    Like

  2. OK, you lost me at “R”. But, I appreciate this.

    I guess it would be unlikely that nothing was going on — it just took some torturing to make the data confess. It’s hard to tell since we (I) don’t really know what the metrics are with Google, but given that the patterns you discover seem slight in relation to the overall monotony, I wonder how big a trend like you find has to be to matter.

    (I wonder how error-ridden the Google stuff is. The overall pattern here is reassuring. But in the Ngrams data — where I’m looking for neat things like first-occurrences of rare terms — I keep finding “books” that are misdated.)

    Thanks!

    Like

Comments welcome (may be moderated)

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s