

What I find fascinating about these graphs is that, first, they identity a cross-over. In 2004 “terrorism” was searched for more frequently than “global warming”, but by 2006 they had switched places. It is not clear how significant the difference is, since google does not provide a scale, though my guess would be that it’s logarithmic. A second insight, apparent by the bottom “news reference volume” graph, is that the news is heavily skewed towards reporting terrorism relative to global warming. In 2004, where “global warming” was nearing “terrorism” in searches, its news volume barely left the x-axis (there is no scale or numbers provided, presumably for privacy concerns). By 2006 the volume of news reference to “global warming” was catching up, but as is apparent by the next graph spanning 2004-2007, they are still greatly disproportionate.
There are severe limits to inference with google trends. A keyword might have other referents. For example, trying to determine which university (Berkeley, Stanford or Yale) is most popular would seem to give the prize to
Furthermore, it’s not clear exactly what it means that someone searches for a specific word. Someone could search for “terrorism” because they are interested in participating in or opposing terrorism. All google trends can tell us is the approximate relative magnitudes of the frequency of various searches over time and by region. It’s a good start.
No comments:
Post a Comment