6

Is it common in academia to include the number of results a Google search on a keyword yields as an evidence for the prevalence of something online?

I'm not intending to base my research paper on this evidence; it is just a supporting point in a paragraph.

For example, I'm writing a research paper on Stack Exchange; one of the paragraphs is about how popular Stack Exchange is in the internet.

In one of the paragraph, I would mention that a Google search on the keyword 'Stack Exchange' yielded about 49,600,000 results, which shows how popular it is among netizens.

Improve this question
10
  • 5
    Are you only interested in online presence of the "something"? A Google n-grams (books.google.com/ngrams) result would be more robust. It covers books, and gives you year-by-year counts.
    – mhwombat
    Commented May 22, 2014 at 11:32
  • 1
    This question appears to be off-topic because it is not about academia
    – 410 gone
    Commented May 22, 2014 at 14:14
  • 1
    @EnergyNumbers Questions about whether a kind of source should be used (e.g. "Can I cite Wikipedia?") seem to be on-topic, according to our history as a community.
    – ff524
    Commented May 22, 2014 at 14:21
  • 2
    @EnergyNumbers I don't think the OP's research area is necessarily web metrics - he/she is asking about making a "supporting point in a paragraph" that some topic X is prevalent online, the kind of thing that could go towards motivation for any kind of research (e.g. "We address problem X, which according to the volume of Google searches has been the subject of a great deal of discussion..")
    – ff524
    Commented May 22, 2014 at 14:39
  • 1
    I agree with ff524, the problem is about the trust/usage of a source, regardless of the field (which, as far as I can tell, is not specified). Also, the comment "This question appears to be off-topic because it is not about academia" is not-constructive.
    – user102
    Commented May 22, 2014 at 15:12

4 Answers 4

Reset to default
13

Do not do this. The "approximate" number of hits Google reports is completely worthless. To see why, look at this number on both the first and the tenth page of Google hits:

First results page

Tenth results page

When I just did this, I got "approximately 9,010,000 hits" reported on the first results page... but only "approximately 48 hits" on the tenth page.

Your results will probably vary, depending on your search engine bubble (another reason why this number is useless).

EDIT: It seems like Google at least does not report the number of hits any more as they used to. Which, per my answer, was an absolutely worthless piece of "information" - the first results page would report "about" a very large number of hits, and the tenth page a much smaller number. I will leave this answer up.

Essentially, I would recommend that if you are interested in the number of hits a search serves up, do not trust the number it self-reports without stress-testing it, e.g., by clicking through to the tenth or fiftieth page of results and seeing whether the reported number of results stays the same (or at least decreases sensibly: if there are 50 results per page, then at the end of the tenth page, you should have N-450 "more hits" if there were N "more hits" at the bottom of the first page).

Improve this answer
9
  • 1
    FWIW... I haven't seen Google counts in a paper, but I have seen academic blogs like Language Log (languagelog.ldc.upenn.edu/nll) use them. However, they always emphasise that they used the count on the last page of the results, for the reason Stephan mentions.
    – mhwombat
    Commented May 22, 2014 at 11:25
  • 1
    I know the "bubble" affects the order in which links are presented, but would it affect the word count? (I don't know the answer, just curious.)
    – mhwombat
    Commented May 22, 2014 at 11:35
  • 1
    +1 - Even if people do this it is not justified. Another example of inappropriate analysis was Google Flu trends.
    – Andy W
    Commented May 22, 2014 at 11:41
  • @mhwombat: good question about the filter bubble and the word count. I have no idea. How many "approximate hits" do the two links report for you?
    – Stephan Kolassa
    Commented May 22, 2014 at 13:20
  • Someone seems not to agree and downvotes. I would be interested in knowing how this answer is not helpful. Would downvoters be so kind to comment on this?
    – Stephan Kolassa
    Commented May 22, 2014 at 14:25
2

I don't think such numbers would be accurate at all, altough they could probably be used as a rough, theoretical estimate. 2 reasons I can think of for why this would not be accurate at all:

  • Typos. Have you ever tried googling for 2 different spellings in order to figure out which is most likely the correct spelling? Unfortunately, the most used(popular) spelling is not always the correct one. What this means is that the number of google hits is not a picture of reality, but a picture of what is in people's thoughts.

  • A phenomenon\activity\product being written about on the web is not equivalent to actually being used alot. What's being written on the web is primarily a picture of what's going on in people's heads, rather than what's actually around them. A new invention could be ground-breaking and get mentioned alot, even though only scientists would use it.

Improve this answer
1
  • 2
    The question is asking about measuring "the prevalence of X online" not "the prevalence of X in reality"
    – ff524
    Commented May 22, 2014 at 13:36
1

I honestly haven't seen something like this in any publication. But I guess you could use it as some by-point, I would in that case provide the number of result of other popular search engines, something like: "xyz is highly sought after in the world wide webs, a simple key search 'xyz' using the most popular search engines yields impressive number of results (google: 49,600,000, bing: ..., yahoo: ..., etc.)"

Consider, however, that the search results also include ambiguous results for the provided key word(s). In your example, it is likely that that result of about 49,600,000 will include hits that contain "Stack" and "Overflow", but not the semantics that bind them into your intention. Please bear in mind that I'm aware that you are able to narrow your search down to a particular key word by using google tools, tweaks and skill, I'm merely providing an example, as depending on the "commonness" of the key word(s), the explicit narrowing down can be complex.

Improve this answer
-1

The popularity of a web site is only very loosely related to the number of search engine results. A better approach is to use traffic statistics, such as the web site's Alexa ranking:

http://www.alexa.com/siteinfo/stackoverflow.com

This places StackOverflow at rank 53, which means that only 52 sites are visited more frequently.

Either way, StackOverflow is so ubiquitous within the programming community that you probably don't need to qualify your claims of its popularity unless your paper is intended for non-programmers.

Improve this answer
2
  • 3
    I believe the OP is not necessarily trying to claim something about the popularity of a website, but the popularity of some generic keyword/topic. (StackExchange was just an example.)
    – ff524
    Commented May 22, 2014 at 15:50
  • Given the way the example is phrased, I think it's reasonable to believe that there is value to the author in having the example answered. If the intent of the question is about the popularity of a web site, this answer is still directly useful.
    – bdean20
    Commented May 22, 2014 at 16:56

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .