1

votes

Why the spike in Google searches in Sept 2011 for autoimmune?

Answered on August 19, 2014
Created August 22, 2013 at 3:51 PM

While using Google Correlate to visualize the search history for disease-related topics, I noticed a massive spike in searches for autoimmune (or autoimmune disease) around Sept 2011.

http://www.google.com/trends/correlate/search?e=autoimmune&t=weekly&filter=autoimmune&p=us

I wondered whether the spike resulted from increased stress due to the 10 year anniversary of 9/11, but there doesn't seem to be a corresponding peak in terms such as stress, depression or anxiety (though it's interesting to note that anxiety is most strongly correlated with inflammation and migraine - which won't suprise many avid paleos).

Anybody have an explanation for this?

F02990386b12528111740ad6279ba29d

(1363)

on August 25, 2013
at 06:10 PM

Makes it a little more difficult to interpret. Is our expected value 50,000 searches with a standard deviation of 100? -Could be anything to cause that many standard deviation difference. Or is our expected value 50,000 with a standard deviation of 20,000? -Whoa, why did we get 5 times the normal search volume? -I would guess something weird with the data in that instance. Obviously neither of these cases are realistic, but we would need more information to come up with any decent guess on why that happened.

F02990386b12528111740ad6279ba29d

(1363)

on August 25, 2013
at 06:09 PM

Makes it a little more difficult to interpret. Is our expected value 50,000 searches with a standard deviation of 1,000? -Could be anything to cause that many standard deviation difference. Or is our expected value 50,000 with a standard deviation of 20,000? -Whoa, why did we get 5 times the normal search volume? -I would guess something weird with the data in that instance. Obviously neither of these cases are realistic, but we would need more information to come up with any decent guess on why that happened.

F02990386b12528111740ad6279ba29d

(1363)

on August 25, 2013
at 06:00 PM

The spike is not unreasonable. and there are other similar sized spikes. The graph makes it look crazy, but it isn't really.

F02990386b12528111740ad6279ba29d

(1363)

on August 25, 2013
at 05:58 PM

But I'm going to go back to my original guess that it's something weird going on with the data, seeing as how it's also highly correlated with "tyler the creator songs" during that time period.

F02990386b12528111740ad6279ba29d

(1363)

on August 25, 2013
at 05:58 PM

My mistake, you are right. It's normalized though, so it's not representing a pure volume increase...

F02990386b12528111740ad6279ba29d

(1363)

on August 25, 2013
at 05:48 PM

Normalized. It's not pure search intensity. The graph I linked is pure search intensity. It looks like there was a spike but nowhere near the magnitude shown in that graph.

32f5749fa6cf7adbeb0b0b031ba82b46

(41757)

on August 24, 2013
at 08:56 PM

That graph overlays 2 searches, but the graph still just shows normalized search intensities. You can change the second search term and the intensity of autoimmune remains unchanged.

F02990386b12528111740ad6279ba29d

(1363)

on August 24, 2013
at 07:21 PM

http://www.google.com/trends/explore?q=autoimmune#q=autoimmune&cmpt=q

F02990386b12528111740ad6279ba29d

(1363)

on August 24, 2013
at 07:20 PM

So that means there was not a proportional spike in google searches for autoimmune during that time period. There was a spike in CORRELATION. Between Autoimmune and peanut butter searches.

F02990386b12528111740ad6279ba29d

(1363)

on August 24, 2013
at 07:19 PM

Wait, this is google CORRELATE. not google TRENDS.

F02990386b12528111740ad6279ba29d

(1363)

on August 24, 2013
at 07:15 PM

It's possible that if there is super low volume of these search terms that it's not some sort of data error or misrepresentation. But it just seem like something is going on with the numbers that is making the graph not represent what it is intended to in that period of time.

F02990386b12528111740ad6279ba29d

(1363)

on August 24, 2013
at 07:08 PM

It's possible that if there is super low volume of these search terms that it's not some sort of data error or misrepresentation. But the odds are something is going on with the numbers that is making the graph not represent what it is intended to in that period of time.

6864d23c49952605b2a97d6256af804d

(726)

on August 23, 2013
at 04:47 PM

Definitely seems odd, and I wouldn't discount data error, but the point of looking at the correlation is to figure out which other search terms varied at the same time. It's not that unusual to have a spike in search terms.

2fd93e91bb14e641a2bac9c6033e84e2

(1614)

on August 22, 2013
at 11:26 PM

Dr. Oz, Dr. Weill, etc... I would bet one of those guys had a show/segment on an autoimmune disease. I heard Dr. Weill on the radio in late December talking about inflammation, which is what got me googling and brought me to Paleo.

61f9349ad28e3c42d1cec58ba4825a7d

(10480)

on August 22, 2013
at 10:41 PM

Was there some celebrity that announced they had an autoimmune disease, or was funding an autoimmune disease charity?

6864d23c49952605b2a97d6256af804d

(726)

on August 22, 2013
at 05:10 PM

The Fukushima disaster happened in March, and shifting the time series by 6 months didn't show much. Good idea though!

A6b7afacdb512b861716bca9e9226ad3

(333)

on August 22, 2013
at 04:56 PM

could it be out of concern regarding the radiation from the nuclear plant in Japan following the tsunami?

A6b7afacdb512b861716bca9e9226ad3

(333)

on August 22, 2013
at 04:49 PM

what a fascinating observation! I'd be very curious to find out what else was going on at that time, or possibly during the previous year. I love your exploration...

F5a0ddffcf9ef5beca864050f090a790

(15515)

on August 22, 2013
at 03:55 PM

That graph is crazy. Check all the other variables - they all went up.

F5a0ddffcf9ef5beca864050f090a790

(15515)

on August 22, 2013
at 03:53 PM

Cortisol levels went up?

  • 6864d23c49952605b2a97d6256af804d

    asked by

    (726)
  • Views
    904
  • Last Activity
    1284D AGO
Frontpage book

Get FREE instant access to our Paleo For Beginners Guide & 15 FREE Recipes!

2 Answers

1
F02990386b12528111740ad6279ba29d

(1363)

on August 22, 2013
at 09:20 PM

Edit: That graph doesn't represent a spike in searches for autoimmune.

autoimmune + craving peanut butter. correlation of two search terms. easily could have spiked from something on television since that probably isn't a common search term.

F02990386b12528111740ad6279ba29d

(1363)

on August 24, 2013
at 07:19 PM

Wait, this is google CORRELATE. not google TRENDS.

F02990386b12528111740ad6279ba29d

(1363)

on August 25, 2013
at 06:10 PM

Makes it a little more difficult to interpret. Is our expected value 50,000 searches with a standard deviation of 100? -Could be anything to cause that many standard deviation difference. Or is our expected value 50,000 with a standard deviation of 20,000? -Whoa, why did we get 5 times the normal search volume? -I would guess something weird with the data in that instance. Obviously neither of these cases are realistic, but we would need more information to come up with any decent guess on why that happened.

F02990386b12528111740ad6279ba29d

(1363)

on August 25, 2013
at 06:09 PM

Makes it a little more difficult to interpret. Is our expected value 50,000 searches with a standard deviation of 1,000? -Could be anything to cause that many standard deviation difference. Or is our expected value 50,000 with a standard deviation of 20,000? -Whoa, why did we get 5 times the normal search volume? -I would guess something weird with the data in that instance. Obviously neither of these cases are realistic, but we would need more information to come up with any decent guess on why that happened.

F02990386b12528111740ad6279ba29d

(1363)

on August 24, 2013
at 07:20 PM

So that means there was not a proportional spike in google searches for autoimmune during that time period. There was a spike in CORRELATION. Between Autoimmune and peanut butter searches.

32f5749fa6cf7adbeb0b0b031ba82b46

(41757)

on August 24, 2013
at 08:56 PM

That graph overlays 2 searches, but the graph still just shows normalized search intensities. You can change the second search term and the intensity of autoimmune remains unchanged.

F02990386b12528111740ad6279ba29d

(1363)

on August 25, 2013
at 05:48 PM

Normalized. It's not pure search intensity. The graph I linked is pure search intensity. It looks like there was a spike but nowhere near the magnitude shown in that graph.

F02990386b12528111740ad6279ba29d

(1363)

on August 24, 2013
at 07:21 PM

http://www.google.com/trends/explore?q=autoimmune#q=autoimmune&cmpt=q

F02990386b12528111740ad6279ba29d

(1363)

on August 25, 2013
at 05:58 PM

My mistake, you are right. It's normalized though, so it's not representing a pure volume increase...

F02990386b12528111740ad6279ba29d

(1363)

on August 25, 2013
at 06:00 PM

The spike is not unreasonable. and there are other similar sized spikes. The graph makes it look crazy, but it isn't really.

F02990386b12528111740ad6279ba29d

(1363)

on August 25, 2013
at 05:58 PM

But I'm going to go back to my original guess that it's something weird going on with the data, seeing as how it's also highly correlated with "tyler the creator songs" during that time period.

0
Medium avatar

on August 22, 2013
at 10:16 PM

I would have to agree with Kyle. I'm no expert but all those other "closely correlating" terms spiked in the same exact way at the same time. Seems odd.

On that note - how do I comment on his post? I could swera I knew how to do this...

6864d23c49952605b2a97d6256af804d

(726)

on August 23, 2013
at 04:47 PM

Definitely seems odd, and I wouldn't discount data error, but the point of looking at the correlation is to figure out which other search terms varied at the same time. It's not that unusual to have a spike in search terms.

F02990386b12528111740ad6279ba29d

(1363)

on August 24, 2013
at 07:15 PM

It's possible that if there is super low volume of these search terms that it's not some sort of data error or misrepresentation. But it just seem like something is going on with the numbers that is making the graph not represent what it is intended to in that period of time.

F02990386b12528111740ad6279ba29d

(1363)

on August 24, 2013
at 07:08 PM

It's possible that if there is super low volume of these search terms that it's not some sort of data error or misrepresentation. But the odds are something is going on with the numbers that is making the graph not represent what it is intended to in that period of time.

Answer Question


Get FREE instant access to our
Paleo For Beginners Guide & 15 FREE Recipes!