Google Trends Can Measure Discrimination?

Data Journalism

June 27, 2023


A few questions that social scientists have been asking for a while are… how do you measure something like discrimination? How do you measure the intensity of discrimination? Also, is it possible to generalize this to a population? For example, X group is strongly discriminated against in the United States, or X group is strongly discriminated against in the world.


  • Method 1: You might be able to do this with laws. For example, if there are laws based on race in a community, then discrimination is clearer. But in a country like the US, as you go out from a local level, there exist more and more laws, thousands! It seems difficult to go law by law through thousands of laws searching for discrimination. Also, this type of search doesn’t account for people who may discriminate against others outside the law.

  • Method 2: Social scientists try to use series of survey questions to detect attitudes towards discrimination of race and gender, but who wants to admit things like being racist or sexist? Even subtle attitude questions about discrimination could be detected by the survey respondent.

If you’re like me, these both don’t seem like great measures and have plenty of room for error. This is definitely not an exhaustive list of possible measures, but none come to mind that are reliable, accurate, and can be applied to large regions or populations. If I misssed some, feel free to comment below..

Another Measurement

Recently, I read a book by Seth Stephens-Davidowitz entitled, “Everybody Lies : Big Data, New Data, And What The Internet Can Tell Us About Who We Really Are.” He shares interesting facts and trends from what he calls the “digital truth serum,” Google Search data.

He proposes that we lie all the time to our friends, neighbors, family, and even to ourselves. But, there is one person we don’t lie to… GOOGLE. Using google trends data, we can see how often people searched a certain term over time and compare it with other searches. While the book overall was interesting and would recommend it to others, I want to focus on one particular part.

One interesting fact about google searches is people often declare statements/confessions - not questions - on google searches. For example, they might say “I’m pregnant” or “I’m gay.” Even dark things like, “I hate myself” or “I wanna kill myself.” It really is a gold mine of interesting data that doesn’t have filters like words your friend might say to you or how a survey respondent might answer.

In his book, he discusses a paper he wrote linking Google searches for “nigger jokes” and “nigger” to election differences in the US for Obama.

All of this led my to my own question and query. Perhaps one other avenue in looking for individuals most discriminated against is using google search data like Seth Stephens-Davidowitz.

So, I made a chart for google search data using statements for “kill [group]” specifically in the US. The results surprised me in several respects.


This is the frequency of searches of these exact words. These are not the frequency of searches that contained these words!


If you are a mobile viewer, the aggregation is by year instead of by month. You can see some of the trends, but the histogram doesn’t display referenced in Observation 2 and you can’t see the monthly seasonal trends referenced in Observation 3. I recommend you view on desktop if possible 🙏

Observation 1:

One of the first things that catches the eye is the spikes in the different colored lines. The highest spike is of the “kill blacks” line which is in June 2020. This was also the same time that the BLM marches and videos of looting were circulating which tracks. Also check out this link from BBC which documents several of the police shootings of African Americans - several of which correspond to spikes on the graph.

If you hover, you can see the “kill chinese” line spikes in March 2020 and has stayed a bit higher than before ever since. You hopefully know the reason for this spike.

The “kill muslims” line spikes in November 2015 the same time was when the paris attacks occurred.

Noticing and researching spikes helps reaffirm our claims about statements. The times when anomosity might be highest toward a group in the US group is also when “kill [insert group]” line spikes. Also, I didn’t name them all feel free to reseach some yourself you see on the graph!

Observation 2:

One of the overall takeways I got from this figure is the little histogram to the left hand side which gives an average relative interest overall. I’m not sure which group I was expecting to be the highest, but I didn’t think it was the “kill jews” search. This search seems to be almost twice as likely as the second higest which is the “kill chinese” score. Followed by the “kill muslims” score, the “kill blacks,” and lastly, the “kill gays” score. You can only compared 5 at a time, but I also tried including other groups which were lower than these including “kill christians,” “kill mormons,” and a few others. “Kill whites” was similar to “kill muslims” in frequency and followed similar spike patterns to “kill blacks,” but on a lesser scale.

Looking at the histogram, this suggested to me that Jews and Chinese are perhaps more discriminated against today than I previously had thought.


If you are a mobile viewer, there is no histogram, but if you imagine taking the average by google search Jews obviously has the highest number, so on.

Observation 3:

One thing that i find very interesting about this graph is the cyclical pattern of the “kill jews” search. It seems to follow a yearly pattern beginning at a low point in July / August then spiking in April. I don’t fully understand why this pattern exists, but perhaps it could be related to Easter - which is in April - when the Jews crucified Jesus Christ, a central figure in Christian religions (a lot of the US is Christian). This theory wouldn’t explain the gradual rising though, so if you have suggestions of why this is the case, please email me or something.


If you are a mobile viewer, you’ll just have to trust me on this one 🙂.

Observation 4:

While it does seem the US has more distaste for jews, chinese, and muslims, still note that African - Americans and gay people are still on the map. All of this is according to our measure using Google Search data, of course.


This is by no means a perfect measure of discrimination against a group, how many people really are searching for this type of thing? Also, it would be so valuable to see demographic breaks by searches. But concerns aside, I think it is a new and interesting take on measure of discrimination and distaste for a particular group. And perhaps it is a valuable signal because of the spikes we see that correlate with real events that might cause more hate/discrimination for a particular group.

If we believe for a second that it is a good measure, we may consider groups we may not hear about in the news as much as subjects of American discrimination (jews, chinese, and muslims) more than we give credit for.

Feel free to leave relevant and thoughtful comments below 🙂