Analyzing people’s online behavior to better understand our usersAdar Dabush IvanchenkoBlockedUnblockFollowFollowingMay 21Google search and the prediction of Trump winningIn 2008 Barack Obama was running as the first African American nominee.
We all know the end of the story, Obama went on to spend two terms as the President of the United States.
Random surveys were taken before and during his tenure showed that the majority of the American population voted based on merits and demerits, rather than race or color.
The survey concluded that Americans no longer judged or voted based on the color of the candidate’s skin.
This idea can further be associated with the academy and media belief of the decline of racism in America.
The results of the surveys are one of the reasons why nobody thought Donald Trump stood a chance during the Republican primaries in 2016.
Trump had managed to insult more than one minority group.
But he won and there were little clues predicting his ascent to the seat of the president beforehand.
Four years before his winning, Data expert and author Seth Stephens-Davidowitz found these clues.
He used Google Trends to find millions of racist queries made by Americans and the results were surprising.
He found out that the ‘N-Word’ was searched for as many times as common words like ‘migraines’, ‘economist’, ‘Lakers’, even after filtering out rap lyrics song searches.
That was startling in itself, but the most revolutionary data Google trends gave him was a map of racism, that was different from what often assumed.
Google search data suggested that racism wasn’t more common in the South vs.
the North, but actually in the East vs.
The highest racist search rate included upstate New York, Western Pennsylvania, Eastern Ohio, Industrial Michigan and rural Illinois, along with West Virginia, southern Louisiana and Mississippi.
Furthermore, the racist queries were not higher in places where there was a higher percentage of Republicans than Democrats.
Four years after his initial research, this map will be used to explain how Donald Trump won the election.
While the surveys asked if the respondents are going to vote and for who, Stephens-Davidowitz’s did something really simple and smart.
Using Google trends to show areas with potentially higher voting percentages, he searched the queries “Where to vote” and “How to vote”.
The logic behind this was that citizens who are actually going to vote are more likely to look up ‘Where to vote’, and ‘How to vote’.
The search bar also gave clues to whom people are going to vote for.
Davidowitz looked at previous election data and found out that people tend to search for both nominees whether they are going to vote for them or not.
But when searching for both nominees, the order of their names does matter.
He discovers that the nominee people write first is the one they are going to vote for, even if they don’t admit it yet.
Thus if you search for Hillary vs trump debate you most likely to vote for Hillary.
If you look for Trump vs.
Hillary, you are probably going to vote for Trump.
The results of the data leave us with a few questions.
Why are surveys aren’t accurate enough?Why do we confide google search bar with the truth?What can UX designers learn from this?How can we use this to our advantage?To answer the above-listed questions, we need to take a look at the social desirability bias.
Social desirability biasResearchers in Denver collected data on three topics from an official source in 1950.
How many people voted, how many had a library card and how many donated to charity.
The researchers then surveyed the same group about the same topics to see if their answers correlated with the data gathered.
The results were shocking.
Although the survey was anonymous, people still lied.
They inflated the amount of charity they gave, said they owned a library card (they did not) and that they voted (they didn’t).
This is just a typical example of what is termed ‘Social Desirability Bias’.
Our desire to answer questions in a way to improve our image in front of an interviewer is one of the major problems and drawbacks of surveys.
Social desirability bias will occur especially when asked about sensitive subjects.
We will exaggerate academic achievements, not admit feelings of low self-esteem, skip taking our medicine, have racist or violent thoughts, and of course, we all are very generous when it comes to charity.
In Seth Stephens-Davidowitz’s book, he quotes Roger Tourangeau who explains the problem of social desirability bias nicely “A person who looks like your favorite aunt walks in… do you want to tell your favorite aunt that you did marijuana last month?”The use of technology is a step in the right direction towards solving this problem, as it seems that the less human contact the interviewer has with the respondents, the more truthful the answers will be.
This is why internet surveys are better than phone surveys, and phone surveys are better than face-to-face interviews.
Unless the survey includes one of the sensitive topics.
People have no incentive to answer the survey truthfully on sensitive topics.
In that the case we should try and find different ways to get the information we need…Everybody liesLying is so inherent in our society that if a person never lies, they might even be seen as cruel in our world standards.
Imagine saying to someone “Oh, you got a new hairstyle!.It looks awful, you should go back to the old one”.
We lie about our health habits, family life, sex life, work, children, sometimes we even lie to ourselves.
Stephens-Davidowitz reveals that the only place that people don’t seem to lie is the Google search bar.
It seems that we tell Google everything.
He refers to the Google search bar as ‘Humanity’s truth serum’.
Have you ever encountered funny or disturbing queries in the suggestion list while searching google?.These are real queries that people have asked Google.
Although the queries we get on Google are biased by design, the Google algorithm will never suggest topics such as porn or racist comments.
It gives us a sense of how much we trust Google.
We search for our deepest secrets and fears, trusting the little white box on Google’s page for confidentiality, but everything we do online leaves a trail.
Stephens-Davidowitz follows that trail of our search habits to get insights into people’s behaviors.
The fact that people do not feel the need to censor their Google searches gives him exciting new ideas.
For his research, he uses Google Trends and we, as UX researchers and designers, should start using it too.
We should also start looking at how people communicate online.
The Online disinhibition effectOne of my guilty pleasures includes reading online comments on explosive and controversial subjects.
This is a place where common sense disappears and internet trolls thrive.
Online comments have so much power to influence minds and opinions of readers that several science pages, news forums, and celebrities have decided to turn off the abilities to comment on their posts.
On September 24, 2013, PopularScience.
com decided to remove the comment section from their website.
“Comments can be bad for science.
That’s why, here at PopularScience.
com, we’re shutting them off.
”This interesting decision from a well-known science and technology magazine was influenced by research, conducted by Professor Dominique Brossard of the University of Wisconsin-Madison.
They have asked 1,183 subjects to read a fake blog post on nano-technology and interviewed them about their perception of the topic.
They further divided them into two groups, with the first reading insulting comments on the topic, and the second reading civil comments on the topic.
The results showed that the group that read insulting comments had more polarized opinions about the subject than the other group.
This shows without a doubt that rude comments are able to change the opinions of readers.
But why are people more open to making racist, violent, or derogatory remarks online than in person?.We don’t seem to yell at each other or say racist remarks face to face.
And yet, we all probably know someone who acts rude online although they are the politest person we know.
This behavior is known as ‘Online Disinhibition’.
Wikipedia defines it as “the lack of restraint one feels when communicating online in comparison to communicating in-person.
”There are a few reasons why people feel like that online.
For example, people are more empowered when making rude comments online because of anonymity and the fact that they don’t have to deal with the receiver’s immediate reaction.
But it is also important to note that a person’s online behavior might be completely different from the true personality of a person.
Even so, we can learn a lot about a user’s needs from their online behavior.
How can we use this to our advantage?As designers, we are trying to understand users needs and create concepts to meet those needs.
Peoples online behavior can help us to create inclusive designs that will satisfy the inner needs of our users.
Where do we get the information?How do we get the information?How do we apply it to our designs?Should we stop conducting surveys?SurveysAlthough it might seem like we should stop using surveys, bias is everywhere and surveys are still a good source of information, as long as we recognize their flaws.
Surveys should be done online when possible.
The more detached people will feel from human contact, we will get more honest answers.
The survey should be completely anonymous to allow the user more freedom when making decisions.
This means respondents shouldn’t be asked to sign in or input personal details when taking such surveys.
The survey questions should be as unbiased as possible.
I am not going to get into it here since there are many articles on how to write a good survey question.
Sensitive subjects should be avoided since we already know that people lie at surveys when asked about it.
Google trendsGoogle Trends can give us useful information that can be used in our product design guidelines.
Get to know a different culture — localization.
The homepage is filtered for a specific country so we can check all the recent trending searches, which can shed more light on the local interests, concerns, or intentions.
This is very useful if we need to design for a culture that is new to us, or not living in the country we design for.
Great for brainstorming.
Each search on Google Trends also shows related topics and queries, which can further help us acquire more information about the users.
We can investigate even more by clicking on a related query and get its related topics.
Just remember to stop at some point, since it is quite addictive…Using Google trends is like having users in the room with us saying “This interests me”.
When debating which word to use when the words are synonyms, Google Trends can help us decide which word is the right one to use.
Let us say you have an E-commerce site in the UK and you wonder which to use, “deal” or “offer”.
Google trends results show that the use of each word is about 50 percent.
However, if we analyze the results by regions we get interesting information.
In Wales, “offer” is more common, but in Northern Ireland, “deal” is.
Now its a matter of where is your user base in the UK.
Social mediaBusinesses are always discussed online, regardless of if it offers B2B or B2C services.
You can read app reviews and search Twitter and other professional groups on Facebook and LinkedIn.
Don’t know what group name to look for?.use Google trends for brainstorming.
Search for your product name or topics related to your niche.
Then use the related topics that google trends suggest to find ideas for group names.
Searching thoroughly on social networks helps you to read the honest and firsthand reviews and reports on your business and competitors.
Final thoughtsAs designers, we need to constantly evolve and improve our research methods.
Investigating people’s online behavior can give us more information and get us closer to the user core motives, enabling us to design products to meet specifications.
Combining it with other design methods, including surveys, screen capture, heat maps, usability testing, a/b testing, and interviews will give us better guidance and allow us to make informed decisions during design.