A study by a leading UK University has shown that age, religion,
political views, ethnicity and even sexuality can be accurately
predicted by a computer program with just a list of Facebook "Likes."
In the study, "Private traits and attributes are predictable from digital records of human behaviour", researcher Michal Kosinski demonstrates how 'Likes', the digital sign of affiliation or 'liking' of status updates, photos, websites or products can reveal much of a user's identity. Use of illegal drugs, alcohol habits and even the relationship status of a user's parents can be inferred by Likes alone.
Michal Kosinski developed Facebook App, "myPersonality" to study the relationship between Likes, demographics and personal attributes. 58,000 US Facebook users volunteered to take part in the survey providing a detailed demography, results of psychometric tests along with access to their Facebook Likes. In return, volunteers received a free personality analysis.
The data from the 58,000 Likes and corresponding psychometric tests was processed through a model to predict individual psychological and demographic profiles based on Likes alone. Researchers' accepting Likes as firm indicators only when 100 users or more had demonstrated a correlating link.
Some Likes indicators were blunt, others less so. Christians were identified by "Jesus Daily" but some more tenuous links were uncovered. Facebook users with "I like lyrics that actually mean something" in their Like list were shown to be statistically more likely to be prone to substance misuse.
The programme proved surprisingly accurate in differentiating between non-variable attributes such as ethnicity, sexuality, politics and religion. The success rate in non-variable characteristics, predicted by Likes alone were;
- African, American and Caucasian Americans 95%
- Sexuality 88%
- Democrats and Republicans in 82%
- Christian and Muslims 82%
Although significantly lower, "good prediction accuracy" was still achieved in the detection of variable aspects
such as the relationship status of a user. The lower detection rate was, according to researchers, explained by a user's tendency to move between the categories of 'in a relationship' to 'out'; the unstable and changeable nature of human relationships proving difficult for a programme to compute. Difficult though not impossible. The relationships status of Facebook users was accurately and automatically predicted by Likes alone in 65% of cases.
Noted as "remarkable" by researchers was the ability of the program to accurately predict in 60% of cases whether an account holder's parents had separated before the user had reached the age of 21. The indication of divorced or separated parents was predicted by a user's increased probability liking relationships and emotionally laden lyrics in their Like list, for example, "I'm Sorry I Love You" or "If I am with you then I'm with you I don't want anybody else".
Unusual associations grouped users - intelligent users were seen to group together in their liking of curly fries and thunderstorms, females in relationships were identified by a common liking in weightwatchers and scapbooking and emotionally stable users outed themselves with likes for skydiving and business administration. Swimming appeared to be a deal breaker; those who had swimming in their Likes scored higher in the most Satisfied With Life (SWL) score whereas drug users were least likely of all to give swimming the digital thumbs up.
As businesses clamour for more information on customers, Facebook users will find that predictive programmes are honed to perfection. The risk of people being vulnerable to giving away more information than they have chosen to give freely, and of this information being misused is noted by Michal Kosinski n the Proceedings of the National Academy of Sciences (PNAS), "commercial companies, governmental institutions, or even one's Facebook friends could use software to infer attributes such as intelligence, sexual orientation, or political views that an individual may not have intended to share. One can imagine situations in which such predictions, even if incorrect, could pose a threat to an individual's well-being, freedom, or even life."
Researchers point out that as users become more aware of their vulnerability to information disclosure, they may lose trust in online services and be deterred from using digital technology. Young people are voting already with their digital feet leaving Facebook in preference for other social networking platforms; Twitter, Pinterest and Instagram which offer the anonymity that Facebook lacks. According to "Inside Facebook" 6 million American users left Facebook in March 2013 alone.
Michal Kosinski and his research team at the Psychometrics team remain optimistic for the future, "It is our hope, however, that the trust and goodwill among parties interacting in the digital environment can be maintained by providing users with transparency and control over their information, leading to an individually controlled balance between the promises and perils of the Digital Age."
Notes:
Proceedings of the National Academy of Sciences (PNAS) http://www.pnas.org/content/110/15/5802.full
Your One Click Personality
http://www.youarewhatyoulike.com
In the study, "Private traits and attributes are predictable from digital records of human behaviour", researcher Michal Kosinski demonstrates how 'Likes', the digital sign of affiliation or 'liking' of status updates, photos, websites or products can reveal much of a user's identity. Use of illegal drugs, alcohol habits and even the relationship status of a user's parents can be inferred by Likes alone.
Michal Kosinski developed Facebook App, "myPersonality" to study the relationship between Likes, demographics and personal attributes. 58,000 US Facebook users volunteered to take part in the survey providing a detailed demography, results of psychometric tests along with access to their Facebook Likes. In return, volunteers received a free personality analysis.
The data from the 58,000 Likes and corresponding psychometric tests was processed through a model to predict individual psychological and demographic profiles based on Likes alone. Researchers' accepting Likes as firm indicators only when 100 users or more had demonstrated a correlating link.
Some Likes indicators were blunt, others less so. Christians were identified by "Jesus Daily" but some more tenuous links were uncovered. Facebook users with "I like lyrics that actually mean something" in their Like list were shown to be statistically more likely to be prone to substance misuse.
The programme proved surprisingly accurate in differentiating between non-variable attributes such as ethnicity, sexuality, politics and religion. The success rate in non-variable characteristics, predicted by Likes alone were;
- African, American and Caucasian Americans 95%
- Sexuality 88%
- Democrats and Republicans in 82%
- Christian and Muslims 82%
Although significantly lower, "good prediction accuracy" was still achieved in the detection of variable aspects
such as the relationship status of a user. The lower detection rate was, according to researchers, explained by a user's tendency to move between the categories of 'in a relationship' to 'out'; the unstable and changeable nature of human relationships proving difficult for a programme to compute. Difficult though not impossible. The relationships status of Facebook users was accurately and automatically predicted by Likes alone in 65% of cases.
Noted as "remarkable" by researchers was the ability of the program to accurately predict in 60% of cases whether an account holder's parents had separated before the user had reached the age of 21. The indication of divorced or separated parents was predicted by a user's increased probability liking relationships and emotionally laden lyrics in their Like list, for example, "I'm Sorry I Love You" or "If I am with you then I'm with you I don't want anybody else".
Unusual associations grouped users - intelligent users were seen to group together in their liking of curly fries and thunderstorms, females in relationships were identified by a common liking in weightwatchers and scapbooking and emotionally stable users outed themselves with likes for skydiving and business administration. Swimming appeared to be a deal breaker; those who had swimming in their Likes scored higher in the most Satisfied With Life (SWL) score whereas drug users were least likely of all to give swimming the digital thumbs up.
As businesses clamour for more information on customers, Facebook users will find that predictive programmes are honed to perfection. The risk of people being vulnerable to giving away more information than they have chosen to give freely, and of this information being misused is noted by Michal Kosinski n the Proceedings of the National Academy of Sciences (PNAS), "commercial companies, governmental institutions, or even one's Facebook friends could use software to infer attributes such as intelligence, sexual orientation, or political views that an individual may not have intended to share. One can imagine situations in which such predictions, even if incorrect, could pose a threat to an individual's well-being, freedom, or even life."
Researchers point out that as users become more aware of their vulnerability to information disclosure, they may lose trust in online services and be deterred from using digital technology. Young people are voting already with their digital feet leaving Facebook in preference for other social networking platforms; Twitter, Pinterest and Instagram which offer the anonymity that Facebook lacks. According to "Inside Facebook" 6 million American users left Facebook in March 2013 alone.
Michal Kosinski and his research team at the Psychometrics team remain optimistic for the future, "It is our hope, however, that the trust and goodwill among parties interacting in the digital environment can be maintained by providing users with transparency and control over their information, leading to an individually controlled balance between the promises and perils of the Digital Age."
Notes:
Proceedings of the National Academy of Sciences (PNAS) http://www.pnas.org/content/110/15/5802.full
Your One Click Personality
http://www.youarewhatyoulike.com
@burrblog
No comments:
Post a Comment