PSYCHOMETRIC TARGETING

Aim & Shoot

Exploring how digital presence is used to detect people’s personal traits & vulnerabilities and how these can potentially be exploited to undermine democracy.

By Claire Daniolou

January 10, 2021

What is psychometric targeting?

Psychometry is a technique through which one can evaluate a person’s beliefs and personality. The word “Psychometry” derives from “psyche” and “metric”. The former stands for an individual’s mind and ideology and the latter refers to a system of measurements. This is a technique used in psychology, health, and other similar fields to provide robust measures for reaching conclusions.

Modern companies use the science of psychometry to create profiles on certain groups of people based not on their demographic data, but the actual underlying traits that dictate their behaviour. This way their marketing strategies are personalised to attract larger audiences more efficiently based on the predictions they make of the subject’s reaction to the content. Since targeting is such an effective marketing technique, political offices have also recognised it and used to promote their agenda. Therefore, from the data retrieved from one’s online activities, persuasive, targeted ads are used to interfere with personal ideas and even manipulate one’s vote. 
 

How does it operate?

“Are you a student? 100£ of Amazon vouchers available for 10 minutes of your time. Tell us about your future career in this survey!” Similarly titled emails often appear in my university mailbox. “What is the harm?”, I have always thought, “I may even buy my sister a nice Christmas gift!”  Unfortunately, that is not exactly the case.

Psychometric targeting mechanisms use basic user digital footprint to develop focused strategies. The data retrieved are sometimes directly acquired from surveys and then used to create user profiles. According to Christopher Wylie, a former employee at Cambridge Analytica, these surveys would commonly be completed by people lured by a small financial incentive or by ones who are, simply, bored. He describes that the latter category mostly refers to “wealthy white women”. However, to examine how data collection is performed on a larger scale can be quite interesting. As explained in his interview, while these people log in the surveys to complete them, data in the form of “likes”, “shares” etc. is retrieved from the surveyed user’s social media profile, as well as from all their virtual friends. Psychologists David Stillwell and Michal Kosinski confirm that “likes” can be used to accurately predict personal traits.

To demonstrate the elementary character of these predictions, here is a simplified visual example:

Imagine person X would like to persuade their colleague, person Y, to reach a certain decision (that is, the star in the GIF above).  X may not know Y personally, but X takes a brief look on Y’s desk. X sees an ice-blue jacket on Y’s desk chair, a dark blue ocean on Y’s screensaver, and a turquoise frame with Y’s partner smiling under a teal blanket on his desk. Which colour should X coat the proposal with? The blue one could unconsciously make Y more prone to like the idea.

Is it really effective?

One of the most widely used psychometric models in social media is called OCEAN as its measures and scales that following traits: Openness, Contentiousness, Extraversion, Agreeableness, and Neuroticism. Psychologists believe that based on these traits, one can predict with high accuracy a subject’s decision-making process. These traits comprise basic pillars through which a person’s personality, hence their way of thinking, can be described. 

The National Academy of Science published an article entitled “Psychological targeting in digital mass persuasion” that concludes psychometric targeted messages resulted in 40% more clicks and 50% more purchases than mismatching or non-personalised messages.

Such results back up the logical assumptions that expect a person to be more prone to respond to a message that appeals to the person’s ideas. Additionally, political ads and speeches use persuasive language to shape audience perceptions. Therefore, if the political ads speak the same "language" as their audience, they would be even more effective. After the first presidential debate of the 2020 U.S. elections, ads from the Trump campaign accused Joe Biden to have been wearing an earpiece to be provided help from his team during the broadcast. The used phrasing “[Biden] DECLINED an earpiece inspection” is indubitably more influential than “[The candidates] did not undergo extensive scrutiny”.

What risks does it pose relating to democracy?

In Netflix’s documentary The Great Hack, it is explained that since digital footprint can predict human behaviour, it can also predict one’s political preferences. One example that illustrates this is the Cambridge Analytica scandal of the 2016 U.S. elections. The company identified the segments of society that were predicted to be undecided with respect to their vote. Afterwards, through the methods described above, Cambridge Analytica micro-targeted this specific audience that it could effectively persuade.

The problem occurs when the message one conveys is biased. If the social media platforms themselves are the ones that decide who has access to their data, they control this bias. Therefore, should social media platforms be partisans, they offer very powerful tools unevenly to the parties involved. That is not necessarily impactful if implemented on a small scale. For private individuals, it is only natural to support the political forces that represent their beliefs and promote their opinions in their social circle. But when it comes to social media corporations, such influence is extended to the whole of the digital population. The effect this has on a global level is imperative for future international socioeconomic situations and should not be freely monitored by a small number of business executives.

What can be done?

At this point, it is important to understand that social media and the recording of user data are inherently linked. The only way these platforms exist is by using these data to suggest, promote, and display relevant content to support the user in their endeavour to connect with the world. Advertisement is also the core source of income of the platforms. So, for social media to be used, data will always be recorded. However, the stakes are high when considering the effect the above may have on imperative issues, such as elections. For if the results of the elections were affected by non-democratic influences, even global health would depend on a handful of irrelevant to health tech companies. Therefore, it is imperative to monitor the power such firms have over election outcomes and embrace the fundamental values of a democratic society. Otherwise, with a cap on freedom of thought, democracy ceases to exist. Consequently, more control is needed to guarantee ethical and safe data usage.

The suggested strategies involve three aspects. Firstly, legislation should hold companies accountable for online harms due to targeting. Some analyses suggest that targeting and misinformation may have even promoted the Rohingya genocide. Secondly, transparency is an absolute necessity. The technologies might not be understood by all, but they affect all. Therefore, data should have a more public nature; what is stored, how it used and protected, and for how long should be readily answerable questions. Lastly, user empowerment is a key attribute. More information and control over how users are targeted should be provided. This should be subject to personal preferences so that people can be able to own their data.

Taking a step back, it is quite shocking to see how people’s “likes” can be used against themselves; the Facebook team that created it perceived it as a way to spread positivity, as is discussed in Netflix’s documentary The Social Dilemma. There is a need for the button to serve its initial cause.