I use large datasets and experimental methods to explain and investigate social psychological phenomena. My current research focuses on using Big Data analytics to investigate the role of personality traits in business-related domains.
If you want to learn more about my research, why not watch one of my talks on Youtube? Here is one on “Big Data, Psychological Profiling, and the Future of Digital Marketing“, and here my TEDx talk on “How Money Can Buy Happiness“.
What can Big Data tell us about the psychological characteristics of individuals?
Whether it is our Facebook profile, Google search queries, or credit card history, the digital footprints we leave as we go about our everyday lives create an extensive record of our personal habits and preferences. Conceptualizing these records as a window into people’s psychology, I investigate how digital footprints can be used to predict a person’s psychological profile, such as their Big Five personality (Personality predictions from online behavioral traces are as accurate as self-reports; Idani, Kohli, Kosinski, Matz, & Stillwell, in prep) or income (Predicting income from Facebook profile information; Matz, Menges, Stillwell & Schwartz). This digital form of psychometric assessment makes it possible to overcome some of the well-known limitations of self-report measures and to investigate individual differences at an unprecedented scale. While requiring participants to complete traditional self-report questionnaires is time-consuming and expensive, digital psychometrics makes it possible to predict the psychological profiles of millions of individuals in a few seconds and at virtually no cost.
What can Big Data tell us about the real-life consequences of psychological drivers?
A large proportion of psychological research is conducted in the lab using undergraduate students. Although controlled experimental designs are important in establishing causality and testing the mechanisms or boundary conditions of psychological phenomena, they do not allow us to capture actual human behavior as it plays out in the real world. While traditionally, the collection of behaviors “in the wild” was difficult and expensive, the enormous amount and diversity of digital footprints now makes it possible to inexpensively study the real-life consequences of psychological drivers over time and at extremely large scale. In my work I therefore combine existing experimental and survey designs with Big Data analytics to learn more about how individual difference are related to important real-life outcomes such as consumption and financial well-being.
For example, in my work on “The financial consequences of kindness: Why agreeableness is a trait the poor cannot afford” (Matz & Gladstone, in prep) I combined nationally representatives survey data, objective bank transaction data, and a large-scale geographic analysis of insolvency rates across the UK to provide converging evidence for the proposition that low-income agreeable individuals finish last when it comes to important financial outcomes such as savings, debt and default rates. A follow up study suggests that this effect can be explained by the tendency of highly agreeable people to place less value on money than their disagreeable counterparts.
Similarly, my work on “Money Buys Happiness When Spending Fits Our Personality” (Matz, Gladstone & Stillwell, 2016) combined a large-scale field study with an experimental design to challenge the well-established assumption that money cannot buy happiness. Using over 76,000 transaction records from a large international bank, I showed that people spend more money on products that match their personality (personality fit), and that those who do so to a greater extent report higher levels of life satisfaction. Indeed, personality fit was found to be a stronger predictor of life satisfaction than both the total amount of money people spent in a year and their personal income. A follow-up field experiment where participants were randomly assigned to two conditions, and given money to spend it on products that matched or mismatched their personality showed that the effect is causal: Personality-matched spending increased happiness. The publication was named as one of the top 10 research articles published in 2016 by Imotions and has received worldwide media attention from outlets such as the Washington Post, the Chicago Tribune, CNBC, the BBC, and the World Economic Forum.
To boost the acceptance and popularity of data-driven approaches among social scientists, I have published my insights and experiences of working at the intersection of Big Data and social science research. My papers “Facebook as a Research Tool in the Social Sciences” (Kosinski, Matz, Gosling, Popov & Stillwell, 2015), and “Using Big Data to Understand Consumers” (Matz & Netzer, invited manuscript for Current Opinions in the Social and Behavioral Sciences) outline important opportunities and challenges of using Big Data and computational methods in social sciences research. To facilitate the actual adaptation of such methods by social scientists, both papers provide practical guidelines on how to successfully integrate Big Data into one’s existing research practices.
How can Big Data help individuals and businesses make better decisions?
My findings from Research Stream 2 (“What can Big Data tell us about the real-life consequences of psychological drivers?”) provide insights into how we can support individuals and businesses to make better decisions. Knowing, for example, that highly agreeable people are at risk of making financial decisions that go against their best long-term interest could be used to create interventions that help them avoid negative financial outcomes such as debt or defaulting. Similarly, knowing that people are more satisfied with their lives when spending money in a way that is in line with their psychological could be used to create recommendation engines that prioritize and suggest those products that are most likely to contribute to an individual’s well-being. To implement such interventions in the real life, however, one needs predictive algorithms like the ones developed in Research Stream 1 (What can Big Data tell us about the psychological characteristics of individuals?). In fact, while researchers can ask participants to complete a questionnaire in the lab, it is not feasible to do so before recommending the products with the best fit online. The combination of insights from Research Streams 1 and 2, therefore makes it possible for the first time to translate psychological insights into automatic and practical tools aimed at help individuals and businesses to make better decisions.
In my work on “Psychographic Targeting as an Effective Approach to Persuasive Communication” (Matz, Nave, Stillwell & Kosinski, preparing resubmission), for example, I explore the practical value of psychologically customized communication. While laboratory studies have shown that matching products and marketing messages to people’s dominant personality trait leads to more favorable attitudes and higher purchase intentions, the validation and implementation of these mechanisms in the real-world was largely hindered by the questionnaire-based nature of personality assessment. By inferring the personality of user segments on Facebook from their Likes and subsequently targeting them through the Facebook advertising platform, I was able to demonstrate the effectiveness of personality targeting in two real advertising campaigns (worth more than $15,000): click-through and conversion rates were up to 15 times higher when consumer audiences were targeted to match the extroversion level of (1) the advertised product or (2) that of the marketing message used to promote the product.
If you want to learn more please visit my Publication page.