Available online at www.sciencedirect.

com

ScienceDirect

Using Big Data as a window into consumers’ psychology

Sandra C Matz1 and Oded Netzer2

The rise of ‘Big Data’ had a big impact on marketing research Applications of Big Data in marketing have largely
and practice. In this article, we first highlight sources of useful focused on (a) assessing customers’ preferences [e.g.,3],
consumer information that are now available at large scale and (b) predicting what customers are most likely to buy next
very little or no cost. We subsequently discuss how this [e.g.,4–6], (c) improving targeted advertising [e.g.,7,8],
information – with the help of new analytical techniques – can (d) understanding brand perceptions [e.g.,9,10], and (e)
be translated into valuable insights on consumers’ describing the competitive landscape [e.g.,11]. See Wedel
psychological states and traits that can, in turn, be used to and Kannan [1] for a review. However, investigations of
inform marketing strategy. Finally, we discuss opportunities how Big Data can help inform some of the more psycho-
and challenges related to the use of Big Data as a window into logical aspects of consumer behavior that is aimed at
consumers’ psychology, and provide recommendations for understanding – rather than merely predicting – con-
how to implement related technologies in a way that benefits sumer attitudes and emotions has thus far only received
both businesses and consumers. scant attention. Davenport et al. [12] (2001, p. 63) note
that holding vast amounts of customer data might help
businesses to ‘know more about their customers’ but does
Addresses
not necessarily allow them to ‘know the customers
1
University of Cambridge, Department of Psychology, Downing Site, themselves’. The focus of this paper is to highlight the
Cambridge CB2 3EB, United Kingdom existing work and discuss the potential of using Big Data
2
Columbia Business School, Columbia University, New York, as a means to better understand consumers’ stable psy-
United States
chological traits as well as more malleable psychological
Corresponding author: Matz, Sandra C (sm917@cam.ac.uk) states.

New sources of consumer information

Current Opinion in Behavioral Sciences 2017, 18:7–12 Traditional approaches to gathering ‘human-centric’ con-
This review comes from a themed issue on Big data in the sumer information include extensive customer surveys,
behavioural sciences focus groups, interviews, observation studies and limited
Edited by Michal Kosinski and Tara Behrend scope secondary data such as scanner panel data [1]. For
example, as part of the Nordstrom’s Personal Touch
program, personal shoppers recorded detailed informa-
tion on customers likes and dislikes, their lifestyle and
tastes through telephone and face-to-face conversations
http://dx.doi.org/10.1016/j.cobeha.2017.05.009 as well as observations made in the store [12]. While the
2352-1546/ã 2017 Elsevier Ltd. All rights reserved. outlined approaches can generate valuable customer
knowledge, they are not only expensive and time-con-
suming – and therefore difficult to scale – but also prone
to numerous well established response biases [13]. For
example, even the most motivated customer will find it
difficult to accurately recall the purchases they made over
The availability of data at large volume, variety, velocity the past four weeks or the exact feeling they experienced
and veracity, often termed as ‘Big Data’, had a big impact when purchasing a specific product.
on marketing research [1] and practice [2]. The wealth
of personal information available about consumers online Thanks to technological advances in the collection, stor-
makes it possible to understand and cater to the individ- age and analysis of large amounts of data, businesses can
ual needs of consumers better than ever before. Whether now gain valid insights on millions of consumers by
it is their Spotify playlists, Facebook profile, Google looking at the digital records that are passively collected
search queries, or mobile location, the digital footprints as consumers go about their daily lives. In fact, observing
consumers leave with every step they take in the digital the behavior of a consumer in a traditional retail store is
environment create extensive records of their personal very similar to analyzing the journey of a customer who is
habits and preferences. By tapping into this rich pool of browsing a company’s online store (e.g., one can examine
consumer data, businesses can enhance consumers’ expe- the characteristics of products the user has looked at and/
rience by better matching the marketing offering to or bought, measure the time they took to make a decision,
consumers’ preferences and do so at the appropriate or implement mouse-tracking technologies to study the
moment. decision process). Similarly, customer forums, product

www.sciencedirect.com Current Opinion in Behavioral Sciences 2017, 18:7–12

8 Big data in the behavioural sciences

reviews and posts in social media make it possible to behavioral responses to products, brands or marketing
observe large and natural ‘focus groups’ at very little to no messages that match their own psychological traits [e.
cost [11]. g.,17–20]. For example, an extroverted and open-minded
consumer might experience more positive emotions and
The sources of information businesses can tap into to report a higher intention towards a retail brand that
learn more about their consumers are almost limitless, and specializes in flashy and unusual clothes, or that uses
it would go beyond the scope of this paper to discuss all of extroverted and creative language to advertise their pro-
them in detail (for an overview see Wedel and Kannan ducts (e.g., ‘Stand out from the crowd and feel unique with
[1], Fig. 2). Among the most vital ones are historical our latest spring collection’). Businesses have long used
purchasing data, credit card records, search queries, such insights for branding and advertising purposes [e.
browsing histories, blog posts, social media profiles, and g.,21].
smartphone sensor data (e.g., GPS location). Importantly,
it is often possible to combine the information extracted However, because unlike demographics and past pur-
from different sources to form a more holistic picture of a chases, latent psychological traits cannot be observed
consumer’s daily habits and preferences. By integrating directly, the opportunities to target consumers and per-
information obtained from a consumer’s social media sonalize advertising based on psychological traits have
profile, their phone logs and sensor data as well as their been limited. If a mobile phone provider, for instance,
credit card spending, for example, one can get a fairly decided to create a strong extroverted brand, it was very
accurate picture of what a consumer has done when and difficult to focus its advertising efforts on extroverted
with whom. consumers short of choosing media channels (e.g., TV
shows) that are predicted based on questionnaires or
These new sources of data not only come from various managerial judgement to have a larger proportion of
sources, but they also come in multiple formats. While extroverts. Instead, the branded marketing message
traditional data have been primarily structured in a had been primarily focused on mass marketing, broad-
numeric format, social media data, are primarily unstruc- casting to large and heterogeneous audiences, thereby
tured including, text, images, audio and video. Accord- limiting its effectiveness.
ingly, different analytical approaches are needed to con-
vert such data into knowledge and insights. In the age of Big Data, however, psychological traits –
including personality, IQ and political orientation – can
Turning Big Data into human-centric be accurately predicted from consumers’ digital foot-
customer knowledge prints. Researchers have demonstrated the ability to
The task of turning vast amounts of – often unstructured accurately infer personal traits from (a) personal websites
– data into insightful consumer knowledge is not easy and [22], (b) Facebook or Twitter profiles [23,24,25], (c)
often requires the application of analytical techniques blogs [26], and (d) language use [27,28,29,30]. This
that are outside of the standard methodological tool box of digital form of psychometric assessment promises to be a
consumer behavior researchers [14]. However, recent game changer in the application and empirical evaluation
years have seen the rise of so-called computational social of psychographic marketing. In an early pioneering study,
science research, a discipline aimed at applying method- for example, Hauser and colleagues inferred cognitive
ologies from the computer sciences to questions asked by styles (e.g., analytic vs. emotional) from clickstream data
social scientists [15]. While the range of possible applica- and showed that matching a website’s ‘look and feel’ to
tions of such methodologies to social science questions is consumers’ dominant motivational orientation can
bounded only by the creativity and imagination of the increase sales by up to 20% [7]. Similarly, Matz and
researcher, here we focus on two types of insights that colleagues showed that inferring the personality of Face-
have recently attracted a considerable amount of atten- book users from their Likes, and matching the content of
tion among researchers and practitioners alike: the pre- real advertising campaigns (products and marketing mes-
diction of (1) relatively stable psychological traits that help sages) to their dominant personality traits can significantly
explain consumers’ general tendency to think, feel and increase click-through and conversion rates [31]. As the
behave in a certain way, and (2) malleable psychological digital assessment of psychological traits becomes more
states that express consumers’ attitudes and emotions in- widespread and readily available (e.g., LIWC for comput-
the-moment and help to put their behavior in context. erized text analysis; ApplyMagicSauce and StatSocial for
personality predictions), consumer behavior scholars will
Predicting consumers’ psychological traits be able to build on this early research and test the
The investigation of stable psychological traits such as effectiveness of psychographic targeting in different
personality, regulatory focus, or need for cognition, has a domains (e.g., retail, charitable giving, political campaign-
long-standing tradition in consumer behavior research ing) and channels (e.g., social media, email, in-store),
[16]. One of the most consistent findings suggests that using different psychological traits (e.g., personality, cog-
consumers show more positive cognitive, emotional and nitive style, motivational orientations), and different

Current Opinion in Behavioral Sciences 2017, 18:7–12 www.sciencedirect.com

 Using Big Data as a window into consumers’ psychology Matz and Netzer 9

outcome measures (e.g. clicks, purchases, long-term can use information on consumers’ mood or emotions
retention). to increase the relevance of the content that is suggested
to the user [47]. Such context-aware recommenders, that
Turning customer data into meaningful psychological take into account consumers’ emotions, have shown
profiles offers tremendous opportunities for a more holis- improved recommendations for music [48], movies
tic Customer Relations Management [CRM; ,32] that [49,50], and images [51].
bridges the gap between online and offline channels.
For example, knowing that a consumer follows a cognitive Combining psychological traits and states
style that is analytical rather than emotional, makes it The combination of psychological traits (variability across
possible for both computers online and salespeople in consumers) and psychological states (variability within
brick and mortar stores to adapt their communication to consumers over time) offers an unprecedented under-
the preferences of the customer. standing of consumers’ unique needs as they relate to the
situation-specific expressions of more stable motivations
Predicting consumers’ psychological states and preferences [33; also compare to the theory of free
As we have outlined, psychological traits play an impor- traits, 52]. For example, extroverted consumers might be
tant role in understanding and predicting consumer more likely to respond to personality-matched advertise-
behavior. However, marketing researchers have long rec- ments [e.g.,19] when they are in an extroverted situation
ognized that they cannot account for the full variation in that highlights and reinforces their extroverted innate
consumer behavior [33]. This is, because psychological nature or when they find themselves in an introverted
traits do not operate in a vacuum, but instead are situation that lacks the excitement and stimulation they
expressed in a certain context, these traits are often need to thrive. The availability of data and analysis tools
influenced by situational factors [34,35]. For example, to investigate personality traits and states in real time,
consumers who are in a positive mood use more heuristic provide a fruitful avenue to exploring the interesting
– rather than systematic – information processing and interactions between personality traits and states and
evaluate products and brands more favorably [for an how consumers may react to offer that leverage such
overview on the effect of mood on consumer behaviour interactions.
see Ref. [36]]. Hence marketers can benefit from paying
close attention to and capitalize on customers’ psycho- Figure 1 summarizes the outlined opportunities of using
logical states. Big Data in the context of consumer research. As we have
discussed throughout the paper, the wealth of personal
However, because of the transient nature of psychological consumer information available at little to no cost makes
states identifying such states in real time is even more it possible to not only predict consumer outcomes, but to
challenging than identifying psychological traits. Similar also understand consumers’ psychological needs and
to psychological traits, psychological states have tradition- motivations at both the state and trait levels. Understand-
ally been tied to questionnaire measures [e.g., the PANAS ing consumers’ psychological states and traits can then be
scale for positive and negative affect; [37]]. However, used to better match the firm’s marketing offerings to
these have been mainly performed for academic purposes customers’ needs and preferences, and hence improve
as the ability of firms to measure and act in real time on business and consumer outcomes.
varying psychological states using surveys is largely
impractical. Fortunately, new data sources and advances Opportunities and challenges
of analytics techniques make psychological traits predict- The combination of information about ‘what one does’
able from a broad variety of digital footprints collected in with deeper understanding of ‘who one is’ offers tremen-
real time [see e.g.,38,39]. Consumers’ mood and emotions dous opportunities to not only boost the effectiveness of
have been successfully predicted from spoken and writ- marketing campaigns but also to help consumers make
ten language [40], video [41], wearable devices [42], better decisions. The pre-selection of content that is in
smartphone sensor data [43], and even information line with consumers’ psychological needs can alleviate
obtained from the environment such as weather or physi- the problem of choice overload [53,54] and help consu-
cal location [44]. mers to maximize the satisfaction and happiness they gain
from their choices [55]. In addition, psychologically-cus-
While marketers have long used retrospective analyses of tomized health messages are known to be effective in
consumer sentiment in the study of online word-of- changing behaviors among patients and groups who are at
mouth [45,46], the ability to assess consumers’ psycho- risk [56,57]. Targeting highly neurotic individuals who
logical states and sentiment in real time provides con- display early signs of depressions with ads that guide
sumer behavior researchers and practitioners with tre- them to self-help pages or offer professional advice, for
mendous opportunities to personalize marketing content example, could have a tremendous positive impact on the
to the immediate psychological needs of consumers. well-being of some of the more vulnerable members of
Context-aware recommendation systems, for example, society, and even save lives.

www.sciencedirect.com Current Opinion in Behavioral Sciences 2017, 18:7–12

10 Big data in the behavioural sciences

Figure 1

Psychological Traits
• Big 5 Personality Traits
• Values
• Regulatory Focus
• Cognitive Styles
• Intelligence
• Discounting
Digital Records
• Spendthrift/Tightwad Consumer Outcomes
• Purchase History • Risk Aversion • Preferences and Liking
• Browsing History • ...
• Brand Perception/
• Search Queries
Awareness
• Social Media Profiles
(e.g. Facebook, Twitter) • Purchase Behavior
• Post-Purchase Satisfaction
• Personal Blogs
• Customer Lifetime Value
• Wearable Devices (e.g. Psychological States • Customer Retention
Smartphones, Fitbits)
• Mood • Competitor analysis
• Product Reviews
• Emotions • ...
• ...
• Alertness
• Attention
• Stress
• ...
Current Opinion in Behavioral Sciences

Leveraging Big Data to infer psychological traits and states and affect customer behavior.

Alongside the benefits psychologically-personalized mar- therefore suggest to use the knowledge of consumers’
keting provides, it also raises new ethical challenges. psychological traits to provide optional services that con-
While psychological targeting can help consumers make sumers can actively opt-in to. Given that privacy concerns
better choices, it could also be used in a way that exploits are known to negatively impact the effectiveness of
‘weaknesses’ in a person’s character. For example, one personalized advertising [61], while giving consumers
could target individuals who are prone to compulsive or more control over their personal information positively
addictive behavior [58] with ads for an online casino, or affects their willingness to click on personalized ads [62],
exclude them from receiving insurance ads. In fact, Face- such an approach is not only in the interest of consumers
book was recently criticized for analyzing teenagers’ but eventually in the best self-interest of businesses. By
emotional or mental state using their Facebook profiles. implementing psychologically-personalized targeting in a
While Facebook said it does not currently use such transparent way that gives data ownership and control to
inferences for targeting, even the collection of such data consumers, businesses can avoid the risk of reputational
raised consumers’ ethical concerns. This more critical damage and instead turn psychological customization into
side of increasingly personalized marketing is reflected a desirable component of their value proposition to
in general public skepticism [59,60]. A 2010 survey of customers.
American Internet users showed that less than 20%
expressed a preference for targeted ad, while 64% viewed Conclusion
personalized advertising as ‘intrusive’ [59]. In 2012, this Taken together, the ability to predict consumers’ psy-
skepticism reached a public peak in response to a chological traits and states from their digital footprints
‘scandal’ involving the U.S. retail giant Target. Using offers exciting new opportunities for digital marketing.
data-driven recommendation algorithms, Target had pro- We expect both researchers and practitioners to go
moted baby equipment to a pregnant teenage girl in beyond the understanding and prediction of psychologi-
Minnesota, whose parents had previously been unaware cal states and traits and towards real-time ‘optimization’
of the pregnancy. With the introduction of even more of marketing actions on the basis of these predictions.
sophisticated prediction algorithms that not only analyze Much like in the scene in the science fiction movie
individual behaviors but make inferences about a con- Minority Report, where advertising billboards are person-
sumers’ intimate psychological traits and states, these alized to the emotional state of the person walking past
concerns are unlikely to change for the better. We them, businesses will be able to optimize the advertising a

Current Opinion in Behavioral Sciences 2017, 18:7–12 www.sciencedirect.com

 Using Big Data as a window into consumers’ psychology Matz and Netzer 11

consumer is exposed to in real-time and at a level of detail 12. Davenport TH, Harris JG, Kohli AK: How do they know their
customers so well? MIT Sloan Manage. Rev. 2001, 42:63.
never before possible. For example, one could use infor-
13. Podsakoff PM, Organ DW: Self-reports in organizational
mation about a person’s momentary heart rate extracted research: problems and prospects. J. Manage. 1986, 12:531-
through their headphones to determine which song to 544.
play next, extract emotions from a person’s facial expres- 14. Kosinski M, Matz SC, Gosling SD, Popov V, Stillwell D: Facebook
sion to change the color scheme of a website, or recom- as a research tool for the social sciences. Am. Psychol. 2015,
mend the next tourist attraction in a new city as a function 70:543-556.
of the person’s predicted personality and their current 15. Lazer D, Pentland AS, Adamic L, Aral S, Barabasi AL, Brewer D,
level of physical activity. We encourage researchers to Christakis N, Contractor N, Fowler J, Gutmann M: Life in the
network: the coming age of computational social science. Sci.
continue to explore these exciting opportunities. (New York, NY) 2009, 323:721.
16. Kotler P, Armstrong G: Principles of Marketing. pearson education;
2010.
Acknowledgement
17. Aaker JL: The malleable self: the role of self expression in
Sandra C Matz was funded by the German National Merit Foundation. persuasion. J. Mark. Res. 1999, 36:45-57.
18. Sirgy MJ: Using self-congruity and ideal congruity to predict
References and recommended reading purchase motivation. J. Bus. Res. 1985, 13:195-206.
Papers of particular interest, published within the period of review, 19. Hirsh JB, Kang SK, Bodenhausen GV: Personalized persuasion:
have been highlighted as: tailoring persuasive appeals to recipients’ personality traits.
Psychol. Sci. 2012, 23:578-581.
 of special interest
 of outstanding interest 20. Wheeler SC, Petty RE, Bizer GY: Self-schema matching and
attitude change: situational and dispositional determinants of
1. Wedel M, Kannan PK: Marketing analytics for data-rich message elaboration. J. Consum. Res. 2005, 31:787-797.
 environments. J. Mark. 2016, 80:97-121.
Wedel and Kannan offer a critical review of marketing analytics methods. 21. Aaker JL: Dimensions of brand personality. J. Mark. Res. 1997,
Examining different types of data (e.g. structured vs. unstructured, firm 34:347-356.
internal vs. external), they review the potential of these methods and
22. Marcus B, Machilek F, Schütz A: Personality in cyberspace:
subsequently highlight current and future directions for new analytical
personal Web sites as media for personality expressions and
tools with regard to (a) optimized marketing-mix spending, (b) persona-
impressions. J. Pers. Soc. Psychol. 2006, 90:1014-1031.
lization, and (c) data security.
2. van den Driest F, Sthanunathan S, Weed K: Building an insights 23. Quercia D, Kosinski M, Stillwell D, Crowcroft J: Our Twitter
 engine. Harvard Bus. Rev. 2016, 94:15. profiles, our selves: predicting personality with Twitter. IEEE
Van den Driest et al. (2016) provide a practitioner-oriented discussion on International Conference on Social Computing 2011, 18:180-185.
how to successfully implement customer-centricity as a powerful source 24. Kosinski M, Stillwell D, Graepel T: Private traits and attributes
of competitive advantage. Focusing on the retail giant Unilever, they  are predictable from digital records of human behavior. Proc.
outline operational and people characteristics that facilitate the devel- Natl. Acad. Sci. 2013, 110:5802-5805.
opment and application of highly functional “insight engines” on the basis Kosinski et al. demonstrate the validity of preference-based personality
of Big Data. assessment via social media. Using data of more than 58 thousand
3. Jacobs BJD, Donkers B, Fok D: Model-based purchase Facebook users, they show that highly intimate demographic and
predictions for large assortments. Mark. Sci. 2016, 35:389-404. socio-psychological characteristics (e.g. political orientation, IQ or per-
sonality) can be accurately predicted from people’s Facebook Likes.
4. Ghose A, Ipeirotis PG, Li B: Designing ranking systems for
hotels on travel search engines by mining user-generated and 25. Youyou W, Kosinski M, Stillwell D: Computer-based personality
crowdsourced content. Mark. Sci. 2012, 31:493-520. judgments are more accurate than those made by humans.
Proc. Natl. Acad. Sci. U. S. A. 2014, 112:1036-1040.
5. Lu S, Xiao L, Ding M: A video-based automated recommender
(VAR) system for garments. Mark. Sci. 2016, 35:484-510. 26. Yarkoni T: Personality in 100,000 words: a large-scale analysis
of personality and word use among bloggers. J. Res. Pers.
6. Linden G, Smith B, York J: Amazon.com recommendations: 2010, 44:363-373.
item-to-item collaborative filtering. IEEE Comput. Soc. 2003,
7:76-80. 27. Netzer O, Lemaire A, Hertzenstien M: When words sweat:
 Identifying signals for loan default in the text of loan
7. Hauser JR, Urban GL, Liberali G, Braun M: Website morphing. applications. Working Paper
 Mark. Sci. 2009, 28:202-223. Netzer et al. use text-mining and machine-learning tools to automatically
Hauser et al. demonstrate the value of dynamic website morphing in a process and analyze the raw text in thousands of loan requests from an
large-scale experiment using data from 835 BT Group (former British online crowdfunding platform. They find that borrowers, consciously or
Telecom) customers. They infer cognitive styles (e.g. analytic vs. emo- not, leave traces of their intentions, circumstances, and personality traits
tional) from clickstream data and show that matching a website’s “look in the text they write when applying for a loan.
and feel” to consumers’ dominant motivational orientation can increase
sales by up to 20%. 28. Park G, Schwartz H, Eichstaedt J, Kern ML, Kosinski M, Stillwell D,
 Ungar LH, Seligman MEP: Automatic personality assessment
8. Trusov M, Ma L, Jamal Z: Crumbs of the cookie: user profiling in through social media language. J. Pers. Soc. Psychol. 2014,
customer-base analysis and behavioral targeting. Mark. Sci. 108:934-952.
2016, 35:405-426. Park et al. demonstrate the validity of language-based personality
9. Culotta A, Cutler J: Mining brand perceptions from twitter assessment via social media. Using an open-vocabulary analysis of
social networks. Mark. Sci. 2016, 35:343-362. the status updates of more than 70 thousand Facebook users, they show
that language-based predictions of personality (a) converge with self and
10. Tirunillai S, Tellis GJ: Mining marketing meaning from online other reports, (b) accurately discriminate between traits (c) are stable over
chatter: strategic brand analysis of big data using latent time and (d) exhibit correlations with external criteria (e.g. life satisfaction)
dirichlet allocation. J. Mark. Res. 2014, 51:463-479. similar to those found for self-reports.
11. Netzer O, Feldman R, Goldenberg J, Fresko M: Mine your own 29. Schwartz HA, Eichstaedt JC, Kern ML, Dziurzynski L,
business: market-structure surveillance through text mining. Ramones SM, Agrawal M, Shah A, Kosinski M, Stillwell D,
Mark. Sci. 2012, 31:521-543. Seligman MEP et al.: Personality, gender, and age in the

www.sciencedirect.com Current Opinion in Behavioral Sciences 2017, 18:7–12

12 Big data in the behavioural sciences

language of social media: the open-vocabulary approach. Utility Theory: In International Conference on Fuzzy Systems and
PLoS One 2013, 8:e73791. Knowledge Discovery. Springer; 2006:970-979.
30. Pennebaker JW, King LA: Linguistic styles: language use as an 45. Das SR, Chen MY: Yahoo! for Amazon: sentiment extraction
individual difference. J. Pers. Soc. Psychol. 1999, 77:1296. from small talk on the web. Manage. Sci. 2007, 53:1375-1388.
31. Matz S, Kosinski M, Nave G, Stillwell D: Personality-matching 46. Jansen BJ, Zhang M, Sobel K, Chowdury A: Twitter power:
 increases the effectiveness of digital advertising. Working tweets as electronic word of mouth. J. Am. Soc. Inf. Sci.
Paper Technol. 2009, 60:2169-2188.
Matz et al. demonstrate the effectiveness of personality-based targeting
in several large-scale experiments on Facebook. They show that inferring 47. Zheng Y, Mobasher B, Burke R: Emotions in Context-aware
the personality of Facebook users from their Likes, and matching the Recommender Systems: In Emotions and Personality in
content of real advertising campaigns (products and marketing mes- Personalized Services. Springer; 2016:311-326.
sages) to their dominant personality traits can significantly increase click- 48. Lee JS, Lee JC: Music for my mood: a music recommendation
through and conversion rates.
system based on context reasoning. European Conference on
32. Odekerken-Schröder G, De Wulf K, Schumacher P: Smart Sensing and Context. Springer; 2006:190-203.
Strengthening outcomes of retailer–consumer relationships: 49. Winoto P, Tang TY: The role of user mood in movie
the dual impact of relationship marketing tactics and recommendations. Expert Syst. Appl. 2010, 37:6086-6092.
consumer personality. J. Bus. Res. 2003, 56:177-190.
50. Shi Y, Larson M, Hanjalic A: Mining mood-specific movie
33. Belk RW: Situational variables and consumer behavior. J. similarity with matrix factorization for context-aware
Consum. Res. 1975, 2:157-164. recommendation. In Proceedings of the Workshop on Context-
34. Kenrick DT, Funder DC: Profiting from controversy: lessons aware Movie Recommendation; ACM: 2010:34-40.
from the person-situation debate. Am. Psychol. 1988, 43:23. 51. Tkal9
ci9
c M, Burnik U, Koir A: Using affective parameters in a
35. Fleeson W, Noftle E: The end of the person–situation debate: an content-based recommender system for images. User Model.
emerging synthesis in the answer to the consistency question. User-adapt. Interact. 2010, 20:279-311.
Soc. Pers. Psychol. Compass 2008, 2:1667-1684. 52. Little BR: Free traits, personal projects and idio-tapes: three
36. Bagozzi RP, Gopinath M, Nyer PU: The role of emotions in tiers for personality psychology. Psychol. Inq. 1996, 7:340-344.
marketing. J. Acad. Mark. Sci. 1999, 27:184-206. 53. Iyengar SS, Lepper MR: When choice is demotivating: can one
37. Watson D, Clark La, Tellegen a: Development and validation of desire too much of a good thing? J. Pers. Soc. Psychol. 2000,
brief measures of positive and negative affect: the PANAS 79:995-1006.
scales. J. Pers. Soc. Psychol. 1988, 54:1063-1070. 54. Schwartz B, Ward A: Doing better but feeling worse: the
38. Zeng Z, Pantic M, Roisman GI, Huang TS: A survey of affect paradox of choice. In Positive Psychology in Practice. Edited by
recognition methods: audio, visual, and spontaneous Linley AP, Joseph S. John Wiley & Sons, Inc.; 2004:86-104.
expressions. IEEE Trans. Pattern Anal. Mach. Intell. 2009, 31:39- 55. Matz S, Gladstone J, Stillwell D: Money buys happiness when
58. spending fits our personality. Psychol. Sci. 2016, 27:715-725.
39. D’mello SK, Kory J: A review and meta-analysis of multimodal 56. Noar SM, Benac CN, Harris MS: Does tailoring matter? Meta-
 affect detection systems. ACM Comput. Surv. 2015, 47:43. analytic review of tailored print health behavior change
D’mello and Kory provide a systematic review of multimodal affect interventions. Psychol. Bull. 2007, 133:673-693.
detection systems. Based on a quantitative review and meta-analysis
of 90 peer-reviewed affect detection systems, they show that (a) the 57. Mann T, Sherman D, Updegraff J: Dispositional motivations and
majority of systems relies on person-dependent models, fusing audio and message framing: a test of the congruency hypothesis in
visual, (b) multimodal systems were consistently better than unimodal college students. Heal. Psychol. 2004, 23:330.
systems, and (c) systems were substantially less accurate for natural than
for acted emotional reactions. 58. Bagby RM, Vachon DD, Bulmash EL, Toneatto T, Quilty LC,
Costa PT: Pathological gambling and the five-factor model of
40. Cowie R, Douglas-Cowie E, Savvidou* S, McMahon E, Sawey M, personality. Pers. Individ. Dif. 2007, 43:873-880.
Schröder M: FEELTRACE: An instrument for recording
perceived emotion in real time. ISCA Tutorial and Research 59. McDonald AM, Cranor LF: Americans’ attitudes about internet
Workshop (ITRW) on Speech and Emotion 2000. behavioral advertising practices. In Proceedings of the 9th
Annual ACM Workshop on Privacy in the Electronic Society; ACM:
41. Teixeira T, Wedel M, Pieters R: Emotion-induced engagement in 2010:63-72.
internet video advertisements. J. Mark. Res. 2012, 49:144-159.
60. Ur B, Leon PG, Cranor LF, Shay R, Wang Y: Smart, useful, scary,
42. AlHanai T, Ghassemi MM: Predicting latent narrative mood creepy: perceptions of online behavioral advertising. In
using audio and physiologic data. AAAI 2017. Proceedings of the Eighth Symposium on Usable Privacy and
Security; ACM: 2012:4.
43. LiKamWa R, Liu Y, Lane ND, Zhong L: Moodscope: building a
 mood sensor from smartphone usage patterns. Proceeding of 61. Goldfarb A, Tucker CE: Privacy regulation and online
the 11th annual international conference on Mobile systems, advertising. Manage. Sci. 2011, 57:57-71.
applications, and services; ACM: 2013:389-402.
LiKamWa et al. demonstrate the validity of assessing people’s mood on 62. Tucker CE: Social networks, personalized advertising, and
the basis of their smartphone sensor data. Using smartphone logged data  privacy controls. J. Mark. Res. 2014, 51:546-562.
of 32 participants collected over a period of two months, they show that Tucker investigates how consumers’ perceived control over privacy
people’s daily mood can be accurately predicted from their communica- affects their willingness to engage with personalized advertising. Using
tion history and patterns of application usage. data from a randomized field experiment, she demonstrates that giving
people more control over their personally identifiable information
44. Park H-S, Yoo J-O, Cho S-B: A Context-aware Music increases– rather than decreases – the likelihood of them clicking on
Recommendation System Using Fuzzy Bayesian Networks with personalized ads.

Current Opinion in Behavioral Sciences 2017, 18:7–12 www.sciencedirect.com