Work

PROJECT HIGHLIGHTS

Coin-Op TV In The News

A project in which we had turned people's TV sets into coin-operated machines to study how they make programming choices got written up in Advertising Age.


Health Motives

Used COMET in an analysis of long-form social media posts by chronic patients. Found clear differences in emotional needs between segments based on patient demographics and the length of time with the condition.


No Goal Worth Saving For

Survey: 5% in a sample of 400 adults have no goal worth saving money for.


Friends' Recommendations Matter More Than Celebs'

Survey: one's choice of a restaurant is most easily influenced, in descending order, by: a recommendation from someone, reviews on Yelp, expert reviews, celebrity endorsement (based on 691 responses).

In a separate survey, we found out that one's self-reported likelihood to follow up on a friend's product recommendation shared in social media varies by product category. Recommendations for movies, restaurants and getaway destinations are more likely to be investigated than lawn-care services, insurance companies, and sports facilities.

Perhaps not surprisingly, people talk about different kinds of products with different frequency. Restaurants and movies are talked about most often. Home repair and lawn care -- least often.


Peering Through A Window to the Soul

Setting up a lab to measure pupillary response to brand stimuli. Theoretically, we should be able to predict one's choice by looking at the changes in one's pupil dilation.


Facebook Ad Formats Vary In Effect

Completed an experimental study of different formats of Facebook ads. Identified formats that are most likely to influence choice in the exposed population -- the effect size can reach 10 percentage points.


SwiTchers And Loyals Have Different Needs

A large study of implicit motivations using an affective Stroop task showed significant differences between brand-related motivations among high- and low- frequency brand users.


GOING OFFLINE

Segmented Twitter users who mentioned a client brand by their offline interests identifiable through their online behavior.


CUSTOMERS Pay It Forward

Analyzed posts in social media that reported on the acts of "paying it forward" taking place in client's stores.


2chainz Bot Attracts Fans

Jared builds a Twitter bot that auto-generates lyrics in the style of the famous rapper 2chainz. The bot has attracted a small core of fans who interact with the bot regularly.


When Do Soccer Fans Tweet?

Based on an analysis of 500,000 Tweets about the FIFA Confederation Cup, established that game events such as kickoff, halftime, and goals trigger the greatest number of tweets posted during the game, as compared to all other topics.


Do You Really Know Your Followers?

Analyzed a client's Twitter followers to segment them by their occupation. Project involved cross-referencing Twitter and LinkedIn. Proved useful in guiding content strategy. Currently tracking changes in audience composition.


We Know What They Did Last Summer

Isolated a sample of a client's confirmed customers who are active in social media, then analyzed years of autobiographical posts to develop detailed personas.


Emotions Influence Consideration Sets

Conducted an experiment to establish whether different purchase triggers can result in formation of different consideration sets. Primed respondents with happy or angry scenarios. Confirmed: consideration sets' size and composition vary depending on the trigger's emotional valence.


What Types of Photos Get Shared?

Identified genres of photos taken by customers that are most likely to be shared by their friends.


Small Items Cause Big Excitement on Black Friday

Analyzed a sample of tweets posted on Black Friday. Identified a category of goods that trigger outsized excitement expressed in social media posts relative to their purchase price.


If You Bought This...

Harvested and analyzed Amazon's "People who bought this..." suggestions to identify product clusters.


Novelty and Familiarity In Entertainment Choices

In a survey, people strongly preferred:

- To play board games that they have played in the past

- To play video games that they have played in the past

- To watch movies they haven't yet watched.

The difference between movies and video games surprised us.


Two Kinds of Trust

Researched two types of trust violations: integrity trust violation (immoral acts) and competency trust violations (incompetent acts). Different types of trust violations require different approaches towards trust repair.

Fame Isn't Everything

An experiment clearly demonstrated that it's not enough for a brand to be known by many; what people know about the brand matters even more. Saw vast differences in share of choice between equally well-known brands.


ALL THE SAME

An experiment shows that most major insurance brands are viewed favorably but with little excitement; six of eight brands share the same average pleasure/arousal score.


COMET Updated

Updated COMET system of human motivations to reflect new knowledge.


Prospect Theory in Health Forums

The Prospect Theory applied to long-form social media posts helps predict patients' likelihood of regimen change.


Women in Social

Survey: in an online sample of 1,400 women (age 25-65), 89% have a Facebook account, 61% have Pinterest, 53% have Twitter, 41% have Instagram, and 6% have an account on a dating site.


Jingle Bells

Survey: 85% in a sample of 1,400 Americans decorate a Christmas tree. About 20% made at least one ornament themselves in the past.


Audience Engineering

Developed an "audience engineering" methodology based on social graph analysis for optimizing Twitter outreach.  One part of it: target people who are followed by people you need to reach.


Being Liked Is Not Enough

Experimentally demonstrated that high arousal + positive valence are more predictive of choice than positive valence alone. In other words, being liked is often not enough, but it depends on how people feel about your competition.


BETTER METRICS FOR Consumer Reviews

Introduced two new metrics useful for analyzing reviews on Tripadvisor:  weekly average score, and the average score of the reviews on the first page. Both are more informative than the overall average score, which doesn't change much after a property reaches a certain number of reviews. Showed recurring nature of the weekly averages.


Machiavelli Was A Yelper

Explored personality types of respondents who are more likely to post online complaints about companies. Established a correlation between likelihood to post and the Machiavellian personality type (high Machs).


Epic Stimuli

Developed stimuli for a client's 210-cell cognitive profiling study.


Apples & Oranges

Analyzed a set of social media posts containing direct product comparisons.


On Slippery Slopes

Six months before the games, conducted a headline risk analysis related to the Sochi 2014 Winter Olympics for a client sponsoring the games. Forewarned is forearmed: the analysis helped the client prepare for contingencies.


We Question the Wisdom of Customer Support Via Social Channels

Experimentally established that the way a company handles customer comments on Facebook can have an effect on how witnesses of the interaction perceive the company.

A separate experiment showed that even constructive handling of customer complaints posted on a brand's Facebook page can have an overall negative effect on the audience, likely because the original complaint is often the only part of the exchange that is seen.


New Year's Resolutions Decoded

Analyzed 800 Facebook posts containing New Year's resolutions.  Resolutions posted by people (as opposed to "pages") vary in how specific they are (intent vs commitment) and in the object of desired change (intrinsic vs extrinsic). Serious pledges are rarely verbalized; instead, people rely on props such as userpics, likes etc as well as sharing third-party content to signal their intent. Identified 10 common types of resolutions posted by pages.


America's Most Photo-Shared Ballparks

Ranked all major ballparks by the amount of photo-sharing activity relative to average attendance.


Researchers Go Bananas

80% of Americans 18-50yo we surveyed peel bananas from the stem end (from the "handle"). Half of respondents (51.3%) feel strongly or very strongly that "that the way you peel a banana is better than any other way." 84.7% "have a perfectly logical reason why I peel bananas they way I do."


Media Planning Around Evolving Memories

Memories of a trip change over time. When targeting past visitors, timing matters not only because of their plans for the future, but also because of their evolving memories of the past.


We Jerzify Ourselves for Science

Used Jersey Shore to explore the predictors of information propagation in social networks, and the network dynamics of Twitter communities. Read our full report, or just the highlights in Ad Age. The experiment was also cited in Spreadable Media.


Online Video in Lean-Back Mode

Built Youtoonies, an application that uncovers and plays retro cartoons as a continuous stream. Built on YouTube's Related Videos api, Youtoonies explores the potential of lean-back web video viewing.


Does this bag make me look old?

Compared the impact of shopping bags from four brands on how others perceive certain attributes of the bag owner. A bag with Sears logo makes others judge you as older and less affluent.

Reviews Affect Market Share

An experiment revealed how online customer ratings for a business can have a dramatic effect on its share of the local market.


Would You Buy A Green Toothpaste?

Toothpaste in a blue tube was preferred by a large margin over toothpaste in a green tube in two separate experiments. The preference forms in under 100ms. A word of caution to the makers of herbal and other natural toothpastes.


What Matters When Picking Movies

Asked English speakers to pick a movie in Chinese to isolate factors that influence entertainment decisions.


Vectors of Desire Point to Choice

Working on a "vectors of desire" theory:  that the direction of human motivation is defined by two forces: maximization of pleasure and optimization of arousal level. Knowing one's current emotional state and one's emotional evaluation of available alternatives, we should be able to predict one's choice.


Why We Go to Church

Completed a study of motives behind church attendance. It's complicated, but the very variety of motives we have identified helps explain the enduring popularity of the institution.


Tweeting Products Works

Showed how exposing users to product tweets increases their level of arousal towards the brand, which in turn improves the odds of a favorable choice.


Science Goes For a Séance

Participated in a group psychic séance. Presumably communicated with dead relatives. No winning lottery numbers were revealed.


More PromotERS Than Preventers In the US

Survey: a study of regulatory focus based on an online sample of 1,000 US adults showed a skew towards promotion over prevention in the general American population.


Forums a Gold-Mine for SEM Knowledge

Compiled a list of long-tail keywords for search-engine marketing by performing content analysis of customer posts on social media.


BIG DEBATE SETTLED

Asked 400 US adults 18+: "In a fight between a pirate and a ninja, who do you think would prevail?"


Effective Facebook Post Genres Revealed

Looked at 20 brands' six months of Facebook posts to identify genres typical for the category. Identified genres with consistently high engagement levels.


Tomorrow's Emotions Influence Today's Decisions

Conducted online point-of-purchase intercepts to understand how expectations of emotions to be caused by a future consumption experience influence buying decisions in the present.


Learning Makes For Better Work

Survey: 71.2% in our sample of 400 adult Americans "love" or "like" their jobs in general. 1.2% "hate" their jobs.

Which of the following makes their work better?

- Learning new things, ideas: 61.5%

- Collaborating with others: 38.2%

- Focusing on your work alone: 34.7%

- Socializing ideas with others 29.5%


How People Recommend Brands on Facebook

Analyzed how people ask their friends for a product recommendation on Facebook; surveyed frequency of mentions by category (some things are more likely to be brought up in a conversation than the others). Correlated market share, share of voice, and share of recommendations.


What Men Want

Modeled men’s fashion shopping behavior by analyzing their in-store tweets.

In a separate experiment, we debunked the belief that men are more rational in their purchases than women. High levels of excitement about a product are more predictive of a purchase made by men than by women.

In yet another experiment, we showed how a man's haircut influences how others estimate his age.

How do men decide when it's time for a haircut?

"I have to comb my hair so it looks nice."

"Brushing and combing become daily task."

"My wife says I have a 'skullet'."

"When my sideburns can reach below my nose."

"The back of my neck gets a bit bushy."

Heuristics.


Tweeting Up a Storm

Analyzed topics and measured the increasing emotional intensity of tweets posted by people living in the path of an approaching hurricane.


Half-Life of Scandal

Measured the half-life of public interest in a scandal by triangulating social media posts, search queries, and syndicated brand tracking data. All 10 scandals we investigated follow the same pattern of public interest.


Insurance Brands in Rap Songs

Analyzed cultural significance of financial service brands in hip-hop lyrics quoted by social media users.

Separately, we measured cultural impact of an ad campaign by analyzing mentions of it by public figures and celebrities outside of the broadcast window.


A Week Without Cable

A deprivation experiment that left five Boston-area families without cable over a holiday week demonstrated how much people rely on the contiguous nature of TV programming. Read the full report.