SOCIAL MEDIA AND WEB ANALYTICS
Table of Contents
WEEK 1 ......................................................................................................................... 3
Lecture 1 – Course Introduction............................................................................................3
Lecture 2 – The Design of Empirical Research .......................................................................4
The E ect – Chapter 1, 2, and 5 .......................................................................................... 12
Intrinsic vs. Image-Related Utility in Social Media: Why Do People Contribute Content to
Twitter? – Toubia & Stephen ................................................................................................ 14
WEEK 2 ....................................................................................................................... 15
Lecture 3 – Causation & Randomized Experiments .............................................................. 15
Put Your Mouth Where Your Money Is: A Field Experiment Encouraging Donors to Share About
Charity – Silver & Small ...................................................................................................... 26
Randomization and Causality ............................................................................................. 26
Lab Regression .................................................................................................................. 28
WEEK 3 ....................................................................................................................... 30
Lecture 4 – A/B Tests: The Essentials ................................................................................... 30
Lab Identification ............................................................................................................... 38
Statistical Challenges in Online Controlled Experiments: A Review of A/B Testing Methodology
– Larsen et al. .................................................................................................................... 40
The surprising power of online experiments – HBR............................................................... 40
Online Experimentation: Benefits, Operational and Methodological Challenges, and Scaling
Guide – Bojinov & Gupta ..................................................................................................... 41
The A/B Test: Inside the Technology That's Changing the Rules of Business .......................... 41
WEEK 4 ....................................................................................................................... 42
Lecture 5 – A/B Testing: Next Steps ..................................................................................... 42
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-Experiment Data –
Deng et al. ......................................................................................................................... 52
The E ect – Your standard errors are probably wrong ........................................................... 53
WEEK 5 ....................................................................................................................... 54
Lecture 6 – Di erences in Di erences................................................................................. 54
The E ect – Chapter 17 Event Studies ................................................................................. 62
The E ect – Chapter 18 Di erence-in-Di erences ............................................................... 63
WEEK 6 ....................................................................................................................... 64
Lecture 7 – Di in Di : Applications .................................................................................... 64
Consumer heterogeneity and paid search e ectiveness: A large-scale field experiment – Blake
et al. .................................................................................................................................. 81
1
, Does Online Word-of-Mouth Increase Demand? (and How?) Evidence from a Natural
Experiment – Seiler et al. .................................................................................................... 81
WEEK 7 ....................................................................................................................... 82
Lecture 8 – Intro to Text Analytics ........................................................................................ 82
Text Mining with R – Chapter 1 ............................................................................................ 83
Text Mining with R – Chapter 2 ............................................................................................ 83
Text Mining with R – Chapter 3 ............................................................................................ 84
Lecture 9 – Sentiment Analysis ........................................................................................... 85
VADER: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text ........ 85
WEEK 8 ....................................................................................................................... 86
Text Mining with R – Chapter 6 ............................................................................................ 86
2
,WEEK 1
Lecture 1 – Course Introduction
What is marketing analytics?
Marketing analytics is the practice of collecting and analysing consumer and firm data to
optimize a firm’s marketing e ectiveness and improve business/marketing decisions.
- A young field with fast progress:
o Since the 1970s: conjoint analysis
o Since the 1990s: structural models
o In the last decade: rise of modern causal inference
- The number of methods is increasing fast
- Most important methods originate outside the discipline of marketing
o Near the applications, the substance, the problem to be solved
o From adjacent fields: economics, statistics, psychology, data science, political
science
Marketing analytics is an unusually diverse discipline, cross-roads of other fields, great place for
a broad perspective on methods.
What is the subfield of digital marketing analytics?
- Digital marketing analytics ↔ social media and web analytics
o Applying marketing analytics to the online world
Websites, online advertising, retail platforms, social media
- Quickly becoming one of the largest fields within marketing
o Lots of data
o Increasingly the main place where consumers and firms interact
What kind of empirical analyses are of interest to us as marketers?
- Descriptive analysis
- Causal analysis
- Predictive analysis
Descriptive analysis: summarise characteristics of a dataset
- What does the data look like?
o Means, standard deviations, distribution of data
o Results are (stylized) facts
- Examples:
o How are users who discuss the US election connected on Twitter?
o What topics are discussed on Yelp reviews?
o Are discussions on Reddit about Albert Heijn di erent from those on Twitter?
Causal analysis: does A lead to B?
- Might also care about the mechanism of how it happens
- Examples:
o Do Facebook ads increase product purchases?
o Does product adoption by influencers increase demand?
o Do tweets by TV studios increase the number of viewers of their show?
3
, Predictive analysis: how can I best predict an outcome?
- When A occurs, so does B
- Examples:
o Is this review posted by a real person or by a bot?
o How many retweets does Nike expect its next tweet to get?
o Who is a new Twitter user likely to follow?
Social media & web analytics needs to combine tools from multiple areas:
1. Statistical/econometric methods
2. Text analytics – text-as-data
3. Network analytics
4. Machine learning
The exact mix of these used in any project depends on:
- The question you want to answer
o Example: can one deliver valuable insight by ignoring the network structure?
- Personal taste
High quality social media & web analytics is incredibly useful
Why?
- Impacts a wide variety of industries
o Media & entertainment, politics, health care, FMCG, fashion & beauty, etc.
- It provides real answers to real problems in marketing and business strategy
o And people care about the answers
Lecture 2 – The Design of Empirical Research
How does the world work?
- We’ll never know everything perfectly
- There’s always scope for new research
- We’ll need to be comfortable with simplifications
A good research question is:
1. Well-defined
2. Answerable
3. Understandable to the audience that you need to deliver it to - context-specific. E.g.,
“how does price of yoghurt influence quantity bought?” should be framed di erently
when the audience is a group of academics (talk about price elasticity etc.) vs. a group of
managers from Danone (talk about how demand changes).
Our goal: conduct research in a way that’s capable of answering the questions we asked
The focus of this class: quantitative empirical research
Empirical research:
- Uses (structured) observations from the real world to attempt to answer questions
Quantitative:
4
Table of Contents
WEEK 1 ......................................................................................................................... 3
Lecture 1 – Course Introduction............................................................................................3
Lecture 2 – The Design of Empirical Research .......................................................................4
The E ect – Chapter 1, 2, and 5 .......................................................................................... 12
Intrinsic vs. Image-Related Utility in Social Media: Why Do People Contribute Content to
Twitter? – Toubia & Stephen ................................................................................................ 14
WEEK 2 ....................................................................................................................... 15
Lecture 3 – Causation & Randomized Experiments .............................................................. 15
Put Your Mouth Where Your Money Is: A Field Experiment Encouraging Donors to Share About
Charity – Silver & Small ...................................................................................................... 26
Randomization and Causality ............................................................................................. 26
Lab Regression .................................................................................................................. 28
WEEK 3 ....................................................................................................................... 30
Lecture 4 – A/B Tests: The Essentials ................................................................................... 30
Lab Identification ............................................................................................................... 38
Statistical Challenges in Online Controlled Experiments: A Review of A/B Testing Methodology
– Larsen et al. .................................................................................................................... 40
The surprising power of online experiments – HBR............................................................... 40
Online Experimentation: Benefits, Operational and Methodological Challenges, and Scaling
Guide – Bojinov & Gupta ..................................................................................................... 41
The A/B Test: Inside the Technology That's Changing the Rules of Business .......................... 41
WEEK 4 ....................................................................................................................... 42
Lecture 5 – A/B Testing: Next Steps ..................................................................................... 42
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-Experiment Data –
Deng et al. ......................................................................................................................... 52
The E ect – Your standard errors are probably wrong ........................................................... 53
WEEK 5 ....................................................................................................................... 54
Lecture 6 – Di erences in Di erences................................................................................. 54
The E ect – Chapter 17 Event Studies ................................................................................. 62
The E ect – Chapter 18 Di erence-in-Di erences ............................................................... 63
WEEK 6 ....................................................................................................................... 64
Lecture 7 – Di in Di : Applications .................................................................................... 64
Consumer heterogeneity and paid search e ectiveness: A large-scale field experiment – Blake
et al. .................................................................................................................................. 81
1
, Does Online Word-of-Mouth Increase Demand? (and How?) Evidence from a Natural
Experiment – Seiler et al. .................................................................................................... 81
WEEK 7 ....................................................................................................................... 82
Lecture 8 – Intro to Text Analytics ........................................................................................ 82
Text Mining with R – Chapter 1 ............................................................................................ 83
Text Mining with R – Chapter 2 ............................................................................................ 83
Text Mining with R – Chapter 3 ............................................................................................ 84
Lecture 9 – Sentiment Analysis ........................................................................................... 85
VADER: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text ........ 85
WEEK 8 ....................................................................................................................... 86
Text Mining with R – Chapter 6 ............................................................................................ 86
2
,WEEK 1
Lecture 1 – Course Introduction
What is marketing analytics?
Marketing analytics is the practice of collecting and analysing consumer and firm data to
optimize a firm’s marketing e ectiveness and improve business/marketing decisions.
- A young field with fast progress:
o Since the 1970s: conjoint analysis
o Since the 1990s: structural models
o In the last decade: rise of modern causal inference
- The number of methods is increasing fast
- Most important methods originate outside the discipline of marketing
o Near the applications, the substance, the problem to be solved
o From adjacent fields: economics, statistics, psychology, data science, political
science
Marketing analytics is an unusually diverse discipline, cross-roads of other fields, great place for
a broad perspective on methods.
What is the subfield of digital marketing analytics?
- Digital marketing analytics ↔ social media and web analytics
o Applying marketing analytics to the online world
Websites, online advertising, retail platforms, social media
- Quickly becoming one of the largest fields within marketing
o Lots of data
o Increasingly the main place where consumers and firms interact
What kind of empirical analyses are of interest to us as marketers?
- Descriptive analysis
- Causal analysis
- Predictive analysis
Descriptive analysis: summarise characteristics of a dataset
- What does the data look like?
o Means, standard deviations, distribution of data
o Results are (stylized) facts
- Examples:
o How are users who discuss the US election connected on Twitter?
o What topics are discussed on Yelp reviews?
o Are discussions on Reddit about Albert Heijn di erent from those on Twitter?
Causal analysis: does A lead to B?
- Might also care about the mechanism of how it happens
- Examples:
o Do Facebook ads increase product purchases?
o Does product adoption by influencers increase demand?
o Do tweets by TV studios increase the number of viewers of their show?
3
, Predictive analysis: how can I best predict an outcome?
- When A occurs, so does B
- Examples:
o Is this review posted by a real person or by a bot?
o How many retweets does Nike expect its next tweet to get?
o Who is a new Twitter user likely to follow?
Social media & web analytics needs to combine tools from multiple areas:
1. Statistical/econometric methods
2. Text analytics – text-as-data
3. Network analytics
4. Machine learning
The exact mix of these used in any project depends on:
- The question you want to answer
o Example: can one deliver valuable insight by ignoring the network structure?
- Personal taste
High quality social media & web analytics is incredibly useful
Why?
- Impacts a wide variety of industries
o Media & entertainment, politics, health care, FMCG, fashion & beauty, etc.
- It provides real answers to real problems in marketing and business strategy
o And people care about the answers
Lecture 2 – The Design of Empirical Research
How does the world work?
- We’ll never know everything perfectly
- There’s always scope for new research
- We’ll need to be comfortable with simplifications
A good research question is:
1. Well-defined
2. Answerable
3. Understandable to the audience that you need to deliver it to - context-specific. E.g.,
“how does price of yoghurt influence quantity bought?” should be framed di erently
when the audience is a group of academics (talk about price elasticity etc.) vs. a group of
managers from Danone (talk about how demand changes).
Our goal: conduct research in a way that’s capable of answering the questions we asked
The focus of this class: quantitative empirical research
Empirical research:
- Uses (structured) observations from the real world to attempt to answer questions
Quantitative:
4