SAMENVATTING DIGITAL
MEDIA SOCIETY
Mariek vda
, Digital methods (oude naam)
Class 1: Introduction digital methods: Close reading, distant reading and common
characteristics of big data
Collection of data create opportunities but also risks for research => harm people,
messing’s, overgeneralize some outcomes
Close reading (quali)
Study of Danah Boyd => making sense of teen life
• Research approach
1) Immersion in teen pop culture and subculture → analyse, get involved
o Consuming media that is popular among teens
2) Participant observation and content analysis of traces on social media
o Observe behavior on social media that is (semi)public
3) Deep hanging out in physical spaces
o Deeper understanding what they are doing (going to the playground, …)
4) Semistructured face-to-face interviews
o All previous info used in the final f-2-f interviews
• Social media certainly make it much easier to peek into people’s lives, but it is also
quite easy to misinterpret online traces
• Example: Gang isignia on MySpace profile → indication of Gang involvement?
o Student => why he lied to Harvard in his letter to enter college => that he quite
the gang => but still posting things about it?
o Alternative explanation:
“Without knowing the specific boy involved, I surmised that he was probably
focused on fitting in, staying safe, or, more directly, surviving in his home
environment. Most likely he felt as though he needed to perform gang
affiliation online—especially if he was not affiliated—in order to make certain
that he was not physically vulnerable.’
Importance of close reading:
• Interpretation based on detailed analysis (Lecture 8 – 11)
• Context is necessary, or thick description, go more in depth
➔ Digital methods allow us to have a more close reading about a context that we try to
study
Distant reading (quanti)
“Big Data”: advances in technology & analysis
• = large amounts of data, also big private companies such as Facebook
• Data being big (multiple datasets in combinations with datasets)
Example: Investigating health and wealth in Rwanda
• Asked people specific questions (demographics, characteristics)
• More valuable than traditional social science survey, less expensive
• Call records (of approx. 1,5 million people)
1
, • Specific info + specific location → high resolution map of wealth in Rwanda
Importance of distant reading: analyzing large numbers of data(sets)
Readymade versus custommade data
• Readymade: repurpose datasets that are created by organizations and
governments
o We will reproduce data that were originally created by companies or
governments or other actors and we can use them for purpose of doing our
social research. (e.g. Data from Facebook)
o Da verwijst dus naar die pisbak omda de onderzoeker die data nog moet
aanpassen voor zijn eigen onderzoek, net zoals die kunstenaar die pisbak
aanpast nr kunst
• Custommade: use tools to create data to answer a specific question
o Start with specifical questions and then use tools (surveys, interviews, …) to
collect this data and answer the question
o verwijst dus nr da kunstwerk omda da door de onderzoeker zelf gecreëerd
wordt from scratch
Digital traces → Big Data → digital methods
• Generally helpful for research: big, always-on, and nonreactive
• Generally problematic for research: incomplete, inaccessible, nonrepresentative,
drifting, algorithmically confounded, dirty, and sensitive
Repurposing
• Created by government for other reasons than social research
• Found versus designed data
o Found data = data designed/found by someone else with a specific purpose
o Designed data = created by researchers themselves with specific purpose
• What should the ideal data set look like?
• Als ge dus bv. web scraping doet dan krijgt ge altijd van die ready made data en
moet ge dus bv. nog kijken of die data wel bij uw onderzoek past (bv. Twitter heeft
een heel specifieke groep van gebruikers die mss ni bij uw onderzoekspopulatie past)
10 characteristics of big data sources
1) Big
Is all that data really doing anything?
Big datasets are never an end in themselves, but do allow for the study of rare cases,
detection small differences, and estimation of heterogeneity
• Still remember: more data is not better or objective
You can use a Scale. But you can also use twitter and look what focus people do rather that
what they say, scrape data from comments,
Analysis of the 2016 US Presidential Campaign on Twitter
2
, • Total of 18 910 250 tweets were analyzed
o 39.1% debate tweets pro Trump hashtag (e.g., #MAGA)
o 13.6% debate tweets pro Clinton (e.g., #ImWithHer)
However…
• 32.7% pro tweets Trump originated from bots
• 22.3% pro tweets Clinton originated from bots
→ we must be careful when looking at big data => A lot of this messages were not put out
there by humans
2) Always-on
• Unexpected events => able to study Unexpected events
• Real-time measurements
• Example: Sandy-related Twitter and Foursquare data
o Combinate this do => found that grocery shops peek the night before the
storm, day after the nightlife was very high, this would be impossible by only
looking at traditional methods
3) Nonreactive
• Measurement in big data sources is much less likely to change behavior
o Participants are not aware that their data is been captured, thus much less
likely to change behavior => more valid results
• However, social desirability bias can still be present
o Achievements only sharing => for better presentation of our self
4) Incomplete
No matter how big your data is, it doesn’t have the ideal information that you want
Usually the following information is missing/ incomplete
• Demographic information about participants E.g., Lying about age, of not saying
their gender
• Behavior on other platforms => all their own website, but not share this, so we have
multiple platforms, but you can also focus on 1
• Data to operationalize theoretical constructs
Construct validity: when there is a perfect alliance between the theoretical construct and
the data that is available → be critical about the context
Example: measuring social capital
• Articulated networks → contacts from email, ….
• Behavioral networks → communication with those contact
Construct validity?
For example: spending more time on the phone with your colleague does not imply that they
are more important that spouses
3
MEDIA SOCIETY
Mariek vda
, Digital methods (oude naam)
Class 1: Introduction digital methods: Close reading, distant reading and common
characteristics of big data
Collection of data create opportunities but also risks for research => harm people,
messing’s, overgeneralize some outcomes
Close reading (quali)
Study of Danah Boyd => making sense of teen life
• Research approach
1) Immersion in teen pop culture and subculture → analyse, get involved
o Consuming media that is popular among teens
2) Participant observation and content analysis of traces on social media
o Observe behavior on social media that is (semi)public
3) Deep hanging out in physical spaces
o Deeper understanding what they are doing (going to the playground, …)
4) Semistructured face-to-face interviews
o All previous info used in the final f-2-f interviews
• Social media certainly make it much easier to peek into people’s lives, but it is also
quite easy to misinterpret online traces
• Example: Gang isignia on MySpace profile → indication of Gang involvement?
o Student => why he lied to Harvard in his letter to enter college => that he quite
the gang => but still posting things about it?
o Alternative explanation:
“Without knowing the specific boy involved, I surmised that he was probably
focused on fitting in, staying safe, or, more directly, surviving in his home
environment. Most likely he felt as though he needed to perform gang
affiliation online—especially if he was not affiliated—in order to make certain
that he was not physically vulnerable.’
Importance of close reading:
• Interpretation based on detailed analysis (Lecture 8 – 11)
• Context is necessary, or thick description, go more in depth
➔ Digital methods allow us to have a more close reading about a context that we try to
study
Distant reading (quanti)
“Big Data”: advances in technology & analysis
• = large amounts of data, also big private companies such as Facebook
• Data being big (multiple datasets in combinations with datasets)
Example: Investigating health and wealth in Rwanda
• Asked people specific questions (demographics, characteristics)
• More valuable than traditional social science survey, less expensive
• Call records (of approx. 1,5 million people)
1
, • Specific info + specific location → high resolution map of wealth in Rwanda
Importance of distant reading: analyzing large numbers of data(sets)
Readymade versus custommade data
• Readymade: repurpose datasets that are created by organizations and
governments
o We will reproduce data that were originally created by companies or
governments or other actors and we can use them for purpose of doing our
social research. (e.g. Data from Facebook)
o Da verwijst dus naar die pisbak omda de onderzoeker die data nog moet
aanpassen voor zijn eigen onderzoek, net zoals die kunstenaar die pisbak
aanpast nr kunst
• Custommade: use tools to create data to answer a specific question
o Start with specifical questions and then use tools (surveys, interviews, …) to
collect this data and answer the question
o verwijst dus nr da kunstwerk omda da door de onderzoeker zelf gecreëerd
wordt from scratch
Digital traces → Big Data → digital methods
• Generally helpful for research: big, always-on, and nonreactive
• Generally problematic for research: incomplete, inaccessible, nonrepresentative,
drifting, algorithmically confounded, dirty, and sensitive
Repurposing
• Created by government for other reasons than social research
• Found versus designed data
o Found data = data designed/found by someone else with a specific purpose
o Designed data = created by researchers themselves with specific purpose
• What should the ideal data set look like?
• Als ge dus bv. web scraping doet dan krijgt ge altijd van die ready made data en
moet ge dus bv. nog kijken of die data wel bij uw onderzoek past (bv. Twitter heeft
een heel specifieke groep van gebruikers die mss ni bij uw onderzoekspopulatie past)
10 characteristics of big data sources
1) Big
Is all that data really doing anything?
Big datasets are never an end in themselves, but do allow for the study of rare cases,
detection small differences, and estimation of heterogeneity
• Still remember: more data is not better or objective
You can use a Scale. But you can also use twitter and look what focus people do rather that
what they say, scrape data from comments,
Analysis of the 2016 US Presidential Campaign on Twitter
2
, • Total of 18 910 250 tweets were analyzed
o 39.1% debate tweets pro Trump hashtag (e.g., #MAGA)
o 13.6% debate tweets pro Clinton (e.g., #ImWithHer)
However…
• 32.7% pro tweets Trump originated from bots
• 22.3% pro tweets Clinton originated from bots
→ we must be careful when looking at big data => A lot of this messages were not put out
there by humans
2) Always-on
• Unexpected events => able to study Unexpected events
• Real-time measurements
• Example: Sandy-related Twitter and Foursquare data
o Combinate this do => found that grocery shops peek the night before the
storm, day after the nightlife was very high, this would be impossible by only
looking at traditional methods
3) Nonreactive
• Measurement in big data sources is much less likely to change behavior
o Participants are not aware that their data is been captured, thus much less
likely to change behavior => more valid results
• However, social desirability bias can still be present
o Achievements only sharing => for better presentation of our self
4) Incomplete
No matter how big your data is, it doesn’t have the ideal information that you want
Usually the following information is missing/ incomplete
• Demographic information about participants E.g., Lying about age, of not saying
their gender
• Behavior on other platforms => all their own website, but not share this, so we have
multiple platforms, but you can also focus on 1
• Data to operationalize theoretical constructs
Construct validity: when there is a perfect alliance between the theoretical construct and
the data that is available → be critical about the context
Example: measuring social capital
• Articulated networks → contacts from email, ….
• Behavioral networks → communication with those contact
Construct validity?
For example: spending more time on the phone with your colleague does not imply that they
are more important that spouses
3
