Theory as a Lens

Today

Theory is the part of this course that sounds least useful and turns out to decide the most. It is what tells you which of the thousands of chat messages in front of you are worth counting.

You leave today with a lens picked for your own project, and a reason for picking it that you could say out loud to somebody who disagreed.

One finding, three explanations

Livestream viewing looks more socially motivated than content motivated. Why?

• Uses and gratifications (Katz, Blumler, & Gurevitch, 1973): viewers are active agents meeting a social need a finished video cannot
• Parasocial interaction (Horton & Wohl, 1956): viewers form one-sided bonds with the streamer, and connection is how that bond feels
• Social identity (Tajfel & Turner, 1979): the community is a group, and chat is a space of belonging

Same pattern, three research programs, three different studies.

Theory is not speculation

• In casual use “theory” means guess. In research it means the opposite of a guess.
• A theory is a formal, systematic explanation of relationships between concepts
• Logically coherent: the parts fit together without contradiction
• Generalizable: it applies across contexts rather than to one case
• Falsifiable: it makes predictions that could turn out wrong
• Generative: it produces new hypotheses rather than closing the question

Social identity, worked

People take part of their self-concept from the groups they belong to (Tajfel & Turner, 1979). Watch what that produces:

• Viewers rate their preferred community above rivals, especially on subjective dimensions like authenticity
• Criticize a streamer and the regulars respond defensively
• In a different domain entirely: partisans rate aligned news outlets as more credible even when the factual content is identical

One idea, three predictions, all of them checkable.

The lens metaphor

• Theory works like a camera lens. Some things come sharp, some fall outside the frame.
• Appraisal theory focuses you on individual evaluation, so you build surveys and experiments measuring personal reactions
• Social identity focuses you on group meaning, so you analyze how members talk about the stream as a marker of who they are
• Same phenomenon, different designs, and neither one is wrong
• You cannot use every lens at once. Trying produces blur, not depth.

Three paradigms

• Social scientific: an objective reality that can be observed and measured. Deductive, so you start from theory and test. Asks does X cause Y?
• Interpretive: reality is built through interaction. Inductive, so you observe and build theory from what you find. Asks what does X mean to the people who live it? Methods include grounded theory and thematic analysis (Braun & Clarke, 2006).
• Critical and cultural: power shapes what counts as knowledge. Asks whose interests does X serve, and whose voices go unheard?

Grand and middle-range

• Grand theories explain human behavior across every context: Marxism, psychoanalysis, structuralism
• Intellectually powerful, and close to untestable. The scope is too wide and the predictions too general to be proven wrong.
• Middle-range theories are specific enough to generate testable hypotheses and broad enough to travel past one case
• Robert Merton named the zone. Every theory in the catalogue ahead sits in it.

Catalogue: relationship and group

• Parasocial interaction (Horton & Wohl, 1956): one-sided bonds with media figures that feel intimate but are not reciprocated. Predicts stronger connection with more consumption, and a sense of betrayal when a scandal breaks. The lens for donations to strangers, and for influencer marketing.
• Social identity (Tajfel & Turner, 1979): group membership shapes self-concept. Predicts in-group favouritism and symbolic boundary keeping. Foundational in political communication and sports fandom.

Catalogue: media and message

• Agenda setting (McCombs & Shaw, 1972): media do not tell people what to think, they tell people what to think about. On Twitch the front page, the recommendations, and the category rankings are all setting an agenda.
• Framing (Goffman, 1974; Entman, 1993): how information is presented shapes how it is read. Episodic framing pushes blame toward individuals, thematic framing toward structures. You can see it in how streamers title their own broadcasts.

Catalogue: audience and exposure

• Uses and gratifications (Katz, Blumler, & Gurevitch, 1973): audiences are active and pick media to meet needs. It is the lens behind most research on why people watch other people play video games (Sjöblom & Hamari, 2017).
• Cultivation (Gerbner & Gross, 1976): long-term cumulative exposure shapes how people perceive social reality. Heavy consumers overestimate what they keep seeing, most strongly where they have no direct experience to check it against.

Two lenses on the same chat

Four messages from one Twitch broadcast:

DragoBoi89: xQc actually playing well today wtf

emotelord: PogU PogU

realnamedfan: @xqcow do u remember when u said hi to me last week

streamfan22: chat is this real

Read them once before we put a lens on them.

Two readings

• Uses and gratifications: four needs, four viewers. Evaluative commentary, collective emotional expression, direct acknowledgment, collective sense-making. All four are equally interesting to this lens.
• Parasocial interaction: realnamedfan is the case. An @-address plus a memory claim about a previous acknowledgment treats a one-sided interaction as a continuing relationship.
• emotelord and streamfan22 never address the streamer, so this lens pushes them to the edges. Same four messages. Two different studies.

Four methods, four questions

• Surveys measure what people report. Natural for uses and gratifications, and limited by the gap between what people say and what they do.
• Experiments isolate cause by manipulating something. Natural for “does X cause Y?”, and limited by artificiality.
• Qualitative methods capture meaning and lived experience. Natural for the interpretive question, and they trade breadth for depth.
• Content analysis quantifies features of the content itself, at a scale nobody could eyeball.

Why content analysis here

Three reasons, and none of them is that it is the best method:

• Fit: the Twitch data is content, millions of messages and stream records
• Pedagogical completeness: you practice operationalization, sampling, reliability, and analysis on one dataset, without having to recruit a single human subject
• Transfer: a precise codebook, tested reliability, and systematic sampling are what make every other method work too

Choosing with the alternatives in view beats choosing by default.

Looking ahead to Tuesday

• Tuesday: variables and hypotheses, where a lens turns into something testable
• Independent and dependent variables, mediators, and moderators
• Bring the theory you are leaning toward. You will use it, not just name it.
• The Librarian Visit Report is due this week, 50 points.
• The Annotated Manuscript and CITI Certification are both due next week
• Read the second half of Chapter 5 if you have not yet