The Coding Process for Timeline Analysis

Coding

Putting the pieces together

Step 1: Cache your data

• External Timeline– Twapperkeeper (eg #smph2011)– The Archivist (Desktop)– Twitter Archiver (newish)

• Internal Timeline– Twapperkeeper (own account only)– TwitterArchiver (any public account)

Data Collection Issue

• Limited back stories– Archivist has a limited cache– Twitter to PDF has a 3K previous limit– TwapperKeeper has a background limit

• Plan in advance, or improvise quickly

Step 2: Visual first pass

• Aim: Get a top level sense of the data– Hashtags

• Archivist Visualisations (online)• Archivist Desktop• Summarizer

– Timelines• TweetStats (online)• Tagxedo (online)

The Archiver

Tagxedo

Step 3: The Unpleasant Bit

• Raw data needs to be processed

– Manual

– Automated

Step 4: Coding

• Benefits of Manual Coding– Subjective judgement– You learn a lot– Accuracy in semantics

• Downsides– You learn a lot.– Time consuming.

Coding Process

• Clean the data– Assign IDs to tweets

• Add the additional notes columns– Time Frames– Categories– Sub Category– Sentiment (optional)

Coding Choice

• Visual or numerical?– Code book versus Code words

• Currently: Excel with visual

Excel.

A valued ally.

Step 3: It’s time to start coding

• Top Level Coding– Pass 1: identify time periods of interest– Pass 2: Tag Conversational (@reply)– Pass 3: Identify Pass-Along (RT, http)– Pass 4: News / Status / Phatic

Step 4

• Sub Category Coding– Pass-Along / Conversational

• Easier to code, relatively stable

– Phatic / Status / News• Harder to code• Can (or should) move• Can be multiple category*

New

Exemplar Category

Pass-Along

Pass Along

• Automated Endorsement• Endorsement• Retweet• Secondary Social Media• User Generated Content

Knowing is only half the battle.

The other half involves strategy

Questions

@stephendann | [email protected]