Transference Coder

⬆

Load your CSV

Drop your blind_coding CSV here, or click to browse.
Pre-filled post data loads automatically.

Choose file

— or —

All data stays in your browser — nothing is uploaded anywhere.

Each CSV's autosave is kept separate automatically. If you ever want a hard reset anyway, clear all saved progress.

0 / 0

— — uncoded

—

0.6× 4×

Step 1 — self-disclosure  iIs the author revealing something personal about their emotional or relational experience with an AI?
self_disclosure

Step 2 — transference  iIs this person unconsciously projecting relational patterns from their interpersonal history onto the AI?
self_disclosure = 0 → auto-set to 0
manual_label_transference

Step 3 — valence (T-positive only)iWhat is the emotional tone of the transference — positive, negative, or mixed?
transference = 0 → leave blank
transference_valence

Step 4 — post intention (SD only)iWhat is the primary communicative purpose of this post?
self_disclosure = 0 → leave blank
post_intention

Step 5 — metadata (all posts)iCode for every post — age, gender, relationship status, and any alternative tool used.
age_range
gender
relationship_status
alt_tool_name
alt_tool_type
alt_tool_name (custom)
Step 6 — types (T-positive only)iWhich transference types are present? Check all that apply — multiple types are common.
transference = 0 → leave blank

              Erotic / romantic E
type_erotic_romantic

              Dependency / separation D
type_dependency_separation

              Idealization / rescue I
type_idealization_rescue

              Self-worth / validation V
type_selfworth_validation

              Rupture / devaluation R
type_rupture_devaluation

              Identity fusion U
type_identity_fusion

              Other T
type_other — describe in notes

Step 7 — coder notes  iRequired if type_other = 1. Encouraged for all ambiguous or borderline cases.
manual_coder_notes
Media coding  iDoes this post contain images or video? View the original Reddit post via permalink to check.
media_present

media_type
media_count
media_shows_ai_persona
media_shows_conversation
media_content_type
media_emotion_expressed
media_conversation_topic
media_relevant_to_transference
media_user_ai_dynamic
media_coder_notes
⚠ Check before saving

📝 Research Memo

Jot observations, patterns, and hunches as you code. This is your research diary — saved locally alongside your data.

— Codebook

How to use this tool

Transference Coder — AI Companion Relationship Study

Upload your CSV

On the start screen, drag and drop your blind_coding CSV file or click "Choose file." The app loads all 1,000 posts automatically and opens at your first uncoded entry. Post text is pre-filled — you only need to fill in the coding columns.

Read the post, then code

The post text appears on the left. Code on the right. Follow the steps from top to bottom — the panel guides you by highlighting the active step and dimming steps that aren't applicable yet.

Step 1 — is self-disclosure present? This unlocks the rest of the form.
Steps 2–4 — only apply if SD = 1. Steps 3 and 6 only apply if transference = 1.
Step 5 — always fill in metadata (age, gender, relationship status).
Media — check the original Reddit post via the permalink before coding.

Use the info icons

Each step header has a small i icon. Hover it to see the core decision question. Click it to open the full codebook panel on the right with definitions, indicators, and decision rules pulled directly from the coding guide.

Save and advance

Hit Save & Next → or press → to save the current entry and move to the next post. Use Space to skip without saving. Use ← to go back. Jump to any specific post by typing its number in the "go to" box.

Flag uncertain entries

Click ⚑ Flag in the post header (or press F) to mark any entry you want to revisit. Flagged posts are tracked in the Overview and get an amber border. You can filter by flagged entries to review them all at once.

Overview & charts

Press Tab or click "Overview" in the topbar to see all posts in a grid. Filter by coded, uncoded, transference = 1, or flagged. Search posts by keyword. Live charts show transference rates by subreddit, type distribution, valence breakdown, and post intention as you code.

Export your data

Click ↓ Export CSV in the topbar at the end of every session. Re-upload that exported file next session — the app reads whatever is already coded and picks up where you left off. Your progress also autosaves to your browser's local storage as a backup.

Working with colleagues? Each person works on their own copy of the CSV independently. Export when done and compare exported files for inter-rater reliability. Your browser's local storage is not shared between people or machines.

entries coded

You did it.

You just hand-coded 1,000 Reddit posts about human–AI relationships. That's not a small thing. Every judgment call, every borderline case, every "I know it's just an AI but..." — you read them all. The data exists because you did this work.