Images and video often contain more than one analytical event. Coding the entire frame can hide whether a claim concerns a person, object, symbol, gesture, setting, or relationship between elements. Region-based coding lets the researcher mark the smallest visual area that can support a meaningful interpretation while keeping the whole image available as context.
Choose the unit before you draw
A region is not automatically a rectangle around an object. It represents your analytic unit. In a child’s drawing, a region might enclose a route, a school building, a cluster of symbolic marks, or a figure. In a classroom video, it might capture a gesture, a shared interactional space, or a sequence within a defined time range. State what the boundary represents in the code definition or memo.
Keep visual and verbal evidence connected
Visual evidence often gains meaning from an accompanying explanation. A child may point to a shape and call it a park; an observer may record that a participant laughed before answering; a video may show a gesture that changes how an utterance is understood. Code the visual region and keep the linked text, memo, or timestamp accessible rather than forcing all meaning into the image alone.
Avoid three common errors
- Over-selection: coding almost the whole image when only one element supports the claim.
- Decontextualisation: cropping or describing an object as if the surrounding scene did not matter.
- False precision: drawing a very exact boundary when the analytic unit is actually ambiguous or relational.
Use memos for interpretive uncertainty
A visual region can support more than one plausible reading. Instead of forcing premature certainty into a code label, record the uncertainty in a memo: what is visible, what contextual information supports the interpretation, and what would change the decision. This is especially important when working across cultures, languages, or age groups.
When video needs both region and time
For video, the frame and the timeline may both matter. A code anchored only to a visual region may not show when an action occurs; a code anchored only to time may not show where the relevant action is. Use a time range for the event and a visual region when its spatial detail is analytically relevant.
Questions researchers often ask
Can one region receive several codes?
Yes. A single visual selection may be relevant to setting, symbolic meaning, interaction, and affect. Multiple coding should reflect distinct analytic purposes, not duplicate labels.
Should I code every visible element?
Only when the research question requires it. Coding is systematic attention, not exhaustive object detection.