Break down what participants said by who they are

Question coding has always held two different kinds of thing in one list. The reasons people would adopt a product, the barriers that stop them, the needs nobody is meeting: that is what participants said. Age, country, household income: that is who the participant is. Both were called facets, and research teams told us the word got in the way of the work.

How it works

• Topics are what participants said. Everything the video conversations surfaced against your topic guide, such as reasons to adopt, barriers, unmet needs. This is the inductive coding layer of a study, the part you read to learn something.

• Attributes are who the participant is. Age, country, income, anything your screener, panel or uploaded data carried in. These describe the sample rather than teach you about it, and they are what most people filter by.

• One page, two views. Question coding looks and works the way it did. A single switch moves you between browsing all topics and browsing all attributes.

• A searchable outline on the left. The outline sits beside the charts with a search box, so you jump to the question you want instead of scrolling the study.

• Filters on the right. Filtering moved off the top bar to a panel on the right. Common filters are one click, saved filter sets still work, and you can still extend a filter with more detailed rules.

• The first time you open it, Conveo explains the change. A short explainer shows once and does not come back after you dismiss it.

Cross tabulation for qualitative answers

Pick a topic, choose breakdown by, and the coded answers split by the attribute you picked, for example household income. Switch the breakdown to a table and two dimensions stack against each other, which is cross tabulation in qualitative research done on coded video conversations rather than on survey rows. It works the same way in the topics view and in the attributes view, so you can run a cross tab analysis on what people said, not only on what they ticked.

Research teams asked for it in four situations

These come from conversations with clients and prospects. We kept them anonymous and name industries only.

• One study, several markets. Teams in consumer goods and technology asked to read each market on its own without building a separate study per country.

• Users against non-users. Teams in food and drink and in health split a sample by usage and wanted the two halves compared on the same questions.

• A recruit written in demographics. Teams that spec an age and gender mix asked to read the answers back by the same cuts they recruited to.

• A large study with no obvious entry point. Teams opening a study with hundreds of interviews asked for a faster way in than scrolling the whole analysis.

What this means for your research

The split puts the deductive coding work, the attributes you already know about people, next to the inductive coding work, the themes the conversations produced, and lets you cross one by the other in a couple of clicks. For a multi-market or multi-segment study that removes the export-to-spreadsheet step that used to sit between a finding and the proof of who held it.

Two limits to keep in mind. A breakdown can only use attributes your study actually collected, and a panel does not always supply every demographic field, so check the base before you read a split as a finding. And qualitative bases get small fast when you cut them: treat a segment difference as a direction to go and read the quotes underneath it, not as a number to report.

The naming change is on the surface. Everything you built before this release, including saved filters and favorites, still works.

Watch Filip walk through the new Question coding page

Filip Schouwenaars from our engineering team walks through the renamed views, the outline, the filter panel and the new breakdown. The update is live for every Conveo organization. The Question coding guide has the detail on bases, evidence and exports.