Interactions allow users to modify a data visualization to meet their needs. Interactions allow users to zoom in or filter for data that interests them and can allow for presentation of more data than could be presented in a static visualization. Common types of interaction include selection (show me something), exploration (show me something else), abstraction/elaboration (show me more detail), reconfiguring (change the order), filtering (show me a subset), encoding (show me the same thing in a different way), and connecting (show me relationships). Interactions can be supported through direct manipulation or dynamic queries. - **Direct manipulation** allows users to choose the data directly from the visualization through zoom, **semantic zoom**, panning, mouseovers, lasso selection or clicking. - **Dynamic queries** allow users to manipulate the visualization using a search box, slider, sort button or dropdown. Semantic zoom is another term to describe drilling down into the data. For example, clicking a bar in a bar chart to reveal related subsets of the data is semantic zoom. ## interaction problems However, interactions can introduce problems. Interactions can result in hidden [[affordance|affordances]] if it isn't obvious how to use the interaction (no, users will not read the manual or instructions on your dashboard). To avoid this, make the interactions as obvious as possible. For example, a mouseover interaction is easier to discover than the same interaction requiring a click. Instead of an interaction tied to a legend element, explicitly create a dynamic query using a dropdown selector. Interactions also require consideration for the media used. Interactions may appear or perform differently on different media (e.g., tablet versus webpage) if not designed correctly. Interactions also require higher [[cognitive load]] for the users to extract meaningful insights. ## avoiding desert fog **Desert fog** is the term for getting lost in a data visualization after too much interaction. Where am I again? How do I get back? To overcome this, use the **overview+detail** technique or **focus+context** technique. In overview+detail, show a detail pane to reveal the results of any drill-down interaction while maintaining the full chart in a separate pane. This is also known as **multiple view coordination**, the technique of showing two or more visuals where action on one results in a change in the other (often called brushing and linking). **Brushing** is where the user selects a subset of the data to apply an interaction effect on another visualization through **linking**. In focus+context, retain the aggregated data while exploding or zooming in on the data the user wants to investigate further through interaction. The **Table lens** technique is an example, where specific rows or columns are zoomed in on while the rest of the table is squeezed to the margins. Outside of the context of a data table, the same effect would be called **fish-eye lens**. > [!Tip]- Additional Resources > - [Toward a Deeper Understanding of the Role of Interaction in Information Visualization](https://faculty.cc.gatech.edu/~stasko/papers/infovis07-interaction.pdf)