Map markers - centre them or put them "pinned" to the exact location?

Question

In the traditional "pin" approach to maps, the pin essentially "points" to the actual latitude/longitude. Alternatively, you could have a symmetric (e.g. circular) icon that is essentially centred on the latitude/longitude. Have there been any usability studies or guides for when to use which type?

Example map pin:

Example symmetric marker:

My question is whether to have markers "point" to their exact location (i.e. using a "pin"), or whether to centre them on top of that exact location (e.g. using a circular icon). In neither case am I referring to clustering, as I will not be clustering markers.

Answer 1

Most of the digital design we see today is a replica of real life objects. We started with skeuomorphic design and slowly transit to simpler flat design to cut away the design bloats. Before the digital drop pin, people were using real pins to mark location on the map.

There is no official statistics (not that I know of) on which location marker fare better than the other. Which marker type to use is entirely up to your design and implementation. That being said, people are more familiar with traditional pin as location markers for the reason I describe above. Choose this if you want a more conservative design approach.

Answer 2

The answer to your question depends on the purpose of your map. Use symbols that fit your need and data.

For example, use

• Pin (= marker symbol, or short marker http://www.designingmapinterfaces.com/patterns/marker/) when highlighting a single (you are here) or limited amount (route from - to) of locations
• Single symbol to draw many features in a layer with a common symbol like a blue dot or circle
• Unique values symbol to apply a different symbol to each category of features within the layer based on one or more fields (e.g. this location is a hospital vs fire station vs school etc.)
• Other variations are based on qualitative differences that can be shown by using a range of colors or quantitative values as shown through a series of unclassed, proportionally sized symbols

Clustering comes into play when you run into the problem of having too many points on the map and/or it's really not necessary to show them all but only hint to the number. This case is very common on smaller scales, e.g. 100k policies in NY and I don't want to show 100,000 points, right?