Cognitive Issues in Geographic Information Visualization

Perception and Thematic Map Symbol Design I

3708.2 - Distortions and perception of different population density maps of Lithuania

Monday, July 3
4:30 PM - 4:50 PM
Location: Maryland C

Introduction

Population density maps of the country are usually compiled at small scales. Population census or (and) registries data normally refer to address points. In order to prevent acquisition of sensitive personal information the data are aggregated by administrative units and, since 2012, by statistical grid cells. Still, in 2016, the most of population density maps of Lithuania used for education and for evaluation of the demographic situation show data aggregated by administrative units: 60 municipalities or, more rarely, by 546 subdistricts. Grid maps can be found in the portal of Statistics Lithuania, but it will take time for the public users to get used to this method. As the majority of users of such maps are not cartography professionals, it is important to assure that they interpret the maps correctly. That is sensitive issue in Lithuania, where depopulation has been gaining pace over the last decade and is among highest in Europe. The purpose of the pilot research performed in the Centre for Cartography was to evaluate the loss information conveyed by maps with higher aggregation levels and to understand how it impacts the knowledge of users.

The method

For the research we used population census data of 2011 provided by the Statistics Lihuania. This is the only census dataset publicly available in a form of statistical grid with maximal resolution of one square kilometer for the entire territory (100x100 meter for the urban areas). Three spatial aggregation levels were chosen for comparison: municipalities, subdistricts and one square kilometer size cells, from which we derived a kernel density map with the radius of 1 km and the neighbourhood search radius of 3 km. Kernel density map was suggested to use in the experiment because the users might have been confused by the uncommon visualization. One kernel density and three choropleth maps were compiled; urban areas excluded due to extreme density values, the rest of information classified using natural breaks into four classes: high density, moderate density, low density, no population. We considered this sufficient because previous pilot studies showed that the most of people clearly distinguish and memorize three classes while five can be too many. The research consisted of two parts.
1. Estimation of differences in population density class level between the sample map and the original grid map (objective evaluation).
2. Analysis of identification of density regions and of memorization of the situation by the aware but inexperienced users. The experiment group consisted of 35 first year students of Geography and Hydrometeorology. They were asked to indicate sparsely and densely populated areas on the given maps, then, without looking at the maps, to draw these areas on a base map.

The outcomes

Municipality level distorts information to the level that makes the maps useless. The distortion map explains poor general knowledge about Lithuanian population density. Even though the subdistricts may seem appropriate for aggregation, the maps are also largely wrong for the countryside areas. The choropleth maps are quickly perceived and memorized; therefore the harm is even bigger. The errors in kernel density map are tolerable thus it can be used instead of the grid map.
The maps differ to the extent that the users did not recognize the same data on different maps and stated that they represented data from different periods of time, sometimes with difference by centuries. There are territories that were memorized both as densely and sparsely populated; they largely overlap with the territories with highest deviation from the grid values.
The research revealed some unexpected issues related with interpretation of population maps that helped us to better understand the problem.

Giedrė Beconytė

Professor
Vilnius University

Professor at the Centre for Cartography, Vilnius University. Has published more than 40 papers in scientific journals as well as conference proceedings and textbooks on spatial analysis and DBMS. Current research interests include thematic mapping, geographic information system design and project management.

Presentation(s):

Send Email for Giedrė Beconytė

Amy L. Griffin

Senior Lecturer
UNSW Canberra

Dr. Amy Griffin is a Senior Lecturer in Geography at UNSW Canberra. She is currently a co-chair of the ICA Commission on Cognitive Issues in Geographic Information Visualization (CogVIS). Her research interests include investigating cognitive and perceptual processes involved in using maps, information visualizations and other forms of geospatial information.

Presentation(s):

Send Email for Amy Griffin


Assets

3708.2 - Distortions and perception of different population density maps of Lithuania



Attendees who have favorited this

Please enter your access key

The asset you are trying to access is locked. Please enter your access key to unlock.

Send Email for Distortions and perception of different population density maps of Lithuania