Crime mapping conference 2009
Download as PPT, PDF0 likes494 views
The document summarizes research into the most effective ways to present crime statistics and maps online. Researchers tested 6 different crime mapping websites from the UK and US with users. Eye tracking and think aloud sessions showed that tables provided the quickest access to data, while graphs connected statistics but took longer to extract information from. Maps were best for area selection, overviews, and comparisons when color coded meaningfully. The research recommends connecting maps and data, usability testing before publishing, and providing context to aid understanding of statistics.
![Questions? Talke Hoppmann, User experience consultant [email_address] Ian Haynes, E-learning director [email_address]](https://image.slidesharecdn.com/cimexncmc2009-100520055528-phpapp02/85/Crime-mapping-conference-2009-24-320.jpg)
Crime mapping conference 2009
- 1. Presenting crime maps online: effectiveness, usability & user preferences Exploring the use of tables, maps and graphs for statistical data presentation on the Internet Talke K. Hoppmann, PhD cand., UX consultant Dr. Katerina Tzanidou, Head of User experience Niki Economidou, UX researcher
- 3. Why user research? Home Secretary’s Policing Green Paper, Jul ‘08 ‘ Summary of responses and next steps’, Nov ‘08 require police forces to deliver crime rates and present information more effectively to engage with the public = Crime maps go public … to find out which form of data display is most effective and to understand problems users face
- 4. User experience research How ‘real users’ interact with the system In-house usability labs Using video equipment, eye-tracking device & recording software Testing on websites or clickable wireframes Implementing findings into IA and design
- 6. Project overview Examine crime statistics in tables, graphs & maps 6 Examples - Beatcrime, West Yorkshire - MyNeighbourhood, West Midlands - GMP, Greater Manchester - Crime mapping, London - Everyblock, Chicago - UpMyStreet Regional, UK National, UK Regional, US
- 7. Research questions Which elements of the sites are (not) working well? What are problems or barriers? On which sites do users perform best/worst? How do they rate the design? What are user interests and possible applications?
- 8. Method 6 x 90 minute one-on-one sessions (3 male, 3 female) Comparing sites & user preferences Design ratings Purpose Method Alternative applications, comments & feedback Post-session interview Search scenario (moving to a new area) for interacting with the site, extracting information Think-aloud interaction Search task for testing performance Eye-tracking Knowledge of online- and crime maps Pre-session interview Search scenario (moving to a new area) for interacting with the site, extracting information Think-aloud interaction Search task for testing performance Eye-tracking
- 9. Eye-tracking findings Examples of different search processes
- 10. Eye-tracking task - total crime in Camden (2006-2007) ‘Crime Mapping’ website (MET, London) Key findings All cases show a strong focus on the pop-up window & graph Patterns Even though data are right there, extracting information takes time Graph seems to be unclear If pop-up doesn’t provide information, users start scanning left and right Timings One of the longest search processes (10-27 sec compared to ~14 sec on average) 1. Strong focus on the graph 2. Extracting information takes long
- 11. Eye-tracking task - total crime level in East Salford, 2008 ‘GMP’ website (Greater Manchester) Key findings First focus on map, then straight to the graphs Highly directed search for information Patterns Most examined graphs left to right Due to information overload, users wanted to verify and check their answers Extracting information took very long Timings Longest search process (13-35 sec) 2 nd focus on graphs 1 st focus on map
- 12. Eye-tracking task – crime rate in South Chicago ‘Everyblock’ website (Chicago) Key findings Gaze plots show two search patterns Search length connected to order (the later, the longer and more varied) Patterns Short searches = merely guessing Long searches = users try to verify information in the map Timings Broadest range of search processes (from 1.6 sec to 45.9 sec) Strong focus Global scanning
- 13. Eye-tracking task – total no. of robberies in Nottingham, 2007/8 ‘UpMyStreet’ website (National) Key findings Good user performance Very clear and easy Patterns Quick to access the information Tables seem to be working best for displaying statistical data Timings Overall shortest search process (ranging from 3.7- 9.2) Note: Ease of use also due to task (find the total crime rate) Immediate focus on the table data
- 14. User performance across sites Site users spent most time on overall Site users spent least time on
- 15. Think aloud findings Examples of different search processes
- 16. Think Aloud Task – no. of burglaries in Bradford South Division ‘Beatcrime’ website (West Yorkshire) Key findings Design = negative, use = positive Data presentation Map is not interactive Table is simple and (relatively) clear Lack of professionalism = impact on trust Time period & crime type are clear Other Clear pull-down menu, but ‘invisible’ radio-buttons The later, the more positively evaluated
- 17. Key findings Top navigation not visible enough Combination both positive & confusing Data presentation Problems of consistency & labelling (e.g. NEXT button) Default search set to postcode Map = labels/tool tips missing, lack of meaningful colour coding Table = time period & crime type clear Other Users did usually not find the data table Think Aloud Task – residential burglaries in Dudley Town ‘MyNeighbourhood’ website (West Midlands)
- 18. Key findings Good and interactive experience Meaningful colour coding Data presentation Higher level of detail needed (“What does average refer to?”) Better comparison options expected Other Pull-down menu easily understood (not always easily applied however!) Most technical difficulties Clearer divides between areas Think Aloud Task – residential burglaries in Hackney (2006/7) ‘ Crime Mapping’ website (London)
- 19. Key findings Lack of consistency Colour coding without meaning Data presentation Bad use of space – map too far from statistics Area selection by map = easy to use Time period unclear, too difficult and overloaded Takes too long to find information Other Table view often preferred to graphs Think Aloud Task – no. of burglaries in Bolton Central ‘GMP’ website (Greater Manchester)
- 20. Key findings Quickest search process Data presentation Easily accessible information Clear, clean, straightforward Good – comparison to average Lacking more detailed information Other Further information links – no further information Map preferred for comparison, but only when meaningful Think Aloud Task – no. of burglaries in Bristol ‘UpMyStreet’ website (National)
- 21. So far… None of the online crime maps or statistics gets it completely right
- 22. Findings Tables provided the quickest & easiest data access Graphs immediately connected to statistics and can contain in-depth information, but may take too long to extract data Maps serve three main purposes 1. for area selection (purely functional) 2. for getting an overview 3. for comparing areas/crime rates
- 23. Recommendations 4. Provide points of reference users are familiar with to further understanding of statistics 5. Interactivity enhances user experience but has to allow filtering information according to user needs 3. Connect maps & other data to allow users to verify the data and build trust in the source 2. Colour-code maps in a meaningful way (i.e. the darker the colour, the higher the crime rate) 1. Carefully consider which data to display in which form, some data are better provided in e.g. tables Conduct user research to find out about problems and barriers prior to publishing statistical data or maps online
- 24. Questions? Talke Hoppmann, User experience consultant [email_address] Ian Haynes, E-learning director [email_address]
Editor's Notes
- #4: Introduction This study was conducted in order to explore different ways of presenting crime maps and statistics on the Internet in order to find out which form of data display is most effective and understand the problems users face when trying to gather information about crime statistics online. The area of online crime mapping became highly relevant in 2008, due to ongoing police reforms and the Home Secretary’s release of the Policing Green Paper "From the neighbourhood to the national: policing our communities together" [1] , which outlines future directions for policing. The paper emphasises that the National Policing Improvement Agency (NPIA) will set up two programmes, which “will seek to create innovative information services for citizens and partners. […] and explore the options for how information can be used to more effectively engage with the public; for example […] investigate how the public could access more police information online. This work will be done in support of the neighbourhood policing programme.” (p.43) The reform process started with the Police Reform Act in 2002. The Policing Green Paper from July 2008 was followed by a three month consultation period which concluded with a ‘Summary of responses and next steps’ in November 2008 [2] . These next steps are now being implemented in different stages, with various milestones in the following two years. With regard to crime mapping online, these policy papers and changes in legislation play a vital role as they require police forces to deliver local crime rates and present this information more effectively to engage with the public. Chapter one of the summary report, which outlines the practical steps taken to improve the connection between the public and the police, stresses that “Local people must get as much information as possible, including ‘crime maps’, regular updates on local action taken and follow-up for victims and witnesses” (p.6). The milestone assigned to presenting crime rates online on a national scale was January 2009. By this date, the Policing Pledge, including crime mapping, is supposed to be in place everywhere in England and Wales. [1] Home Secretary (17 July 2008). From the neighbourhood to the national: policing our communities together . Available online at: http://files.homeoffice.gov.uk/police/policing_green_paper.pdf [2] Home Secretary (28 November 2008). From the neighbourhood to the national: policing our communities together. Summary of Green Paper consultation responses and next steps. Available online at: http://police.homeoffice.gov.uk/publications/police-reform/green-paper-responses?view=Binary Relevance In the context of this study, the essential points of the green paper are the use of “innovative information services” and providing information “to more effectively engage with the public”. According to the green paper, the most effective ways of delivering this kind of information to the public need to be explored so as to make the data engaging and foster the connection between the public and the police.
- #7: Relevance To this end user testing was carried out on six different crime map websites to examine user preferences and uncover possible problems and barriers users might face in gathering crime statistics online. These websites represented a broad range of data displays and covered different geographical areas. It was deemed essential to include both regional and national websites and to not only look at British websites but include a US American crime map website as well. Accordingly, four of these websites had a regional focus. These websites were: 1. Beatcrime, West Yorkshire Police Authority (http://www.beatcrime.info/), 2. MyNeighbourhood, West Midlands Police (http://www.myneighbourhood.info/myn2/html/home), 3. Greater Manchester Police (http://www.gmp.police.uk/live/nhoodv3.nsf/index.html?ReadForm), and 4. Crime Mapping, London Metropolitan Police (http://maps.met.police.uk/). The website examined for national crime figures was UpMyStreet (http://www.upmystreet.com/enter-location/local/police-crime/l/) and finally the website on US crime statistics, for the Chicago area was EveryBlock Chicago (http://chicago.everyblock.com/). Across these six example websites crime statistics were presented in the form of graphs, tables and maps so that it was possible to compare the user’s performance, preferences, and problems encountered in relation to each of these ways of displaying data.
- #8: Methodology Empirically user testing was employed to examine the most effective presentation of crime statistics online. The main research questions the user experience research sought to answer were: Which elements of the sites are (not) working well? What are problems or barriers? On which sites do users perform best/worst? How do they rate the design? What are user interests and possible applications?
- #9: Through the triangulation of interviews, eye-tracking and think-aloud tasks, it was possible to compare the findings of each stage of the five-tiered research process. Each user testing session was split into five main sections: 1. a pre-session interview, 2. eye-tracking tasks, 3. scenario-based think-aloud tasks, 4. comparing design to functionality, and 5. a post-session interview. The table below illustrates which method was employed for obtaining what kind of findings. Method Findings relating to… Pre-session interview Internet use, use & knowledge of online maps and crime maps Eye-tracking tasks Search tasks for testing user performance, i.e. time needed to identify relevant information Think-aloud tasks Search scenario driven tasks for interacting with the different sites to gather information, revealing user expectations, references, problems and discovering which elements work well and which do not Comparison of design & functionality Post-session interview Six participants took part in this qualitative study and interacted with the different sites in 90 minute research sessions.