Arvi A Arbuis –Instructor (Week 5) horizontal relationship in which we recognize a

common factor across a particular crime type, such as

Subject: LEA103 graffiti in similar types of crime locations. Visual
thinking is a private activity involving exploration
and confirmation. In the exploratory phase, maps
Topic: Mapping Crime: Principle and may be crude and are not intended for display or
Practice publication. A computer-printed map of burglary
patterns for the most recent week might be marked
Chapter 2: What Crime Maps Do and with handwritten information provided by
investigators or with other data not in digital form.
How They Do It Information might be transcribed from a mental map
to a paper map. Another possibility is that the tools of
exploratory spatial data analysis (ESDA) are used to
What crime maps do. find anomalies in data, such as an unexpected cluster
of incidents, that could point to unexpected
Maps are often thought of solely as display tools. relationships.
Infact, maps have a wide-ranging role in the process
of research, analysis, and presentation. Mapping is At this stage the analyst may generate a formal
most effective when those broad capabilities are hypothesis, or educated guess, to explain the process
recognized and used to their fullest extent. The map producing the observed crime pattern. Did the
is the end product of a process that starts with the observed cluster of burglaries pop up by chance? Is
first-responding officer’s report that is processed by there some recognizable cause? Is a serial burglar
data entry personnel, entered into a database, and operating in the area? Do officers in the field have
transformed into a symbol on paper. In this narrow insight to offer? By developing a hypothesis, the
interpretation, a map is merely a picture or part of a analyst is in the mainstream of scientific research,
database. But maps can be useful in other ways. using a venerable methodology—the so-called (and
MacEachren (1994) and MacEachren and Taylor awkwardly called) hypothetic-deductive method.
(1994), following DiBiase (1990), noted the
distinction between visual thinking and visual Maps and other graphics are integral tools for
communication in the use of maps and graphics exploration and hypothesis testing. Do preliminary
(figure 2.1). maps confirm the hunch that a burglary pattern is
likely the product of a repeat offender who is using a
bus route, and apparently a specific bus stop, to visit
a neighborhood and commit his offenses? If so, the
preliminary information will help the hypothesis gel
into something useful.

At the core of this method is a potentially repetitive

process involving:

n Development of a hypothesis on the basis of the

best available information derived from both theory
and field data.

Visual thinking n Development of a method for testing the

hypothesis, perhaps involving statistical and graphic
In visual thinking, the map is used to generate ideas testing or modeling.
and hypotheses about the problem under
n Analysis of the data.
investigation. By inspecting a map, for example, we
may notice a relationship, or correlation, between
n Evaluation of the results.
environmental factors that otherwise might have gone
unnoticed. This correlation may be vertical in the
nA decision to accept or reject the original
sense that we see connections between different
phenomena, such as crimes, land uses, and hypothesis.
demographics. Alternatively, we may see a
n Reevaluation of the original hypothesis, if it was A background problem that goes largely
unsatisfactory. It may be modified to take into unrecognized in the community of mapmakers is that,
account new knowledge. If so, the process begins for some people, maps have no meaning. They may
anew. grasp neither scale nor symbolization. As a result,
they have no sense of distance, relative or absolute,
The confirmation stage tells the analyst whether the and are unable to draw meaningful conclusions from
hypothesis does indeed have a factual basis that will a map.
withstand scrutiny. If it does not, we reevaluate and
make necessary adjustments, perhaps gathering more This problem is, in part, a legacy of the
information to add depth to what is already known disappearance (until recently) of geography from
and to shore up the hypothesis, which itself may now school curriculums. But it may go deeper, seemingly
have been modified to take new data into account. having to do with gender- and race-specific
differences in personal mobility that, in turn, may
MacEachren (1994) cautions that investigators should hinder the development of spatial experience and
realize that maps and other graphics are prone to reduce individuals’ abilities to take advantage of
error resulting from their underlying data, maps as tools. For example, in the past, women’s
inappropriate design, or even the margin of error traditional roles in childrearing have limited their
introduced by the normal process of abstraction. If mobility, thus denying them opportunities to learn
possible, the analyst should not rely on any one data geography by directly experiencing places. Race has
source, whether it be a map, field observation, or had a similar indirect effect through the mechanisms
survey, if other sources can be used to complement of discrimination and depressed economic status.
each other. Insofar as minority groups have experienced
disproportionate levels of poverty, their mobility has
Visual communication been limited and their geographic learning
correspondingly stunted. (See Montello et al., 1999,
As we move from visual thinking to visual for a discussion of related questions.) While the
communication, we go from the private realm to the police are very geographically aware, in part due to
public activities of synthesis and presentation. much field experience, individual members of the
Synthesis implies merging various types of community may not be. An argument might be made
information—in this case, geographic information for giving special attention to maps intended for the
system (GIS) layers—into a coherent final product. community. For example, digital photos of landmarks
Although synthesizing is essentially scientific, human could be embedded in a community map as visual
judgment is at the core of this filtering and refining anchors to show residents how the map relates to
process. Synthesis is assisted by the ability to find their environment. All messages, including maps, are
overlaps (intersections) between layers in a GIS. But laced with nuance. “The medium is the massage,”
even then decisions have to be made about what to wrote McLuhan and Fiore (1967), arguing that
put in, what to leave out, and what importance to literate people had been rendered visually
attach to each layer. A presentation puts all the incompetent by an excessive dependence on text.
relevant pieces together in a map. The map can be Since that famous remark, personal computers have
highly persuasive if it provides information germane provided an interactive platform, allowing what is, in
to the question at hand and is well designed. As effect, environmental manipulation on the fly. Maps,
MacEachren (1994, p. 9) noted, “People believe text, and data have moved from the realm of the
maps.” passive to the active and interactive, encouraging
perception of the map as a tool rather than as a mere
display device.
How crime maps do what they do
Peterson (1995) has outlined several theories and
A detailed discussion of how maps communicate models that have been advanced to explain how
through processes of visual comprehension is beyond visual information is processed:
the scope of this guide. However, a few points are
made here to explain the underlying process and n Stage model. Visual information moves through
underscore the idea that people see maps differently three memory stages. The first (iconic) is very short
due to differences in, for example, their eyesight, and deals with initial recognition. The second (short-
aptitude for visual comprehension, and prior training. term visual store) is longer but has less capacity so
complexity becomes an issue. Moving from iconic to
short-term demands attention. The information is
then sent to long-term visual memory. Long-term
images provide cues to help with recognition of new
Chapter 1 characterized thematic maps as falling into
visual stimuli.
the following broad categories: statistical, point,
choropleth, isoline, surface, and linear. How do we
n Pattern recognition theory. Iconic images are
choose the most appropriate type for mapping crime
converted into something recognizable through
and crime-related phenomena? Some decisions jump
pattern matching.
out at us while others are open to interpretation.
For example, if we want to see the precise locations
n Computational model. This sophisticated three-
of burglaries for the last month, then we use a point
dimensional model is similar to the process of
map of addresses of incidents. Or perhaps a city
abstraction in cartography. (For additional discussion,
council member has asked the police department for
see Peterson, 1995, chapter 1.)
a map summarizing the number of incidents of
graffiti per structure by city neighborhoods. This
These theoretical considerations are reminders that
calls for a choropleth map, with neighborhood
producing a map is only half the story. We also have
boundaries making up the geographic units. Links
to be concerned with how it is interpreted by the
between victim and offender residences demand a
intended audience. The storage of cues for the
linear representation. A generalized picture of crime
interpretation of visual images in long-term memory
risk or incidents is seen best with an isoline or
means that familiarity provides a substantial
surface map, and census information depicting the
advantage in the interpretation of maps. We may be
relationship between poverty and race can be shown
oblivious to the fact that our map is extremely
using either a statistical or choropleth map.
familiar to us but means little or nothing to those who
have no reference points in their long-term memory
Because of the infinite potential combinations of
or who have had insufficient time to study and
crime-related conditions that can be depicted on
process the details.
maps, we can combine map types to put more
information on the same map. For example, we can
Another way of visualizing the process of moving a
combine nominal and ratio data, such as a choropleth
concept from the analyst to the map user is illustrated
map of drug-related crime by patrol beats and add the
in figure 2.2, showing that the cartographer’s and
locations of drug markets on the same map. Crime
map user’s realities are both abstractions of reality.
mappers should be aware of the potential for
The cartographer creates a cartographic abstraction
combining thematic map types, provided that the
and translates this into a map that is read by the map
result is not overloaded with information—or just
user and transferred
plain incomprehensible. An overloaded map will
to the user’s mind.
have so much information that the eye is unable to
take it all in. It will prevent the reader from
discriminating between what is important and what is
not.

Examples of thematic maps

Perhaps the best way to get a feel for the kinds of

maps used to display crime data is to look at
examples and to think about why each type of map
was selected. A good place to start is the Web site of
the National Institute of Justice Crime Mapping
Research Center (http://www.ojp.usdoj.gov/cmrc),
which provides links to police departments across the
United States. Another useful Web site is maintained
by Hunter College in New York
(http://everest.hunter.cuny.edu/capse/projects/nij/cri
me.html). (See the appendix for additional
information.)

Choosing a crime map

Thematic maps using point symbols: When there are too many points to be mapped, using
The dot map point data may result in a mess of superimposed
points that have little or no meaning. This could
happen if calls for service are mapped using
When should point symbols be used? The first addresses in a large city. The point data may need to
prerequisite is that you have locational detail be summarized by areas to make the data legible.
information specific to your points, such as street Point maps also get too crowded if long time periods
addresses or coordinates in latitude/longitude or some are summarized for more frequent crime categories.
other system, such as State Plane (explained in Thus, even though you have reasonably precise
chapter 1). The second prerequisite is that the locational information, aggregation by areas in the
audience needs locational detail. If you have point form of a choropleth map may yield a more legible
data, but the audience wants information summarized map than the presentation of each individual point.
by patrol areas or neighborhoods, then the point data
can be added up, or aggregated, to the areas of
interest. Examples of point, or dot, maps are shown Thematic maps using
in figures 2.3 and 2.4. statistical symbols

At its most primitive, a statistical map consists of raw

numbers written in the subdivisions of the map. The
advantage is that the reader can see exactly what the
statistic is. The downside is that maps designed in
this way are difficult to read quickly. It could be
argued that, in effect, they defeat the purpose of the
map, which is to facilitate visualization of the data.
Admittedly, this form of map does put data in its
geographic context, but in an inconvenient format.
Cartographers argue that if you want to see only the
raw numbers, then a table, not a map, is needed (see
next section, “Thematic maps using area symbols”).

Statistical symbols commonly take the form of pie

charts, bar charts, graduated circles, or dots
representing incident counts (dot density) placed in
the relevant map subdivisions (figure 2.5). This
allows multiple variables to be mapped at the same
time. Examples could include bar charts with bars
representing both crime and poverty or graduated
circles like those in figure 2.6, showing the U.S.
House of Representatives vote on an Omnibus Drug
Bill provision requiring a 7-day waiting period for
the purchase of handguns. At first glance, the
symbols in figure 2.6 look
 of the 90-degree segments is proportional to the
number of yes votes (top part of the circle) or no
votes (bottom part of the circle), with the left side of
the circle representing votes by Democrats, the right
by Republicans.1 The map shows both nominal data
(party affiliation and yes/no votes) and quantitative
data (the number of votes), as well as location of
votes by State. Although reading this map takes some
effort, it is rich in information and gives that
information a clear geographic context.

More typical graduated symbol maps used in crime

analysis applications are shown in figures 2.7, 2.8,
2.9, and 2.10. Note that points and proportional
circles can be

like pies, but circle segments are all 90 degrees. This

is actually a graduated circle map, in which the area
 If the size range is too small, readers will have
difficulty extracting meaning from the map. Also,
some symbols are more effective than others in
conveying the message. Solid symbols probably work
best in most cases because they engage the eye more
effectively.

A disadvantage of using statistical symbols on maps

is that they may overlap one another and result in an
illegible
mess. Map design must take into account the final
size of the map, the scale to be used, and the
possibility of overcrowding.

The use of statistical devices of various kinds on

maps is limited only by the analyst’s imagination. For
example, it may be useful to accompany a map with a
scatter diagram showing a collateral relationship,
such as calls for service by time of day (chart) and
calls for service by location (map). Mapping software
offers numerous possibilities because the programs
usually can make both charts and maps and combine
them in layouts in useful ways.

Thematic maps using area symbols

When we think of making maps that represent areas,

it’s the choropleth map that usually comes to mind,
with administrative or political areas shaded
according to their statistical values, whether they are
frequency counts, averages, or other relevant
measures. When is a choropleth map appropriate?
Strictly speaking, mapping total numbers, such as
crime counts, using choropleth mapping is
unacceptable owing to the misleading impression
given by unequal areas. For example, if the largest
and smallest areas have the same frequency, they will
be shaded the same on the map, which fosters
possible misinterpretations based on per capita or
density considerations. However, many departments
overcome this by using a regular grid for choropleth
mapping. This has the advantage of equalizing areas
but the possible disadvantage that the units of the grid
may not be “natural” local areas. In other words, the
areas exist only on police department maps and may
be difficult to interpret in the field.

For places with boundaries following the rectangular

land survey, a grid of square miles (sections) or
quarter-square miles will be substantially visible in
the local street pattern system. This applies to most
combined if this helps convey the essential areas west of the Appalachian Mountains. East of the
information to readers and avoids overloading the Appalachians, survey systems were usually based on
map. Note also that the size range for symbols is a irregular “metes and bounds” and do not lend
themselves to grid-based maps, except where cities
judgment call.
have a regular block grid. (See also chapter 4, crimes and the statistical effect is to inflate the crime
“Definition in geographic space.”) Examples of rate. Is this realistic?
choropleth maps are shown in figures 2.11 and 2.12.
One could take this scenario to the extreme: If small
Choropleth maps are best used for area averages, areas are used for crime rate calculation, we may
such as crime rates, population density, and have areas with crimes but no resident population,
percentages, as well as nominal-scale information producing an infinite crime rate! Bear in mind that
such as land use. Care is needed in the interpretation the crime rate concept is a loose one and crime rate
of all maps, and choropleth maps are no exception. maps are only an approximation, particularly for
Take, for instance, a choropleth smaller areas. Although neither the offenders nor the
victims are necessarily residents of the areas where
the crimes happen, they would nevertheless be
represented in crime rate maps.

Experience has shown that it may be helpful to add a

disclaimer note on some maps to point out, where
appropriate, that some areas with high crime rates
have small or zero residential populations. Another
form of this disclaimer would note that rates have not
been calculated, or have been intentionally omitted,
for areas with small or no residential population. This
will have to be handled carefully, however, since
some readers may conclude that the police
department has something to hide if data are
manipulated in what appears to be a selective way.

Note that choropleth maps can be produced in a

three-dimensional format with the height of each area
proportional to its data value.2 The advantage of this
type of presentation is visual appeal and vividness.
The disadvantages are that it can be difficult to
decipher the actual data value, and that a tall column
will hide other areas, as is the case in three-
dimensional surface maps.

Thematic maps using surfaces

Although crime maps using isocline’s have been

around in some form since the 1960s, they have only
recently become widely used.3 This is a result of the
availability of algorithms in mapping programs that
perform complex calculations at high speed. With the
addition of three dimensional capabilities, surface
maps with textured surfaces are now within the reach
of most crime analysts. Such maps are tempting as
they are visually more appealing than two
dimensional renditions. But the same caveats noted
above apply. Just because a surface map or three-
dimensional rendition can be produced, it does not
necessarily follow that it is the most appropriate or
map of a crime rate based on population. Map values
useful form for visualizing the data. For example, it
are expressed in terms of numbers of crimes per
may be difficult to add legible landmark icons or
population unit. But what population? Normally, the
even boundaries to a three dimensional map,
residential population as enumerated in the census is
depending on factors such as scale and amplitude, or
implied. What about downtown business areas with
the degree of pawedness, of the map, as well as the
negligible residential populations? They still have
angle of view. (See figure 1.14 and chapters 4 and 6
for other examples of three dimensional maps.) n Routes between victims’ and offenders’ home
Ultimately, deciding whether to use a surface map addresses (e.g., Pyle et al., 1974; Frisbie et al., 1977,
involves balancing scientific and artistic judgment, p. 88).
and in many cases the decision can be made only
through experimentation. Fortunately, maps can be n Passes along streets by patrol cars to illustrate patrol
produced rapidly with desktop computers, so density.
experimentation can and should be a routine part of
the mapmaking process.
n Traffic density.
Thematic maps using linear symbols
Virtually no flow maps have been seen in the recent
Flows between points are shown with linear symbols, literature on crime mapping, even though such maps
with their width or thickness generally proportional are in use in police departments. Their absence is not
to the volume of the flow (figure 2.13). Maps of this due to lack of data. It may be that the apparent lack of
type had their origins in economic geography, first this type of map in the literature is due to the absence
showing passenger flows on Irish railways in the of readily available flow-mapping algorithms in the
1830s and, later, commodity flows among nations GIS programs most popular among crime mappers.
(Campbell, 1993, p. 264). These maps can be used in While it is unlikely that flow mapping will ever be a
crime mapping, to show, for example: major component of crime maps, it will be
increasingly used owing to its obvious utility in
n Links between where vehicles were stolen, where limited applications.
they were recovered, and suspects’ addresses.
Classifying map Information where data are typically sorted into categories as part
of normal processing. But the situation becomes
more intricate when moving from nominal and
Generally, information on maps is classified in some ordinal data to ratio-type data.
way; data are not symbolized individually. For
example, all burglaries are shown with the same It is less obvious how to classify numerical data when
symbol on a point map. It would be absurd to show several alternatives present themselves. Common
each crime with its own symbol. In effect, maps mapping software packages offer options, including a
contain two levels of abstraction: default, for grouping numerical data in thematic maps
but rarely explain how to choose among these
n The overall level of detail and the scale used to approaches. Dividing up the data range in a way that
present the data. best represents it involves the abstraction issue again.
n The way data are symbolized, because there is a Total abstraction would be represented by the use of
continuum from highly detailed to extremely one shade for all areas on the map. This says that
generalized in the symbolization process. there are data, but little or no specific information is
supplied about them. At the other extreme, each area
To some extent, the choice of scale controls the level would have its own shade, and if city blocks were
of abstraction of the content because it is impractical shown, the map would have thousands of shades.
to load a smallscale (large area) map with local Obviously, neither of these alternatives is useful, and
detail. MacEachren (1994, p. 41) argues that for the solution lies somewhere between.
categorical information, “features that end up in the
same category should be more similar to one another Greater accuracy dictates the use of more classes of
than features in different categories.” data, although readers pay a price for this in terms of
comprehension as the map moves along the
What does this mean for crime maps? A map of drug continuum of abstraction toward reality and
offenses might group relateddrug categories together. complexity. The underlying question is, What is this
Generically related robberies could be put together in map being used for? MacEachren (1994, pp. 42–43)
the same category and symbolized the same way on a suggests that if we are in the visual thinking stages of
general crime or violent crime map. If a map were exploration and confirmation, we will need more
specific to robberies, however, symbolization might detail (more classes), but as we progress toward
be separated into commercial and street or weapon synthesis and presentation it becomes more important
type or time of day. This type of adjustment is to show general trends rather than detail, hence fewer
intuitive and naturally occurs in the crime context classes. Furthermore, limitations on human visual
comprehension must also be taken into account—the
limit is about six levels of color or gray scale shading n Equal ranges or intervals. The data range
in the context of a map. (difference between maximum and minimum) is
calculated and divided into equal increments so that
Are there natural breakpoints in crime data? For the within-class ranges are the same, such as 1–3, 4–
example, in a robbery map of a city we could embed 6, 7–9, and so on.
the State, regional, and national robbery rates as
breakpoints. This might be informative but could get n Equal count (quantiles). Approximately the same
a political “thumbs down” if the local jurisdiction number of observations is put in each class. The
compares unfavorably. (Conversely, it could be a number of classes determines the technical definition
popular approach.) Choices available to of the map (quartile if there are four classes, quintile
cartographers in common desktop mapping packages if there are five classes, and so forth). The term
are represented by the drop-down menus shown in quantile is the generic label for data with
figure 2.14. observations divided into equal groups. This software
option gives the user the opportunity to enter the
The choices available, and the relative ease of using number of classes desired. (This is the default in
them, invite experimentation. How will a particular MapInfo®.)
database look when mapped in a particular way?
What method conveys the crucial information with n Equal area. Breakpoints between classes are based
the least distortion and best visual impact? Good on equality of area rather than equality of range or an
maps are likely to result from a working environment observation count. If areas in a choropleth map vary
that encourages experiment because it is ultimately greatly in size, this type of map will differ from an
through trial and error that most learning is done. equal count map based on the same data. If areas are
This is said not to invite a “shotgun” approach but, roughly equal in size (such as city blocks), the result
rather, to encourage the responsible testing of options will be similar to an equal count presentation.
under the assumption that alternative methods of
representation are tested for a reason other than the N Natural breaks. In this approach, gaps or
sake of doing something different. depressions in the frequency distribution are used to
establish boundaries between classes. This is the
default in ArcView®, which employs a procedure
know as Jenks’ Optimization that ensures the internal
homogeneity within classes while maintaining the
heterogeneity among the classes. (For more details,
see Dent, 1990, pp. 163–165, and Slocum, 1999,
chapter 4.)

n Standard deviation (SD). SD is a statistical

measure of the spread of data around the mean, or
average. In the literature of stocks and mutual funds,
for example, SD is often used as a risk index, since it
expresses the amount of price fluctuation over time.
In the context of crime, SD can be a useful way of
expressing extreme values of crime occurrence or
portraying various social indicators. Generally,
classes are defined above and below the average in
units of 1 SD. The drawback is that this method
assumes an underlying normal distribution, or bell-
shaped curve, something of a rarity in social data.

n Custom. As the label suggests, this option allows

users to determine class intervals according to their
own criteria, such as regional or national norms and
thresholds determined for policy reasons. Table 2.1
Each of the alternatives typically employed in data summarizes the criteria for selecting methods to
mapping is introduced here and illustrated in figures define class intervals for maps, providing a guide
2.15 and 2.16. with respect to data distribution, ease of
understanding, ease of computation, and other and the infinite set of display permutations, whether
standards. (For a comprehensive discussion of issues in raw form or as some derivative measure such as a
relating to the determination of class intervals for rate or percentage.
maps, see Slocum, 1999, chapter 4.)
Mapmakers can gain a preliminary understanding of
what the numbers mean through the process of
exploratory spatial data analysis. It is quite helpful to
understand what the distribution of a set of numbers
looks like when expressed graphically. Is this a
normal (bell-shaped) distribution with most
observations clustering around the mean, or average,
and a few very low and a few very high values? Is it a
skewed distribution with extreme values to the right
(high values) or the left (low values)?

In the unlikely event of a normal, symmetrical, bell-

shaped distribution, maps created by all of the
classing methods look similar. Almost always,
however, real world data are somewhat skewed, and
different classing methods produce maps that look
different and convey different visual impressions to
readers. Consider in some detail what will happen
when different methods are applied to a data set that
has a strong positive skew (figures 2.16 and 2.17).53

Let’s review each histogram (or frequency curve) and

Maps and statistics: Exploratory map in figures 2.15 and 2.16, method by method.
spatial data analysis
n Equal count. On the histogram, the right tail
Some statistical methods have been mentioned in the (highest values of the distribution) is prominent
preceding discussion, and consideration of statistical because there are few extremely high values. Thus,
concepts is unavoidable when considering how best the program has to seek the lower rank ordered data
to visualize numerical data. As noted earlier, because values (farther to the left on the histogram) to come
we can lie with statistics, we can also lie with up with the 13 observations for the class. (Note that
statistical maps. Indeed, maps have been used the number of observations per class is uneven,
throughout history as propaganda tools (Campbell, ranging from 13 to 17.) The resulting map tends to
1993, pp. 229–235), so potentially we can have visually exaggerate the seriousness of the problem
honest error as well as pure cartographic deceit. because color saturates more map areas.
Perhaps the greatest danger in the mapmaking
process is that people tend to believe the information n Equal range. Because the distribution is right-
in maps (what MacEachren, 1995, p. 337) called the skewed, equal range will tend to favor lower data
connotation of veracity), and they also believe that values. The two lowest classes have 23 and 26
maps are unbiased (the connotation of integrity). members, respectively, while the higher classes have
7 and 3. The map contrasts with the equal count
But mapmakers, like other elements of society, are version, now visually minimizing the problem.
culturally conditioned, selectively including and
excluding data according to the values of the n Natural break. This method appears to have
responsible parties. Given that maps can harbor many struggled to come up with natural breaks, which is a
possible errors and biases, both intentional and problem, along with breaks in awkward places. The
accidental, it is incumbent on the crime analyst to be result here is quite similar to the equal range
aware of possible sources of error and to work to breakdown, with cuts between classes shifted to the
avoid them. Nowhere is there more scope for left (lower values) as compared with the equal range.
distortion and misinterpretation than in the It comes as no surprise that the equal range and
preparation of maps based on numerical data. This is natural break maps are quite similar.
due to the potential complexity of the information
n Standard deviation. Here, the breakdown of class
intervals is set with reference to the average, or mean,
so that an interval of 1 SD is established to the left of
(below) the mean (blue line, 0.47), and above the
mean at the same distance. The effect of this on the
right-skewed distribution is a symmetrical
breakdown, with about as many observations in the
lowest (10) and highest (9) classes and in the two
middle classes (22 and 18). The visual impression
conveyed by the associated map is close to the
severity of the equal count method. This is due to the
similar number of observations in the top category.

The basic point to be made from this discussion is

that cases that may fall in a given class by one
method may be in a different class by another. The
only certainty is that the highest and lowest values
will always be in the top and bottom map classes,
respectively. What method is preferable?
MacEachren (1994, p. 47) noted that, “for any
skewed data, quantiles are a disaster for a
presentation map!” In the above example, quantiles
result in such a large data range in the highest class as
to be almost meaningless. Standard deviation classes
may be helpful in some situations where the only a map can show skewness in geographic space.
distribution is not extremely skewed. Are the high values distributed geographically in a
random way or clustered? Either method yields
Note that a frequency curve shows skewness in the useful information. If the high crime rates are
rank-ordered data values, but clustered, it may indicate a hot spot. If the high rates
are random, the net impact on the community may be
about the same, but we are now unable to point to a
hot spot.

We can see an empirical relationship between map

scale and skewness, which is minimized in a small
area (large scale) and maximized in a large area
(small scale). Think of it this way: A very small area
in the community, say 1 square yard, can have no
spatial skewness because only one event can happen
there. But as the spatial scope increases (smaller scale
maps covering larger areas), the potential for
skewness increases because there can be bimodal, or
split, distributions in space (as well as time). A clump
of events can occur in one small area with the rest
empty—an extremely skewed pattern.

This is what the crime scene is like on a regional,

national, or global scale. Clusters correspond to
opportunities presented by the underlying controlling
condition, population distribution. At the smallest
scale (region or world), the crime map is for all
practical purposes the same as the population map,
but at larger scales (city or neighborhood) we refine
the view and see that the presence of people actually
means variations in rates conforming to varied social
and physical environmental conditions. Also, at
larger scales we will see different patterns depending accurately portrays the data? Again, this examination
on the denominators used to calculate crime rates. of the data is the ideal. Not all analysts will have the
tools or the time to go through this step.
Another way to visualize a distribution is the use of a Nevertheless, these possibilities are outlined here to
box plot, which shows how data are spread in raise awareness of what constitutes the best practice.
relationship to the mean, median, mode, and
quartiles, with outliers symbolized in a special way. Map design
(Outliers are values more than 1.5 box lengths from
either the 25th or 75th percentiles.) If we examine the The field of map design has generated substantial
HOMRATE data set using the box plot routine in the literature in cartography focusing on how people
Statistical Package for the Social Sciences (SPSS), comprehend maps and the impact of various design
the result appears as shown in figure 2.17.5 Note that elements such as symbol size, color, and line
the box plot is an alternative way of visualizing the thickness. Also of interest is the impact of the
same data shown in figures 2.15 and 2.16. In the box arrangement of a map within the map frame as well
plot in figure 2.17, the red box represents 50 percent as the merits and demerits of various types of maps.
of the data values, with the median shown by the bold
line across the box. The 75th and 25th percentiles are Debate continues, for example, over the dominance
the top and bottom of the box, respectively. The ends of choropleth maps to represent numbers. Opponents
of the Ts represent the smallest and largest observed point to the most obvious choropleth defect: its use of
values that are not defined as outliers. Although the one data value to represent an entire area, an
box plot seems to be repetitive, it provides a different absurdity that becomes acute when most of a
perspective on the data— one that complements the geographic subdivision contains virtually no human
more frequently used histogram. (For a detailed activity because of the terrain or the existence of
explanation, see SPSS documentation, such as SPSS bodies of water.6 Muehrcke (1996) quoted Ronald
for Windows Base System User’s Guide Release 6.0, Abler, Executive Director of the Association of
p. 186.) American Geographers, who, in 1987, said something
to the effect that choropleth maps were an
Only the immediate objective and the available tools abomination that GIS would soon eliminate through
limit the amount of exploration and preprocessing the use of dasymetric mapping.7 (The death of the
crime mappers do. Perhaps the single most important choropleth map has been slow!) However, the
exploratory step is the creation of a histogram, box increased use of density surfaces in various desktop
plot, or comparable graphic with which to visualize GIS programs is a move away from such heavy
the shape of the data and answer the fundamental reliance on either dot or choropleth modes of
question: Is it severely skewed or in some other way representation and is consistent with the concept of
not normal (e.g., bimodal or double peaked)? How areal averaging—without slavish adherence to
will this affect maps, and what type of map will political or administrative boundaries.
permit a presentation that minimizes distortion and
Map design is at once a technical and an artistic
effort. Dent (1990) devotes 5 chapters and about 120
text pages to designing thematic maps; we can give
only broad consideration to a few issues relating to
typical parts of a map and how they should be
organized.

The reader is referred to the following textbooks for a

full explanation, particularly on such details as
lettering and labeling: Campbell, 1993; Dent, 1990,
1993; Robinson et al., 1995; and MacEachren, 1994,
1995. It may be helpful for newcomers to mapmaking
to make a flow chart in which the design of the map
is adjusted to ensure that the map fulfills its stated
purpose. This activity may become less important
 n Graticule (spherical coordinate system:
latitude/longitude, State plane).

n Borders and neatlines (the lines that bound the body

of a map, usually parallels and meridians).

n Symbols.

n Labels.

Most of these elements are necessary in a typical

crime map. The principal exception is the graticule
(spherical coordinate grid), which normally serves no
useful purpose.8 Also, credits are rarely used because
data are likely to be locally derived. However, if data
sources are not self-explanatory, credits clarify
exactly where the data came from. This information
could be listed under the body of the map using the
keywords “source” or “data source.” We could add
author to the list to assist in the process of
accountability, inconspicuously noting the name or
initials of the analyst cartographer in a corner of the
map.

A useful approach to learning more about design is to

look at examples of crime (and other) maps that have
been deemed acceptable by their respective
audiences. Fortunately, there is no shortage of maps,
whether on crime or on other phenomena. The easiest
access is via the World Wide Web, and the appendix
of this guide lists some useful Web site addresses.
Chapter 3 uses examples to discuss various
applications of crime mapping. Here we confine
ourselves to outlining principles.

Dent (1990, chapter 13) has noted several elements of

map composition: balance, focus of attention, and
internal organization.
over time as intuition and experience take over from
n Balance refers to the need to arrange parts of the
reliance on a formally structured process. Dent (1990,
map in a way that enhances its visual symmetry.
p. 316) lists the following elements of the thematic
However, the crime cartographer may have little
map, which could serve as a checklist for inclusion:
flexibility with respect to balance owing to inherent
content limitations. For example, the jurisdiction may
n Title (or caption).
be extremely asymmetrical, making it difficult, if not
impossible, to map without leaving considerable
n Legend.
white space on the paper. Cities with long
“shoestring” annexations, like Los Angeles, or States
n Scale.
with long panhandles, like Oklahoma, are good
examples of difficult map shapes. This problem
n Credits.
sometimes can be solved by chopping the city or
other area of interest into its component parts. An
n Geographic content (showing information that may
inset, or miniature map of the whole, is used to show
not necessarily be included in the subject matter of
how the pieces fit back together. Another solution is
the map, such as orientation or north arrow).
to routinely map individual precincts or districts
under the assumption that the managers of those
areas are first and foremost interested in seeing derived from them.
patterns in their areas of responsibility. The drawback
to this is that crime patterns do not pay much n Value refers to the degree of lightness or darkness
attention to administrative or political boundaries, so of a color. GIS programs can help you select color
that looking at individual subdivisions in isolation values by providing color “ramps” (or series of
from the rest of the area may cause someone to miss related shades or values of a hue) in a visually logical
hot spots or other useful patterns by fragmenting sequence ranging along an intensity spectrum. Colors
them. vary along a continuum from light to dark. For
example, reds may range from light pink to deep red,
n Focus of attention is a concept based on the and blues may range from sky blue to navy blue.
assumption that people read maps like they read the
printed page, by moving their attention from upper n Chroma is understood through the concept of color
left to lower right.9 Hence the optical center of a map saturation. A less saturated color appears to contain
is somewhat above the geometric center suggesting more grays, and a saturated color has no gray and
that, ideally, the most significant information should appears as the “pure” color. In photography, some
be closer to the optical center. Again, this is easier to films have a reputation for conveying more saturation
manage in theory than in practice. Still, it is a useful than appears in natural scenes (bluer blues, greener
concept to bear in mind because crime analysts will greens), “larger than life” color that is pleasing to
sometimes have enough discretion in design that the some viewers but excessive to others.
focus of attention can be exploited to advantage.
Choices of color in maps need to be made quite
n Internal organization refers to the alignment of the carefully because color may have strong emotional
parts of a map or individual maps on a page or within connotations for some readers. For example, should
a frame. Map elements should be arranged in a red be used for a map of violent crime, given the
logical way rather than placed haphazardly on the symbolic connection to blood? It is tempting to
page. The core contents of the map, for example, overload crime maps with warm colors, such as red
should dominate the space, and other elements should and orange, but the analyst should be mindful of the
be secondary. According to Dent, contrast also is symbolic effect and the impact this may have on the
important to visual perception. Line, texture, value, intended audience.
detail, and color are powerful tools because they
allow map elements to be differentiated from one Just as color makes maps and other graphics come
another. More contrast makes objects stand out, less alive, color also enhances our ability to mislead
allows them to fade into the background. Line people with maps through the use of inappropriate
thickness, or weight, can assist in this process, and hues and values. For example, a crime category that
using more than one line weight on the map can add is a local political “hot potato” could be visually
interest. Texture can add variety and draw attention to minimized through the use of cool colors in subtle
an important part of the map. Value refers to the use shades lacking saturation. The use of color in maps
of lighter or darker shades of color, and detail draws and graphics is complicated by the fact that a
the eye in. As noted elsewhere, however, detail is a significant portion (8 percent of males and 0.5
two-sided coin. It adds interest, but when used to percent of females) of the population is at least
excess it can cause clutter and make the map partially colorblind.
illegible. If a map is to be reduced for publication,
fine detail may be completely lost in the reduction Crime mappers can take advantage of various models
process. Experiment with enlarging and reducing on of color sequencing. GIS software typically defaults
a photocopier to learn more about how this works in to a part-spectral plan with shades from yellow to
practice. brown. In a full-spectral plan, colors range from
warm to cool, and in a double-ended plan, data values
Color is extremely important in the process of area representing an increase (or above average) are in
differentiation. It is also a complex issue owing to the one color and a decrease (or below average) in
physiological, psychological, and physical processes another. Increases (or higher values) are typically
involved. Dent (1990, chapter 16) notes that color has shown in warmer colors, decreases (or lower values)
three dimensions: hue, value, and chroma. in cooler (see Dent, 1990, p. 387). GIS software
normally permits the customizing of colors to fit your
nHue is the term given to color labels— red, yellow, purpose.
and blue, the primary colors—and the millions of
permutations
Design, abstraction, and legibility lettering when it is projected, your creativity is
wasted.10Also consider the “demographics” of the
audience to be addressed: are they younger? older?
Map design and abstraction are inseparable. The map more educated? Less educated? predominantly
design defines the level of abstraction to be imposed. female? If the audience is not similar to the general
“To represent is to abstract,” wrote Muehrcke, and population, some adjustments in map design may be
“abstraction frees us from the tyranny of our physical needed. Research has shown, for example, that there
existence” (1996, p. 275). He presumably meant that may be subtle differences in the way men and women
it gives us license to, so to speak, “mess with reality.” read maps (Kumler and Buttenfield, 1996). This begs
Many of the issues that concern cartographers, such the corollary question of exaggeration in maps to
as the degree of distortion on world maps, are of little gain legibility. Sometimes detail must be retained,
concern to crime mappers. Where should we focus but this may result in objects running together owing
our attention when it comes to thinking about to the thickness of the lines representing them. Line
abstraction in our maps? What can we afford to thickness may need to be adjusted and objects may
ignore? Are any map elements indispensable on most need to be moved slightly to maintain visual
crime maps? separation. Bear in mind that line work on maps often
greatly exaggerates the true dimensions of linear
Abstraction, the reduction of detail on maps, permits features. A typical State highway map may be used
us to design our maps in ways that make them as an example. On this map, interstate highways are
attractive and effective. Abstraction is like the 1/16th of an inch wide. The representative fraction
sculptor’s chisel—it determines what remains of the (RF) of the map is 1:380,160, or 6 miles to the inch,
raw material and what form the finished product will which represents a width of 660 yards. This is
take. As noted earlier, most map elements are probably, on average, double the width of most
dispensable at one time or another, depending on the interstate highways. By comparison, area features
context. First and foremost, the analyst must consider such as a city block, a city, or a county, should be
the audience and the medium of presentation. Will accurately rendered because exaggeration is not
the map go to one person and be seen at arm’s needed to make them visible. Even point data
length? Will it be a page in a report? (If so, will it be generally exaggerate the size of the location at which
in color? How would the map look if it was a crime incident occurred or the address of the victim
converted to gray scale?) Will it become a or offender. Point symbols are actually markers for
transparency for an overhead projector or a 35-mm general locations and should be interpreted as
slide? Will it be incorporated into a digital projector approximations owing to (a) the size of the point
production in Microsoft PowerPoint® or comparable symbol and (b) normal problems with address
software? If a map is to be projected, lettering size interpolation touched on in chapter 4 of this guide. It
and line weight become quite critical. You may have is tempting to see point data as the epitome of
a brilliant map with potentially great visual impact, accuracy, but this accuracy is relative.
but if two-thirds of the audience can’t read the