This section explains some common style elements and produces most of the graphs displayed in the section Style Comparisons.

The DEFAULT style is the parent for the styles used for statistical graphics work. You can see all of the elements of the DEFAULT style by running the following step:

proc template;
   source styles.default;
run;

The source listing of the definition of the DEFAULT style is hundreds of lines long. If you run PROC TEMPLATE with the SOURCE statement for most other styles, you see parent = styles.default (or in the case of the HTMLBLUE style, you see parent = styles.statistical, which inherits from the DEFAULT style), and you do not see all of the elements in the style unless you also run the preceding step with a SOURCE statement for all parent styles.

Only a few of the style elements are referenced in the templates that the SAS System provides for statistical procedures. The most commonly used style elements, along with the defaults for the noncolor attributes of the DEFAULT style, are shown next (Color applies to filled areas, and ContrastColor applies to markers and lines):

You refer to these elements in graph templates as style-element or as style-element:attribute (for example GraphDataDefault:ContrastColor). The default values are not shown for the color attributes since they are typically defined indirectly. For example, Graph:BackgroundColor (the color that fills the box outside the graph) is defined elsewhere in the style as colors(’docbg’). The style also defines ’docbg’ = color_list(’bgA’) and ’bgA’ = cxE0E0E0. This shows that the background is a shade of gray that is much closer to white (CXFFFFFF) than to black (CX000000). You can see the background color in Figure 21.27. This shade of gray might seem darker (closer to CX000000) than you might expect based on just the RGB values. Your perception of a color change is not a linear function of the change in RGB values.

You can use the following program to see the color and other attributes for a number of style elements:

proc format; value vf 5 = 'GraphValueText'; run;

data x1;
   array y[20] y0 - y19;
   do x = 1 to 20; y[x] = x - 0.5; end;
   do x = 0 to 10 by 5; output; end;
   label y0 = 'GraphLabelText' x = 'GraphLabelText';
   format x y0 vf.;
run;

%macro d;
   %do i = 1 %to 12;
      reg y=y%eval(19-&i) x=x / lineattrs=GraphData&i markerattrs=GraphData&i
                                curvelabel="  GraphData&i" curvelabelpos=max;
   %end;
%mend;

%macro l(i, l);
   reg y=y&i x=x / lineattrs=&l markerattrs=&l curvelabel="  &l"
                   curvelabelpos=max;
%mend;

ods listing style=default;

proc sgplot noautolegend data=x1;
   title 'GraphTitleText';
   %d
   %l(19, GraphDataDefault)
   %l( 6, GraphFit)
   %l( 5, GraphFit2)
   %l( 4, GraphPredictionLimits)
   %l( 3, GraphConfidence)
   %l( 2, GraphGridLines)
   %l( 1, GraphOutlier)
   %l( 0, GraphReference)
   xaxis values=(0 5 10);
run;

The results in Figure 21.27 display the attributes for a number of the elements of the DEFAULT style.

When there is a group or classification variable, the colors, markers, and lines that distinguish the groups are derived from the GraphDatan elements that are defined with the style. In the DEFAULT style, these are elements GraphData1 through GraphData12. There can be any number of groups even though only 12 GraphDatan style elements are defined in the DEFAULT style. The following steps create a data set with 40 groups, display one line per group, and produce Figure 21.35:

data x2;
   do y = 40 to 1 by -1;
      group = 'Group' || put(41 - y, 2. -L);
      do x = 0 to 10 by 5;
         if x = 10 then do; z = 11; l = group; end;
         else           do; z = .;  l = ' ';   end;
         output;
      end;
   end;
run;

proc sgplot data=x2;
   title 'Colors, Markers, Lines Patterns for Groups';
   series  y=y x=x / group=group markers;
   scatter y=y x=z / group=group markerchar=l;
run;

The colors, markers, and line patterns in Figure 21.35 repeat in cycles. The GraphData1 – GraphData8 lines in Figure 21.27 exactly match the Group1 – Group8 lines in Figure 21.35. After that, there are differences due to the cyclic construction of the grouped style. This is explained next.

The DEFAULT style defines a marker symbol only in GraphData1 through GraphData7. The seven markers are: circle, plus, X, triangle, square, asterisk, and diamond. With the explicit style reference in Figure 21.27, the actual symbol, when no symbol is specified, is the circle. This is what you see for GraphData8 through GraphData12. With the group variable in Figure 21.35, the symbols repeat in cycles. Hence, Group1, Group8, Group15, and so on, are all circles. Similarly, Group2, Group9, Group16, and so on, are all pluses. The DEFAULT style defines 11 different line styles for GraphData1 through GraphData11. You specify line styles by specifying an integer. The default lines styles are: 1, 4, 8, 5, 14, 26, 15, 20, 41, 42, and 2. Hence, Group1, Group12, Group23, and so on, all have the same line style, which is a solid line (line style 1). Similarly, Group2, Group13, Group24, and so on, all have line style 4. There are twelve different colors, so Group1, Group13, Group25, and so on, all have the same colors. Overall, there are color/line/marker combinations that appear before any combination repeats. You can use the %MODSTYLE SAS autocall macro (see the sections Creating an All-Color Style and Style Template Modification Macro) to conveniently change these style attributes.

The HTMLBLUE style is an all-color style for the first 12 groups of observations. Most analyses have fewer than 12 groups. Markers and lines change for groups 13–24 and then again for groups 25–36. Figure 21.36 shows how colors, markers, and line styles change in the HTMLBLUE style. Figure 21.35 and Figure 21.36 through Figure 21.42 show how these elements change in other styles.