How to Easily Create Stunning Bar Charts in SAS

The resulting dataset, my_data, is now ready for graphical analysis. Notice the clear structure: the team and position variables are character types (indicated by the $ in the input statement), and points is numeric. The following examples will utilize the counts of these categorical variables to generate the bar chart visualizations.

Example 1: Generating a Basic Vertical Bar Chart

The most common application of a bar chart is to visualize the frequency of observations across a single categorical variable. In this first example, we use the VBAR Statement within PROC SGPLOT to count and plot the occurrences of each team (A, B, and C) in our dataset. By default, the VBAR Statement automatically calculates the frequency (count) of the variable specified on the vertical axis.

The code below defines a descriptive title for the output and then executes PROC SGPLOT, directing it to use my_data. The simple command VBAR team; instructs SAS to render a bar for each unique value of the team variable, with the height of the bar proportional to its frequency in the dataset. This visualization quickly reveals the relative distribution of observations across the teams.

/*create bar chart to visualize frequency of teams*/
title "Bar Chart of Team Frequency";
proc sgplot data = my_data;
    vbar team;
run;

Executing this code produces a vertical bar chart. As shown in the resulting image, Team A has 6 observations, Team B has 4, and Team C has 3. This type of visualization is essential for initial data exploration and quality checks, ensuring a balanced distribution or highlighting unexpected concentrations within the data.

Creating Horizontal Bar Charts with the HBAR Statement

While vertical bar charts are standard, horizontal bar charts often offer superior readability, especially when dealing with variables that have long category names. The flexibility of PROC SGPLOT allows us to switch from a vertical orientation to a horizontal one simply by replacing the VBAR statement with the HBAR Statement.

Functionally, the HBAR Statement is identical to VBAR; it calculates and plots the frequency of the categorical variable. The key difference is the axis orientation: categories are listed on the vertical axis, and the count is displayed along the horizontal axis. This subtle change can dramatically improve the user experience for reporting or presentation purposes.

To generate the horizontal version of the team frequency chart, the following minor modification is made to the previous code:

/*create horizontal bar chart to visualize frequency of teams*/
title "Bar Chart of Team Frequency (Horizontal)";
proc sgplot data = my_data;
    hbar team;
run;

The resulting horizontal chart presents the exact same frequency data, but now the bars extend horizontally. This visualization is particularly useful when comparing many categories, as the text labels for each category remain fully legible without requiring rotation or truncation, enhancing overall visual clarity.

Understanding Grouped Bar Charts in SAS

While basic bar charts are excellent for univariate frequency analysis, complex datasets often require visualizing the interaction between two categorical variables. This is achieved in SAS using grouped bar charts, which are categorized into two primary types: stacked and clustered. Both types utilize the GROUP option within the VBAR or HBAR statement, specifying a second categorical variable to define sub-groups.

The choice between a stacked or clustered format depends entirely on the analytical goal. A stacked bar chart is ideal for showing the “part-to-whole” relationship, illustrating how the sub-groups contribute to the total frequency of the main category. Conversely, a clustered bar chart is superior for direct comparison of the absolute values of the sub-groups across different main categories, as the bars for each sub-group share the same baseline.

In the following examples, we will use the team variable as our primary category and the position variable (Guard or Forward) as our grouping variable. This allows us to visualize not just the total observations per team, but also the breakdown of positions within each team, providing a richer, two-dimensional view of the dataset’s composition.

Example 2: Implementing a Stacked Bar Chart

A stacked bar chart provides an immediate visual summary of compound frequencies. To create this type of visualization in SAS, we simply add the GROUP = variable2 option to the VBAR Statement. In our case, variable1 is team and variable2 is position.

When the code executes, PROC SGPLOT first tallies the total frequency for each team. It then segments the resulting bar according to the frequencies of the defined positions (Guard or Forward), assigning a distinct color to each position. This immediately shows the contribution of Guards versus Forwards to the total count for Team A, Team B, and Team C, thereby visualizing the internal structure of the teams.

/*create stacked bar chart*/
title "Stacked Bar Chart of Team & Position";
proc sgplot data = my_data;
    vbar team / group = position;
run;

The resulting stacked chart clearly depicts the frequency distribution. For instance, Team A has slightly more Forwards than Guards, while Team B has a balance, and Team C has more Guards. The stacked format is particularly intuitive when the goal is to assess the proportional composition of the main categories, as the total height of the bar still represents the overall frequency.

Example 3: Constructing a Clustered Bar Chart

If the primary analytical need is to compare the counts of Guards directly against the counts of Forwards across all teams, a clustered bar chart is the preferred visualization method. The clustered approach separates the grouped variables and places them adjacently, allowing for direct comparison against a common baseline (the zero axis).

To achieve this clustered display, we build upon the stacked chart code by adding a single, crucial option to the VBAR statement: GROUPDISPLAY = CLUSTER. This setting overrides the default stacking behavior and instructs PROC SGPLOT to cluster the bars based on the position variable.

/*create clustered bar chart*/
title "Clustered Bar Chart of Team & Position";
proc sgplot data = my_data;
    vbar team / group = position groupdisplay = cluster;
run;

The output provides the same frequency information as the stacked chart, but the visual interpretation is different. For each team, we see two separate bars (one for Guard, one for Forward). This configuration is significantly better for comparing the absolute counts—for instance, quickly noting that Team A has the highest count of Forwards compared to Teams B and C—without needing to mentally adjust for the stacked segments.

Advanced Customization and Bar Chart Best Practices

While the examples above focus on frequency counts, PROC SGPLOT is capable of much more sophisticated aggregation. Instead of plotting frequency (the default), you can plot the sum, mean, or median of a numeric variable (like points) by using the RESPONSE option in the VBAR or HBAR Statement, combined with the STAT option.

For example, to visualize the total points scored by each team, you would modify the basic code: vbar team / response=points stat=sum;. This level of customization ensures that the generated bar chart accurately reflects the specific metric required for the analysis. Furthermore, visual aesthetics are paramount in data reporting. SAS allows control over colors (using the FILLATTRS option), data labels (using the DATALABEL option), and axis formatting (using YAXIS or XAXIS statements).

When designing bar charts, always ensure that the axes are clearly labeled and that the visualization style—stacked versus clustered—is appropriate for the narrative you wish to convey. Clustered charts are generally preferred when the goal is comparing categories, whereas stacked charts are best when the goal is understanding the internal proportional breakdown of a whole. Mastering these subtle differences allows for the production of highly informative and authoritative statistical graphics in SAS.

Conclusion and Summary of Bar Chart Types

Generating effective bar charts in SAS is a straightforward process facilitated by the robust PROC SGPLOT procedure. By correctly employing the VBAR Statement or HBAR Statement, along with judicious use of the GROUP and GROUPDISPLAY options, analysts can produce visualizations that range from simple frequency counts to complex multivariate comparisons.

The three methods demonstrated—basic frequency, stacked, and clustered—provide the foundation for nearly all bar chart needs. Analysts should select the chart type that best highlights the critical patterns in the data: the basic chart for single-variable distribution, the stacked chart for part-to-whole relationships, and the clustered chart for direct group-to-group comparisons. Adhering to these principles ensures clean, accurate, and insightful graphical results.

The following resources offer additional guidance on creating and customizing visualizations within the SAS environment: