A More Advanced Pie Chart for Analysis Services Data

MSAS Architect Bill Pearson leads the hands-on creation of a somewhat more advanced Reporting Services’ pie chart, using dual category groups and a series group based upon an Analysis Service data source.

This article examines additional capabilities we can employ in creating advanced Pie charts with Reporting Services. It extends the examination of simple Pie charts that we undertook in Simple Pie Chart for Analysis Services Data, and focuses on the creation of a somewhat more sophisticated Pie chart than we created in the earlier article, using dual category groups and a series group, among other nuances, again based upon an Analysis Service data source. This article continues the examination of Reporting Services charts for Analysis Services data sources that we began in an earlier article of my MSSQL Server Reporting Services series, Introducing Reporting Services Charts for Analysis Services. In that article we summarized the many different chart (or chart data region) types that are available, and looked ahead to individual articles surrounding each type, where we would specify details – and real world innovations – involving the use of each in reporting Analysis Services data. We noted that the focus of these related articles, interspersed among other topics within my MSSQL Server Reporting Services series over time, would be the design and creation of Analysis Services chart reports of various types, and the exploitation of the rich and flexible features contained in Reporting Services that enable us to make report data more meaningful, and easier to understand, from the perspectives of our information consumer audiences.

In many cases, which I try to outline in my articles at appropriate junctures, the functionality of well-established, but expensive, solutions, such as Cognos (PowerPlay, Impromptu, and other applications), Business Objects, and myriad other reporting / OLAP applications can be met by Reporting Services – at a tiny fraction of the total cost of ownership. And the flexibility and richness of the chart data regions alone in Reporting Services exceed, in many ways, the rather fixed options available in other enterprise reporting solutions.

As I have repeated in many of my articles in this column, one of the first things that become clear to “early adopters” of Reporting Services is that the “knowledgebase” for Analysis Services reporting with this tool is, to say the least, sparse. The vacuum of documentation in this arena, even taking into consideration the release of several books surrounding Reporting Services in recent years, continues to represent a serious “undersell” of Reporting Services, from an Analysis Services reporting perspective. I hope to contribute to making this space more accessible for everyone, and to share my implementation and conversion experiences as the series evolves. In the meantime, we can rest assured that the Analysis Services potential in Reporting Services will contribute significantly to the inevitable commoditization of business intelligence, via the integrated Microsoft BI solution.

Note: For more information about my MSSQL Server Reporting Services column in general, see the section entitled “About the MSSQL Server Reporting Services Series” that follows the conclusion of this article.

Overview

As we observed in Introducing Reporting Services Charts for Analysis Services, Reporting Services enables us to present both summarized and detailed data in colorful, easy-to-read charts of various designs, from which we can chose the layout and type that best meets any given business requirement. We learned, in Simple Pie Chart for Analysis Services Data, that, among the types offered, the simple Pie chart type is perhaps among the most popular. In this article, we will extend our examination of the Pie chart type and get some hands-on exposure to the creation of an enhanced Pie chart, continuing to explore general characteristics. This will serve as a basis for other, more in-depth, practical exercises in coming articles, where we will extend the value of our chart-enhanced reports in myriad ways. Among these ways, just for starters, are the capability to format chart and other objects within a host of options, to drill down to see the details behind the graphical / numerical summaries, to combine chart reports with other types of reports, and to access many other options in the powerful Reporting Services tool set.

My objective within this article is to assist the reader in quickly assembling a report containing a working, enhanced Pie chart (relying upon, for instance, already assembled datasets and other underlying support within an existing sample report), and to move efficiently into targeted reporting nuances that meet real world needs. While this initial introduction will focus more on the creation of a relatively straightforward, albeit enhanced, Pie chart, the report we create will serve as a basis, in prospective articles, to demonstrate more detailed intricacies that I have found useful in meeting business requirements of my own clients and readers. The ultimate objective, as is typically the case within my various series’, is to provide hands-on opportunities to learn overall, start-to-finish procedures, before homing in on specific options of interest (although we will certainly deal with many of these options in even our early exercises, as a part of completing the stated objectives of these sessions).

Introducing Pie Charts for Analysis Services

In Introducing Reporting Services Charts for Analysis Services, we learned that the Pie chart type is available in the following variants:

• Simple Pie
• Exploded Pie

In this article we will focus upon the Simple Pie variant, although we will enhance the Pie chart beyond the simple Pie chart that we created in Simple Pie Chart for Analysis Services Data. (We take up the Exploded Pie variant within a relevant context in a sister article of the MSSQL Server Reporting Services series.)

We noted in our introductory article that Pie charts are typically used to present percentages (as in “percent composition of a whole,” etc.). In generally describing the type, we observed that the Pie chart presents value data as percentages of the whole. Categories are represented by individual slices. The size of the slice is determined by the value.

In this article, we will explore an enhanced Pie chart data region in detail, and gain practical exposure to the creation of a basic example of such a chart that is employed in reporting from an Analysis Services data source. In introducing the enhanced Pie chart, we will:

• Perform a brief overview of the Pie chart type, discussing its variants and typical uses;
• Open the sample Report Server project, AdventureWorks Sample Reports, and ascertain connectivity of its shared Analysis Services data source;
• Create a clone of an existing sample Analysis Services report, containing a matrix data region, with which to launch our overview;
• Examine the Pie chart type from the standpoint of the existing report, noting how we add it to an open report (and thus save time in leveraging existing datasets and other support structures) in the Layout tab;
• Modify the existing primary dataset within the sample report clone, adding a filter to limit the size of the data presentation;
• Make modifications to the report layout to support the stated client reporting needs and practice session objectives;
• Create a complete, working sample of an enhanced Pie chart data region, within the existing report, and alongside a pre-existing, populated matrix data region, which will allow us to verify its accuracy and completeness once we have the chart in place;
• Examine all relevant property settings within each of the General, Data, X Axis, Y Axis, Legend, 3D Effect and Filters tabs;
• Discuss the results obtained with the development techniques that we exploit throughout our practice session.

Objective and Business Scenario

In this article, we will continue to perform a relatively straightforward examination of the Pie chart type, from within a copy of an existing sample Reporting Services 2005 report that we will create for this purpose. Our focus will be to create a working, enhanced Pie chart, using an Analysis Services data source (the Adventure Works DW sample OLAP database / Adventure Works cube that accompanies the installation of Reporting Services), while discussing various characteristics of this chart type as we progress.

We will examine relevant chart properties, and get further hands-on exposure to the manipulation of those properties to support the delivery of information to meet the needs of a hypothetical group of organizational information consumers. Other articles within the MSSQL Server Reporting Services series will advance beyond the practice session that we undertake here, using both the simple Pie chart we created in Simple Pie Chart for Analysis Services Data and the enhanced Pie chart which we will create in this article, as a basis from which we can concentrate on in-depth procedures and nuances that we can use to achieve precision in meeting specific requirements, and to deliver data presentation effects, that we might encounter within the environments of our respective employers and / or clients.

The Business Need

For purposes of our practice procedure, we will assume that a group of report developers and analysts, composed of members of the Sales, Marketing, Information Technology, and other departments of the Adventure Works organization, have expressed the need to present some of the information displayed in the existing Sales Reason Comparisons OLAP report through a new Pie chart report. The group has stated that they want to leverage this “conversion” process to learn more about the construction and characteristics of an enhanced Pie chart report in general. Moreover, they assure us that they will extrapolate the techniques they learn to scenarios where they will design, create and deploy reports of this type in the future.

Once we understand the business need, we propose using a copy of the existing Sales Reason Comparisons report (which, among other samples, accompanies the installation of Reporting Services). Our tandem objectives here, we explain, are 1) to streamline our procedures (by using existing connections, datasets, and other structures that are already in place within the pre-existing report), and 2) to provide a ready means of verifying at least some of the accuracy and completeness of the new report (a feature that might be useful in initial report testing). We mention, as an aside, that we can always delete the existing matrix data region prior to deploying the final report.

Once we obtain agreement on this approach, we begin the process of creating the enhanced Pie chart report to satisfy the information consumers.

Practice

Our first objective is to create a copy of the Sales Reason Comparisons sample report, within which we will perform the addition and setup of a Pie chart data region, for the reasons noted above. (We typically work with a similar report copy in articles that are focused on the individual chart types elsewhere in the series.) We will perform our practice session from inside the MSSQL Server Business Intelligence Development Studio. For more exposure to the Business Intelligence Development Studio itself, and the myriad design, development and other evolutions we can perform within this powerful interface, see other articles in this column, as well as within my Database Journal series Introduction to MSSQL Server Analysis Services. In this article, we will be commenting only on the features relevant to our immediate practice exercise, to allow us to keep to the focus of the article more efficiently.

Preparation: Create a Clone Report within the Reporting Services Development Environment

For purposes of our practice session, we will create a copy of the Sales Reason Comparisons report, one of several samples that are available with (and installable separately from) the MSSQL Server 2005 integrated business intelligence suite. Making preparatory modifications, and then making the enhancements to the report to add the functionality that forms the subject of our lesson, can be done easily within the Business Intelligence Studio environment. Working with a copy of the report will allow us the luxury of freely exploring our options, and will leave us with a working example of the specific approach we took, to which we can refer in our individual business environments.

Open the Sample Report Server Project

For purposes of our practice session, we will open the AdventureWorks Sample Reports project, which contains the sample reports that ship with the Reporting Services component of the MSSQL Server 2005 suite. We will complete our practice session within the sample project, so as to save the time required to set up a development environment from scratch within the Business Intelligence Development Studio.

To open the AdventureWorks Sample Reports project, please see the following procedure in the References section of my articles index:

Ascertain Connectivity of the Shared Data Source

Let’s ensure we have a working data source. Some of us may be running “side-by-side” installations of MSSQL Server 2000 and MSSQL Server 2005. This means that our installation of the latter will need to be referenced as a server / instance combination, versus a server name alone. (The default for the Adventure Works DW project sample’s connection is localhost, which will not work correctly in such a side-by-side installation, as MSSQL Server 2000 will have assumed the identity of the local PC by default.)

If you do not know how to ascertain or modify connectivity of the Analysis Services data source, please perform the steps of the following procedure in the References section of my articles index:

Create a Copy of the Sales Reason Comparisons Report

We will begin with a copy of the Reporting Services 2005 Sales Reason Comparisons OLAP report, which we will use as a basis for our Column chart types practice exercise. Creating a “clone” of the report means we can make changes to select components (perhaps as a part of later exploration with our independent solution), while retaining the original sample in a pristine state for other purposes, such as using it to accompany relevant sections of the Books Online, and other documentation. Such uses may form a part of learning more about Reporting Services (particularly about an OLAP report using an Analysis Services data source), and other components of the Microsoft integrated business intelligence solution in general.

If you do not know how to create a copy of an existing report, please perform the steps of the following procedure in the References section of my articles index:

We now have a clone Analysis Services report file within our Reporting Services 2005 Project, and are ready for our hands-on practice session with the enhanced Pie chart type, which we will begin in the next section.

Preparation: Modify the Analysis Services Report Clone for Use within Our Practice Session

We will next make a few modifications to prepare the report for our practice session. Let’s open the report clone we created above in Layout view (for those of us not already there), upon which we can commence our overview steps.

DBJ_OLAP_Report.rdl opens in Layout view.

Filter the Primary Dataset to Limit the Data Retrieved

First, we will modify the primary dataset underlying the existing report, called ProductData. We will simply place a filter on returned data to limit the size of our ultimate report to two (versus four) Calendar Years’ activity.

The Query pane, with the newly added Calendar Year column, appears as partially presented in Illustration 2.

The multi-tabbed Dataset dialog opens.

="CY 2003"

Because we are stating that we want only Calendar Years “greater than or equal to CY 2003,” we know we will retrieve only two years’ data into the dataset (the Adventure Works cube contains Calendar Years 2001 through 2004).

The Filters tab of the Dataset dialog appears as presented in Illustration 5.

Modify the Report Layout to Accommodate Our Focus

Next, we will make some changes to the report layout to more easily accommodate the focus of our practice session. We will start with a larger report canvas.

The Body bar, atop the report canvas, assumes the focus (becomes darker). Body also appears in the Properties pane (by default to the lower right of the design environment). The point here is to ascertain that the report body is, indeed, selected.

Because, from this point, we will be performing procedures that relate only to the chart type with which we are working in this article, and because we create similar files for different chart types in other articles, let’s rename the .rdl file to clearly associate it with this article and the Pie chart type.

RS062_Pie_Chart

The Save File As box is dismissed and we see the new name appear in the Solution Explorer. We are now ready to begin the procedural section of our practice session.

Procedure: Create a Pie Chart within the Report

Add a Chart Control to the Report Canvas

At this point, we will select the Chart item from the Report Items listed within the Toolbox pane (the position of which is defaulted to the left of the Layout tab), for inclusion within our report. Once the general control is placed on the report canvas, we will adjust it, setting properties as appropriate to meet the report specifications.

The toolbox window should appear similar to that depicted in Illustration 8. Mine is pinned to the left side of the design environment, where I find it most convenient. This is, of course, subject to your own choice. (I also dock my Fields and Server Explorer panes in this area to maximize design real estate, as an aside.)

The cursor becomes a small chart icon, in combination with crosshairs, when held above the report canvas. This indicates that we can click to “anchor” the point, from which we wish to define the space that the chart will inhabit.

The chart item appears, in its generic manifestation, as depicted in Illustration 10.

The chart item is now in place, and we are ready to specify its “Pie chart” nature, as well as to populate it with the primary dataset that is already in place.

Populate the Pie Chart item to Meet the Business Requirements

While our next steps might focus upon simply dragging fields from the Fields pane of the Report Designer, we will populate our chart from inside the multi-tabbed Chart Properties dialog. Because few charts I encounter in the business environment can be built completely through the “drag-and-drop” process anyway (expressions are often involved in combination with the data fields, as an example), I have developed the habit of working from the Properties dialog. (The practice of tabbing through the Properties dialog can often remind us of other settings we need to make, additional options that are available in meeting client / employer needs, and so forth, as an another benefit.)

The Properties dialog appears, defaulted to the General tab, where we will begin our examination of the chart properties in the subsections that follow.

General Tab

Sales Reason Comparisons

The Style Properties dialog box appears.

The Style Properties dialog appears, with our settings, as shown in Illustration 13.

We return to the General tab.

We can manipulate settings for chart borders and lines via the Borders and Lines tab, depicted with default settings in Illustration 14.

We return to the General tab, once again.

The two-tabbed Style Properties dialog that appears is identical to that which we saw when we clicked the Chart Area Style... button in the immediately preceding steps. Since we will be leaving these settings, too, at their defaults, we will dismiss the Style Properties dialog at this point.

We return to the General tab, once again, which appears, with our settings, as depicted in Illustration 16.

We will move to the Data tab in the subsection that follows.

Data Tab

The values in a Value series determine the size of the slice in the pie. Multiple Value series appear as separate slices. Settings for the x- and y-axis are ignored for Pie charts, as we shall see. If we have multiple Category or Series groups, the group labels are displayed in the chart legend.

The Edit Chart Value dialog appears, defaulted to the Values tab.

Internet Sales

The Expression Editor opens.

The field values of the ProductData dataset appear within the pane at the bottom right of the Expression Editor.

The Expression Editor appears, with our input, as shown in Illustration 19.

We are returned to the Edit Chart Value dialog, which appears with our input as depicted in Illustration 20.

The dual – tabbed Style Properties dialog appears again, just as it did for the Chart Area Style... and Plot Area Style... buttons we examined earlier. We see the same Border and Line and Fill tabs, which we will leave at default at this point.

This will result in the point labels being displayed within the slices of the Pie (when accommodated by slice size), making the chart more compact and, in this case, making the labels themselves somewhat easier to visually associate with their respective slices.

The Style Properties dialog appears, with our settings, as depicted in Illustration 22.

The Style Properties box closes. We return to the Edit Chart Value dialog box, which now appears as shown in Illustration 23.

Here we can define Hyperlink actions, including Jump to report, Jump to bookmark, or Jump to URL. We will not be working with a drillthrough scenario in this practice session, and so will leave undisturbed the default setting of None, as depicted in Illustration 24.

The Data Output tab affords us a means of defining XML data output options for the values in the chart. We will not leverage these capabilities within this introduction to simple Column charts, and will therefore leave the settings of the Data Output tab at their defaults, as shown in Illustration 25.

We are returned to the Data tab of the Chart Properties dialog. Next we will work with the Category groups area of the tab, just under the Values area. Categories are displayed as individual slices in the pie. Each Category is displayed in the chart legend, when the legend is selected for use.

The Grouping and Sorting Properties dialog appears, defaulted to the General tab.

Year

As we have no input for the Filters or Data Output tabs (which resemble their counterparts in other areas of the Report Designer, as we have, and shall, see), within the context of this newly added group, and as we intend to leave the Sorting at the settings naturally found within the cube structure (the Sales Reasons will sort, where appropriate, in alphabetical order), we will accept our input at this stage.

The Grouping and Sorting Properties dialog appears, with our modifications, as shown in Illustration 27.

We will next add another Category group – this time a “group within a group,” for the Sales Territory.

The Grouping and Sorting Properties dialog appears, defaulted to the General tab, as before.

Territory

Because, once again, we have no input for the Filters or Data Output tabs, within the context of this newly added group, and as we also intend, as earlier, to leave the Sorting at the settings naturally found within the cube structure (the Sales Reasons will sort, where appropriate, in alphabetical order), we will accept our input at this stage.

The Grouping and Sorting Properties dialog appears, with our modifications, as depicted in Illustration 28.

We are returned to the Data tab of the Chart Properties dialog, which appears, with our Values and Category groups settings, as shown in Illustration 29.

Our next step will be to add another group – this time a Series group. Series are displayed as individual slices in the Pie chart. Each series is also displayed in the chart legend. At this point, we will add a Series group to contain the Sales Reason (“Manufacturer,” “On Promotion,” “Price,” etc.) that is associated with the grouped Internet Sales totals that the report will present.

The Grouping and Sorting Properties dialog appears, once again, defaulted to the General tab.

Sales_Reason

The Grouping and Sorting Properties dialog appears, with our modifications, as shown in Illustration 31.

Because we again have no input for the Filters or Data Output tabs within the context of this newly added group, and as we intend to leave the Sorting at the settings naturally found within the cube structure (the Sales Reasons will sort in alphabetical order), we will accept our input at this stage, just as we did with the previously added Category groups, Year and Territory.

We return to the Data tab, once again, which appears, with our Values, Category groups, and Series groups settings, as depicted in Illustration 32.

We will move to the X Axis tab in the subsection that follows.

X Axis Tab

The X Axis tab is disabled for a Pie chart data region. All settings are therefore “grayed out.”

Y Axis Tab

The Y Axis tab, like the X Axis tab, is disabled for a Pie chart data region. All settings are therefore “grayed out.”

We will move to the Legend tab next.

Legend Tab

We advise our client colleagues that, since the default position for the legend is to the right of the chart area, and since our simple pie chart may be wider than expected, due to its nature and the amount of data we are presenting, placing the legend underneath the chart will offer another means of compressing the overall presentation.

The Style Properties dialog box appears, defaulted to the Font tab.

The Font tab of the Style Properties dialog appears, with our settings, as shown in Illustration 33.

We note the presence of the Border and Line and Fill tabs, but we will leave the settings within each at default, at this point.

We return to the Legend tab, once again, where we will leave all else at default. The relevant portion of the Legend tab appears, with our settings, as depicted in Illustration 34.

We will move to the 3D Effect tab in the next subsection.

3D Effect Tab

We inform our client colleagues that the 3D Effect tab offers us a means of transforming the ordinarily “flat” appearance of our Column chart to a highly customizable, three – dimensional presentation. Here we can enable 3-D visual effects (via the checkbox to the immediate left of the Display chart with 3-D visual effect label). The four variables that we can manipulate are:

• Horizontal rotation
• Perspective
• Wall thickness
• Vertical rotation

Once 3-D visual effects are enabled, a slider becomes enabled for each of these variables, which we can move to adjust each variable until we achieve just the degree of readability we desire within the chart.

We will leave the settings of the 3D Effect tab at default at this point. The 3D Effect tab appears, with default settings, as shown in Illustration 35.

We will examine the final remaining tab, Filters, in the next subsection.

Filters Tab

The Filters tab is but one option we have, within Reporting Services, to filter the data that is displayed within our chart. We advise our client colleagues that we performed all desired filtering at the dataset level earlier within our practice session. When this is adequate (that is, when we can afford to filter at the dataset level for the entire report) we may achieve performance gains at report runtime, due to the overall retrieval of less data from the Analysis Services data source. But, we caution the client representatives, due consideration should be given to the various points at which we can filter within a given report, to ascertain that we optimize performance while retaining complete and accurate information for presentation.

The Filters tab allows us to choose either simple dataset columns or expressions to filter data at the chart level. This might make sense as a filter point if we were, say, using multiple data regions (charts, matrices, tables, lists, or a combination of these, perhaps) that were sharing the same common dataset(s), but where each region had different filtering requirements and needed to present different subsets of data from the underlying dataset(s). Whatever our needs, Reporting Services, once again, offers flexibility in ways to meet the challenges involved.

Let’s filter our Pie chart to show results of only the European and North American Territories.

="Europe"

We note that an “or” is automatically inserted in the “And/Or” column on the far right of the first row.

8.  In the Operator column (to the immediate right of the Expression column), select the “=” operator

9.  Type (or cut and paste) the following into the Value column (to the immediate right of the Value column):

=North America"

We are simply asking to see the Internet Sales values for two of the Sales Territories, at this point. In reality, we might parameterize territory selection to make the report do “more with less,” in offering information consumers an ad hoc choice of territories at runtime.

The Filters tab of the Chart Properties dialog appears as shown in Illustration 37.

The Chart Properties dialog closes, returning us to the placeholder chart item in Report Designer, Layout tab. We will adjust the height of the chart item to make the preview we conduct next easier to read.

The Layout tab appears, with our adjustment, as depicted in Illustration 38.

We will conclude our practice session in the next section, where we will verify the operation of our new Pie chart.

Verify Operation of the Pie Chart

Let’s ascertain the accuracy and completeness of our construction efforts. We will execute the report with the following steps:

The new report generates, displaying both the original matrix and new Pie chart data regions.

The Pie chart data region appears as depicted in Illustration 40.

We can easily verify the displayed simple Pie chart totals against the matrix data region by comparing any given Sales Reason’s Internet Sales Amount displayed in the Pie chart data region to the total Internet Sales Amount corresponding to the respective Sales Reason in the matrix data region. For example, if we add CY 2003 – Europe – Price – Internet Sales (1,519,252) and CY 2004 – Europe – Price – Internet Sales (2,066,776), we obtain a total that corresponds with the total Internet Sales Amount (consisting of a combination of amounts for Calendar Years 2003 and 2004) displayed for Europe (3,586,028) in the original matrix data region that we have left within the report.

Our enhanced Pie chart meets the expressed business requirements and demonstrates many details surrounding its property settings. The client representatives express satisfaction with our efforts, and state that, with a few cosmetic changes (including the simple removal of the existing matrix data region, and the subsequent realignment of the Pie chart on the canvas, realignment / resizing of the point labels to better fit the slices of the pie, etc.), the report will be ready for deployment to the targeted information consumer group. Moreover, they assure us that the details they have examined within the practical exercise we have undertaken can be extrapolated to their creation efforts of other, perhaps even more sophisticated, Pie charts.

As we can see from our examination above, Reporting Services offers a wide range of options for Pie chart creation and manipulation, to assist us in the delivery of information within the business environment. We extend our examination of chart types, specifically examining each type, together with the properties and methods we can manipulate for the precise presentations we seek to be able to deliver, in other articles of this series.

Conclusion

In this article, we performed a relatively straightforward examination of a Reporting Services enhanced Pie chart, from within a copy of an existing sample Reporting Services 2005 report that we created for this purpose. Our focus, as we stated in the introduction, was to create an enhanced, working Pie chart, using an Analysis Services data source (the Adventure Works DW sample OLAP database / Adventure Works cube that accompanies the installation of Reporting Services), and to discuss various characteristics of the Pie chart type as we progressed.

We examined relevant chart properties, and got some hands-on exposure to the manipulation of those properties to support the delivery of information to meet the needs of a hypothetical group of organizational information consumers. We noted that this article would serve as a basis, in some cases, for other, more advanced articles within the MSSQL Server Reporting Services series, where we use the Pie chart we created here as a basis from which we can concentrate on in-depth procedures and nuances that we can use to achieve precision in meeting specific requirements and data presentation effects that we might deliver in the business environment.

About the MSSQL Server Reporting Services Series ...

This article is a member of the series MSSQL Server Reporting Services. This monthly column is designed to introduce MSSQL Server Reporting Services (“Reporting Services”), presenting an overview of its features, with tips and techniques for real-world use. For more information on the series in general, please see my initial Database Journal article, A New Paradigm for Enterprise Reporting.

» See All Articles by Columnist William E. Pearson, III