Plot borders and layout

The plurality of plots in survey reports have no borders, as is the case with this unusual column plot from Akerman RealEstate 2016 [pg. 5]. The plot is unusual because it’s base rests on the bottom of the page, the x-axis line extends to the far left and right, and because there are no axis titles: those titles are shown next to the appropriate column.

Howes Percival SocialMedia 2018 [pg. 7] unusually outlines plots with dashed lines. Intermixed with that design choice are plots outlined in solid borders like the one below on the lower left.

Osborne Clarke Consumer 2018 [pg. 6] borders its plots with a solid line on all four sides. Some people may feel that this enclosure adds nothing to the presentation, merely more ink on the page.

In a final example, from Goodwin Law Rollover 2018 [pg. 6], horizontal lines bracket the plots at the top and bottom.

Plots and numbers of columns

Osborne Clarke Consumer 2018 [pg. 9] applies on most of its pages a two-column layout, and sometimes puts a plot in each column.

Another design choice has two columns of text but sometimes spreads a wide plot across both columns. Here’s an example from Norton Rose AustAgribus 2018 [pg. 14].

King Wood AustraliaDirs 2015 [pg. 25] spreads an unusual polar chart across the right two columns of a three-column layout. The gray bars on either side are the edge of the page, by the way, not a design element.

Going one column further, Burness Paull CorpScotland 2016 [pg. 15] stretches an unusual pie chart across all three columns.

DLA Piper Debt 2015 [pg. 8] extends a plot into the right margin.

Number of plots per page that have a plot

Our sense was that survey reports commonly place only a single plot on a page, but we scrutinized six reports to test that impression: Baker McKenzie Cloud 2015, Berwin Leighton ArbConflict 2010, Carlton Fields CA 2013, DLA Piper RE 2017, Fulbright Jaworski Lit 2013, and Seward Kissel HedgeFund 2015.

Going through each report, we counted how many pages (with a plot) had only one plot on them, how many had two plots, and how many had three plots. No report from this group had a page with four or more plots. Also, this exercise excluded all pages without a plot and did not count tables as plots.

The graphic below displays the results for each report as a proportion bar chart. The bar totals 100\% of the pages with plots and is broken into two or more segments, where the segment length tells what percentage of those pages had one plot (the first, purple, segment closest to the survey name), two plots (the second, turquoise, segment), or three plots (the third, rightmost, yellow, segment). Thus the darkest segment of each report’s bar indicates the percentage of plot pages that had only a single plot. For example, the Seward Kissel report had two-thirds of its plot pages with a single plot and one-third with two plots.

My hypothesis failed. Three of the reports indeed have long dark segments — most of their plot pages contain only a single plot. But double-plot pages dominate the other three reports. The pattern that is most evident from this small sample is that three-plot pages are rare. Only two of the six reports had a three-plot page, and only a single page at that (approximately 10\% of the plot pages).

Report elements (text, plots, other) as percentage of page space

The law firms in our collection of surveys make all kinds of decisions about the design of their reports. To start to understand the outcome of those decisions, we analyzed four reports. With each report we estimated the distribution of text, plots, and other design elements.

The four reports we chose were Eversheds, “21st Century law firm: Inheriting a new world” (2008); Hogan Lovells, “Risk and Return: Foreign Direct Investment and the Rule of Law” (2014); Norton Rose Fulbright, “Shipping” (2017); and White & Case, “Mining & Metals 2017: A tentative return to form” (2017). To be forthright about our criterion for selection, we picked four we had not written about before.

We decided to study the first five substantive pages of each report. Accordingly, we skipped the cover, executive summary, table of contents, and other introductory pages. For each of the next five pages we concentrated on the space available between the four margins (We ignored material in the margins on the left and right.). We estimated by eye what percentage of that space is filled with three elements: text (including space between lines), plots or tables of data, and other components such as pictures, embellishments, or pull-out quotes.

The eyeball estimation admits a degree of subjectivity, but we parsed all four reports consecutively so we applied a consistent standard. We are confident the findings are directionally correct even if other reviewers might estimate slightly different percentages.

For each of the four surveys, the next plot shows s a column for each of the three elements (Norton Shipping had no “Other” components so its column is missing). The columns indicate the average over the five pages reviewed of the percentage of space taken up by that element, for each report.  Together, they should total 100% for each report.  The first plot shows the elements side-by-side as columns.

The next plot shows the same data but stacks the elements into a single bar. People differ in their ease of interpreting the two plot formats.