Data tables

5.4.1 Data tables

performance analyses are made. Medium- to long- term budgetary allocation is essential prior to im- In the first case, the evidence is the data itself, or plementation to ensure the required continuity. aggregates arranged in appropriate strata. Data However, planning and strategy development tables are essential for this proof and typically con- should also account for enforced budgetary re- sist of columns of independent variables (time, ductions, through prioritization, sample scheme space or major strata) with their associated num- flexibility, and a clear description of their bers in the rows described by minor strata or their consequences.

combinations. Thus, a time series of catch, by fish- ery, by species would typically be represented as in Table 5.3.

Key presentational principles are shown in this Trial or test programmes prior to implementation table: not only verify the methodology, they also test the • Arrangement: strata type and the independent acceptance and level of support from the data sup- variable in columns; major strata (in this case fish- pliers. This support can be maintained during ing method) and aggregate data highlighted; and implementation in many ways, including the minor strata (in this case species data) are sub- demonstration of the benefits and the provision of sidiary to major strata. incentives, or by the application of penalties.

Implementation phase

• Format: numbers right-aligned; same font as Continuous feedback is required to support text with slightly smaller point size; table margins data collection programmes and to ensure their indented from text and centred on page; optimum continued applicability, as emphasized in the in- use of white space; and frames (grid lines) are mini- formation cycle. Thus, training in new or revised methods as programmes proceed, and consulta- tion to ensure data collection practices are the most up-to-date and meet all stakeholder require- Table 5.3 Total catch in metric tonnes by fishery by ments, should be maintained as an integral part of species, 1996–9 (hypothetical). implementation.

Fishery/species

5.4 20 700 FISHERIES INFORMATION

PRESENTATION 12 000

7 000 7 500 8 000 Fisheries data most often relate to time series or Demersal trawl

yellowfin

24 000 23 400 31 200 spatial distributions (two- and three-dimensional).

22 500 22 500 30 000 Sometimes both time and space – the four dimen-

cod

1 500 900 1 200 sions – are required to make sense of the data. In Pelagic trawl

hake

40 000 45 000 65 000 other cases, data sets may consist of two or more

12 000 13 500 23 000 variables, which may or may not be related to one

mackerel

28 000 31 500 42 000 another. To cope with the often-large data sets that Purse seine

herring

12 000 32 000 are generated it is usually necessary to present 25 000

7 000 18 000 17 000 data in ways that (1) provide the evidence upon

herring

5 000 14 000 8 000 which statistical conclusions are made and (2) Total

sardine

95 300 118 700 provide a means to reveal the information that the 141 900

Chapter 5

mal, clear, of minimum weight, and placed to de- • presenting many small numbers in a small lineate major strata.

space, and making large data sets coherent; • encouraging visual comparison of differences and revealing the data at both lower and higher lev-

5.4.2 Data graphics

els of detail; and

Using simple principles of graphic presentation • serving a clear purpose and being closely inte- usually leads to a better view of the information grated and relevant to the statistical and verbal content of data. Indeed, graphics are often the descriptions of the data. only way of revealing information since they Figure 5.3 contains some of these principles using can be used to accurately depict large data the data from the major strata in Table 5.3. This sets, which are nearly always multivariate. There graphic depicts some design principles that sup- are numerous ways of visually presenting data: port the way information in presented as shown histograms (column, bar), line and scatter above, including: plots, proportional plots (pie, area, radar) to three- • The representation of numbers as physically dimensional surfaces. The most common is the measured on the surface of the graphic should be column histogram, which enables standard pre- directly proportional to the numerical quantities sentation of the independent variable along the represented. (Note: the common use of three- x-axis, for example the passage of time, and a dimensional bars adds nothing to the information variety of ways of presenting data for single and content, and a poor choice of perspective can even aggregated strata.

lead to visual distortion.)

In general, graphic presentations should reveal • Clear labelling should be used to minimize dis- the information content of data by:

tortion and ambiguity about the data. Legends • showing the data by representing the substance should be placed in support of the data. rather than the method of presentation;

• Variation in the data should be shown, not de- • avoiding distorting the data – either graphically sign variation. The choice of stacked columns here or by showing the data out of context;

provides visual comparison of the variations in

Metric tonnes (thousands) 40

Fig. 5.3 Total catches (metric Longline

Demersal trawl

Pelagic trawl

Purse seine

tonnes – thousands) by fishery, 1996–9 (hypothetical).

both fishery and total annual catches. (Note: mul- tiple graphics within documents should use the same or very similar formats, as they often also form a visually coherent series.) • Maximize the ink related to the data (areas, points, lines and text), which draw the eye to the data itself. Minimize the ink related to design (grids, axes, fills and other decoration – often called chartjunk). (Note: the common use of cross- hatching creates moiré effects that can visually disturb the information content of the graphic.) Although not presented here, the often-depicted times series of money (and derived indicators) should usually use deflated and standardized units of monetary measurement rather than nominal units. For a wider appreciation of data graphics, refer to the Graphics Press series by E.R. Tufte (1987, 1990 and 1997).