description of the survey methods, (c) an evaluation of the quality of the data, (d) a discussion of qualifications or limitations of the data, and (e) references to related statistics from other sources. To set the background for the statistical tables, the text might begin with a summary of findings (particularly appropriate for the preliminary reports).

The typing, page size, amount of reduction, etc. are discussed fully in chapter III-6. Although that chapter emphasizes the technical aspects of setting up and printing tables, the guidelines are similar. For readability, verbal materials in the text are usually arranged in two columns and the amount of reduction may be less than the reduction for statistical tables.

2.1 Separate series by industrial sector

There will be three preliminary reports, one for each of the three major areas covered by the census. These reports will contain selected statistics for industry and geographic areas.

For the final reports, each of these sectors will again be handled by a separate report series. There will be a pattern to the content in each of the series. For manufacturing establishments, final publication will present industry data for Providencia as a whole (13 tables); data for provinces, counties, or wards (8 tables); and manufacturing activity according to other characteristics such as size of establishment and type of ownership (4 tables). Publications that group tables by these three systems of classification will be issued for mineral industries and for electricity and gas industries.

Uniformity in the tabular content of the final reports benefits the data user. The

user will frequently want to refer to all the industry data for in-depth studies of an industry or group of related industries or to all the geographic data for regional analysis. Furthermore, this feature of parallelism in tables aids in the preparation of the census publications. Exhibit III-5-9 demonstrates the close concordance of the manufactures and minerals census tables in the final report series. As part of the final report phase of the publication program there will be a report covering the household industries. Additionally, there will be an overall summary release bringing together key totals from all three census

areas.

2.2 Time schedule

The 1975 Providencia Industrial Census will have a series of preliminary advance reports followed by the final report series. The advance reports will be issued within 12 months of the end of the reference year (1975), that is, by 31 December 1976. The final reports will appear on a flow basis during 1977 with all of them in print by the end of that year. There will be a single bound volume comprising all of the final reports (with redundant text and pages eliminated) published early in 1978.

3. DESCRIPTION OF TABLE OUTLINES

3.1 Table content

Table outlines (formats) consist of the title, headnote, boxhead, stub captions, and footnotes, if any, of each table to be included in the census publication. They do not have to be prepared in complete detail but should contain enough information to enable the table planner of the Publications Division to count spaces and lines accurately, and the systems analyst of the Data Processing Division to relate the table content to the questionnaire by means of a "key" technique.

Exhibits III-5-1, 2, and 3 contain the formats of the tables in the preliminary (advance) reports on manufacturing, mineral industries, electricity and gas industries. The preliminary reports are scheduled for release by the end of 1976. Exhibits III-5-4, 5, and 6 illustrate the table formats for

the final reports on the three industry sectors. The tabular content of the final summary reports with key totals from all three industry sectors is presented in exhibit III-5-7. Finally exhibit III-5-8 shows the detail for the tables in the special report for household industries.

Exhibit III-5-10 illustrates the relationship between tables and inquiries on the forms, using the manufacturers final report series to show how all of the information collected from respondents is published and in a logical sequence.

3.2 Table design

For most censuses, statistical tables consume a substantial portion of the reports. The design of the tables is described in detail in chapter III-6.

1. GUIDELINES FOR TABULAR PRESENTATION

The tabular presentation of Providencia's Industrial Census reports is based on a set of rules, the application of which depends greatly on good judgment. However, these rules involve principles and technical details which every statistician needs to know about in order to instruct clerks, typists, and printers in the formatting of tables.

It is of major importance to keep in mind who is going to see and use the table. One should always plan the table from the viewpoint of the user. In observing this rule, everything possible should be done to make tables easy to read, understand, and interpret. The table form does not add new meaning to the figures; it is intended only to make the meaning which is already there easier to find.

Good style in tabular presentation implies a clear and economical display. This means that consistent practices must be developed relating to the layout or design of a table; the way table titles are written; the way classifications are handled; the use of footnotes and headnotes; the use of vertical and horizontal rules; and consistency in capitalization, abbreviation, punctuation, and indentation.

As the term implies, a text tabulation usually appears as part of the text of the report. Text tabulations are simple ruled tables without a table number or title. In other respects, the form is the same as that of the formal statistical table. Since the text tabulation has no identification, it must be introduced by an explanatory statement.

Text tabulations are difficult to use because, in the arrangement of the page, they cannot be shifted away from their introductory statement (this is also true of leader work).

The table number is an identifying device; it positions a table within a series. In a single report, no two tables should bear the same number. The tables should be numbered in one series, using Arabic numerals for example, and starting with 1. When there are text tables as well as a separate series of formal tables, the text tables having titles should be identified by capital letters or possibly Roman numerals (usually there are not many text tables). Capital letters are preferred, as they are easier to read and understand than Roman numerals. The table number is followed by a period and two spaces (or a 1-em dash if letterpress). The table number may be set in boldface type for easy identification.

wording and arrangement of the component parts

of the title depend on the comparative significance of these two factors in the given presentation.

3.22 Component parts.--Each title should indicate, specifically or by implication, four basic characteristics of the tabular data. The parts are joined by appropriate connectives and punctuation to form the title as a whole. The parts can be described as "title segments." (1)

(2)

Universe segment: Specifies the nature and limits of the group covered by the table. For example: Employed persons; Manufacturing establishments; Production workers. The universe segment usually appears as the first or second component part of the title. Placement at the beginning of the title is suitable in a series of tables involving many different universes or in a table presenting general information on the universe without stressing a particular characteristic. If the universe segment is long, part of it can be put in the headnote; however, universe qualifications which distinguish one table from another should remain in the title. On the other hand, if the basic universe is identical for all tables in a series and the nature of the universe is made clear in the report title or running head, the universe segment may be omitted completely from the title and headnote.

Classification segment: Specifies how

the universe data are classified and cross-classified. For example: By age and sex; By number of hours worked; By occupation and industry. This segment may precede or follow the universe segment (or may be split). Placing the classification segment at the beginning of the title is most suitable in a series of tables with the same or similar universe and in which the emphasis is on differences in the classification of data. For example: Age of production workers; Weekly salary of production workers. The extent to which varying detail of classification should be shown in the title depends largely on whether the primary purpose of the table is such as to make detailed description essential. Where possible, the title should be kept short by indicating no more than the basic classifications, unless this results in an over-statement of the table contents. For summary tables containing a number of subjects, the terminology,

(3)

(4)

"General characteristics of .." may be

used and the separate characteristics can be identified in the boxhead spanner or the stub center heads; only one such table title would be appropriate in any series.

Area or space segment: Specifies the administrative divisions or subdivisions, geographic area, or place to which the data refer. For example: For rural areas; For Providencia; For Provinces; For urban places. Normally this segment is placed toward the end of the title immediately preceding the time reference. The area or space segment may be omitted if the report relates to one area and this fact is shown in the report title or the running head.

Time reference segment: Specifies the point in time to which the data refer. For example: 1970; 1963 and 1975; 1920 to 1970; First quarter 1974. As indicated, the time reference may be a single date, a time period, or a series of dates or time periods. The time reference is customarily placed at the end of the title or interspersed where it applies. A terminal time reference is set off from the rest of the title by a colon; an interspersed time reference is set off by commas. If the entire report refers to one time period, the time reference may be omitted from the table titles. This type of omission may be confusing, however, and it is generally better to include the date in the title.

A

3.23 Wording of the title.--Table titles are written in telegraphic style, not in complete sentences. Verbs and unnecessary words are omitted. For example, words such as the following are unnecessary: Table showing...; Comparison of...; Relationship between.... table title should be brief but adequate, and a headnote may be used to qualify an otherwise long title. Titles of simple tables can easily be made both brief and complete; for tables consisting of complex classifications, however, it is considerably more difficult.

As a rule, a less complete title which will be read is better than a complete title which will be ignored because of length or complexity. Furthermore, a title may understate the contents of a table but should never overstate.