Considerations

"There are three kinds of lies: lies, damned lies, and statistics"(Disraeli)

Statistics Canada

• About Statistics: Power from Data! - A useful site about statistical procedures and terminology.
• Using the Data - A useful Statistics Canada site which outlines how read and effectively use Statistics Canada tables.
• Statistical Methods - This covers the basic concepts that define Statistics Canada data, the underlying methodology of surveys and key aspects of data quality.

Below is a list of criteria to consider when looking at statistical data

Source

• Always give the source of your figures including the data of collection.
• Could the source be biased?
• Why were the statistics being collected?
• Independent sources may be more reliable than a source producing its own statistics with an interest in the outcome.
• It is hard to get data on private companies.

Quality

• How were the statistics collected - Read the introduction. Was the survey done over telephone, in person, etc.
• What was the sample size and composition? The larger and more random the better.
• Were the questions biased in wording, could the answers have been recorded in a biased way?
• Remember the value of any set of statistics is limited by the reality of the raw data. Sophisticated analysis cannot make up for poor data in the first place.

Date

• Check the date that the information was collected, it is often a number of years prior to the publication date.
• Are the figures annual, quarterly or monthly?
• Remember that in the USA the format is mm/dd/yyyy while in the UK it's dd/mm/yyyy. Canada uses both formats but usually it is dd/mm/yyyy.

Units of Measurement

• What exactly is being measured?
• Check the units of measurement carefully, you may often see '000 at top of columns of figures.
• Is the figure a percentage or a ratio?
• Are the figures "seasonally adjusted" and what does that mean?
• Are figures "indexed"? If so, what does it mean and what was the base for the indexing?

^ Top of Page

Analysis and Presentation

• Know the meaning of common, analytical terms such as mean, average and median.
• Be aware of the implications of scales in graphic presentations - differences can be enhanced or reduced by the astute use of scale. Horizontal and vertical axes should always be labeled.
• How long a time period is being used? Is it long enough?
• If you want to change monthly figures to annual figures should you be summing them or averaging them? (You should not average figures that are already averages.)
• Is the glass half full or half empty?
• Statistics Canada data are subjected to rounding and suppression to maintain confidentiality . Name of individual companies and organizations are never given.

Data From Different Sources

• Check that you are comparing "apples to apples".
• Is the year used calendar, fiscal or some other sort.
• Always read the footnotes at the bottom of a table of figures.
• Remember things like Maritime versus Atlantic provinces, CMA versus GTA, CMA Toronto versus City of Toronto.

Definitions

• Especially important when using census data.
• For example, language - Do you want home language, mother tongue, or first official language spoken.
• Family - In the census there is a census family type and an economic family type.
• Ethnic Origin - Respondents are asked to specify as many groups as applicable. Do you require figures by single ethnic origin or multiple ethnic origin?

IMPORTANT!

Statistical information is covered by copyright and licensing agreements the same as other materials. Users should be aware of the restrictions imposed by licensing agreements that have been entered into by the purchaser of the information. This is particularly important when using Statistics Canada material from CD-ROMs and from data supplied under the Depository Services Program and the Data Liberation Initiative. Such material can only be used for study and research purposes, it MUST NOT BE USED FOR COMMERCIAL PURPOSES OF ANY SORT. At Ryerson this also applies to Census and CANSIM data acquired from CHASS (Computing in the Humanities and Social Sciences, University of Toronto) datacentre.

^ Top of Page