Data & Statistics

The two terms are often used interchangably. Although there are some commonly understood distinctions, there are also grey areas: statistics are certainly a kind of data, and data are used to generate statistics. 

Statistics are:

• facts or figures
• time series
• tables, charts, or graphs
• to support an argument
• 'ready to use'

Data can be used to:

• test hypotheses
• generate custom tables
• look at responses of individuals
• analyze in SPSS, SAS, or Stata
• do Regression, t-test, ANOVA, etc

• Microdata is the original, unprocessed (except to protect privacy of participants) information: for example, income reported by each household, height and species of each tree in a park.
• Aggregate data is summarized and combined in some way: average income in a census block, number of oak trees in a city park.

First things first: slow down! Take time to understand what it is you are really trying to gather.

Don't focus on the numbers in the table right away. Instead, carefully review the details around the edges, such as:

• What information is given by the title or header?
• What are the row and column labels?
• Are there any footnotes or references underneath the table?

All of this information can help you understand the context of the information contained within the table.

Questions to ask (and answer!) when looking at numerical data or statistics:

• What's being counted or summarized?
• What units are being used: thousands of dollars (CAD? USD?), individual spectators, percentage change (from what?), percentage of total, etc.
• Who collected and/or summarized the data?
• What questions were asked or what sources used to find, solicit, compile, collect, create the data?
• What was the purpose of collecting the data in the first place?
• How does all of that fit with what YOU want to do with these numbers?

Many factors can affect what data is collected and why. Other factors affect what can be shared with others once data has been collected. A few common issues that arise with published data and statistics are outlined below.

Privacy: If the population being counted is small and there is a possibility individuals or businesses may be identifiable, there are legal and ethical requirements to anonymize data before sharing it.

Methods: If there are concerns regarding the methods used to collect the data or if it is not possible to confirm the accuracy of the data, it might not be made available. Some statistical calculations require specific criteria to be considered valid, for example, if the number of data points is too small, or if the method of obtaining the data was inconsistent, the statistic calculation is not considered to be accurate and may not be published.

Mandated Measurements: Many surveys, including the national census, are required by law or regulation. In some cases, the specific questions and responses collected are explicitly outlined by a government agency or ruling, such as with Statistics Canada. These regulations can change over time, so the questions asked 10 or 20 years ago might not be the same as those asked today. Consequently, comparing data over time may not be so straightforward, or, in some cases, not possible.

Pre-existing Categories: Standardized methods and categories are often used by many groups to easily share and compare data sets and statistics. It is convenient to use these standards, but they might not perfectly match your specific question.

• For example: NAICS codes are commonly used in Canada and the US to collect economic and labour statistics based on industry. Each specialized industry will have a single NAICS code that is a subset of a larger category, which is in turn part of an even larger category, etc. The hierarchical arrangement defined by NAICS might not always fit the way you would like to categorize the industry. These codes can also change over time, something to keep in mind if you are looking at statistics from different decades: NAICS 1997 had three codes for internet-related industries. NAICS 2012 has 57.

There is always a chance an existing dataset does not meet your needs, especially if you trying to address a highly specific gap. This might serve as motivation for you to collect data yourself, or it may mean that you find a proxy variable/dataset. Regardless, it always means patience and persistence when data seeking, and even, at times, the flexibility to alter your research question or project focus given data availability.