Access to personal study rooms and computer workstations will be closed during the 4-week province wide emergency shut down beginning Monday, April 5th. Contactless Pickup and Digital Delivery will continue to be available Monday through Friday. 

Remote hours of operation are 8:30 am - 4:30 pm from Monday to Friday. Please refer to our homepage and Frequently Asked Questions for the latest updates on our remote services, accessing the library building, and hours of operation. Be sure to login to your account to access all of our resources from off-campus. 


Data for 85-222, Treatment of Experimental Data

Please look over these notes before you begin. Jump to the data sources here.

For further assistance, email or come to the data centre during open hours.

Types of data

There are two main types of data you will be using. Time series data is data that has repeated observations on something such as a country or region over time - for example, Canada's GDP annually. Cross-sectional data is data on multiple units at the same time - for example, a survey where 500 people were asked the same questions, or the population of 100 countries in 2003.

How many observations

You need many observations to conduct an analysis. For a bivariate analysis (an analysis with two variables where you use one variable to predict the other) 30 observations may be enough. 20 is iffy, 10 ridiculous.

While it is mathematically possible to run statistical procedures with fewer observations, you are unlikely to get useful results, and your professor is unlikely to accept your paper.

With annual time series data it can be difficult to get enough observations - many time series don't go back more than 20 years or so. However, some time series are available monthly, giving you 120 observations in 10 years, which is plenty. With cross-sectional data you need enough units - data on Canada's 10 provinces would not be enough, but data on 100 Canadian cities would.

Frequently data - particularly international data - will have missing observations.

Combining data

Sometimes you can combine data from different sources to get the variables you need. With time series data you need to make sure the time periods match up. With cross-sectional data on countries or regions this can also work. With survey data this is not possible, as there is no way to make different groups of people match up.

Environmental and scientific data

Automotive, transport, related

Finance and economic data


Archives of datasets specifically prepared for teaching and learning
These datasets are usually ready-to-use but may be limited in scope. Most include some data with an engineering or experimental focus.

Still looking? Email or come to the data centre during open hours.