Language selection

Search

Evidence Informed Decision Making

Data11-17-2021
Description: girl looking at wall with multiple papers
Photo by Brandon Lopez / Unsplash

It should feel fairly obvious that we inform our decisions by having data and evidence to back them up; but it can feel a bit overwhelming depending on the topic. We’ll go through a few ways to ensure you’re practicing this delicate art correctly.

Firstly, providing evidence to decision-makers means giving sound, logical advice supported by credible evidence in a way that suits the audience. This means being able to extract and present insights from different sources and requires critical thinking to leverage the best sources and interpret the context.

It’s important to note that data doesn’t always include an analysis, this means that when you’re looking for evidence, you may have to interpret raw data by looking at sample sizes, algorithms, approaches etc. I know, it sounds daunting but we’ll get to this a little later.

At the heart of our roles is the need to provide unbiased, comprehensive, accurate and considered advice to our senior officers and the government of the day. We do this when we are:

  • Reporting on performance
  • Seeking a decision
  • Evaluating a program
  • Making recommendations
  • Writing a brief
Description: information stand
Photo by Philip Strong / Unsplash

We need evidence!

To provide successful advice, we need evidence. What is evidence? The true definition is the facts, signs or objects that make you believe that something is true. We can also understand evidence by what it gives us. For example, we can think critically and scrutinize how it directly impacts our work and the context we are using it for. You should be asking the following questions

Will a source of information…

  • Add objectivity?
  • Add credibility?
  • Assist in evaluating alternatives?
  • Inform about the unexpected yet be trustworthy?
  • Strengthen an argumentation?
Data, if it’s not accurate, if it’s not timely, if it’s not kind of current, can be dangerous. Pause, take a deep breath and go okay, is this data accurate? Is this data credible? Is this data current? So pause and think about that for a moment before you run off with a piece of information. Make sure it’s real time actionable data that you can do something with.
- Taki Sarantakis, President, Canada School of Public Service
Description: magnifying glass
Photo by Clément Falize / Unsplash

Interpreting evidence

We need our evidence to come from reliable and trustworthy sources to inform correct decisions, policies, etc. It can be difficult to know which sources to trust, so I've listed a few places to start your search:

  • Research or administrative data
  • Expertise, experience and knowledge of stakeholders (this may include government sources)
  • Scholarly, peer-reviewed articles or books
  • Contextual information

Interpretation:

This is the tough work and requires, patience, focus and often a ton of research and background knowledge. Interpretation is also where humans play a vital role in the data space. It's our responsibility to develop meaning, and explanations for the data we've collected.

You'll want to:

  • Analyze the data to extract meaningful insight
  • Consider the experiences, preferences, and values of local stakeholders
  • Assess the influence of the location and population on choice of solutions
  • Learn from the experiences of others

Applying what we’ve learned

Think about explaining what you’ve learned without bias and present all sides of the research. This will help create an informed decision while considering all factors.

  • Consider the sources of evidence
  • Try to find the best possible practicable solution
  • Engage with stakeholders throughout the process
  • Develop adapted products to inform decision-makers
Description: newspaper
Photo by Roman Kraft / Unsplash

Information is not always evidence

Not all evidence is equal

It can be difficult to discern information from evidence. At this point, you’re probably asking what the difference is. Well, would you consider all types of information as evidence useful for decision making? All the sources listed below provide information, but the key is to discern and choose the appropriate information for your purpose. Not all evidence is equal, and recommendations may change as evidence quality evolves. The best way to approach this is for stakeholders to be engaged in defining your goals and reviewing the consideration of evidence for decisions. Evidence starts a conversation, it doesn’t end it. Consider:

  • Eyewitness accounts
  • Scientific measurements and experiments
  • Survey results
  • Program evaluations
  • Computer modelling
  • Expert opinion
  • Legal precedent
  • A panel discussion on TV
  • Discussion in an internet forum (social media, comments page, etc.)
  • Newspaper or TV news article
  • Advertising
  • Arguments by lobbyists

Evaluating evidence

Data plays a great part in evidence. Some data is collected as part of an administrative task, while other data can be collected during studies by academics and during program evaluation. To evaluate evidence, we need to evaluate the studies and processes behind it. How do we know whether these data are of good quality? Statistics Canada has developed 6 dimensions of data quality to help us answer this question.

This video is intended for learners who want to acquire a basic understanding of data quality. No previous knowledge is required.
Transcript

Data quality in six dimensions: Evaluating and ensuring quality

We are exposed to data every day. For example in news stories, weather reports and advertising. But how do we know whether these data are of good quality? In this video, you will be introduced to the fundamentals of data quality, which can be summed up in 6 dimensions or six different ways to think about quality. You will also learn how each dimension can be used to evaluate the quality of data.

Learning goals

By the end of this video you will learn about basic quality concepts, data quality expressed as 6 dimensions and the interactions between these dimensions. This video is intended for learners who want to acquire a basic understanding of data quality. No previous knowledge is required.

Steps in the data journey

(Diagram of the Steps of the data journey: Step 1 - Find, gather, protect; Step 2 - explore, clean, describe; Step 3 - analyze, model; Step 4 - tell the story. The data journey is supported by a foundation of stewardship, metadata, standards and quality.)

(Text on screen: The steps in the data journey are supported by a foundation of stewardship, metadata, standards and quality)

This diagram is a visual representation of the data journey from collecting the data to cleaning, exploring, describing and understanding the data to analyzing the data, and Lastly to communicating with others the story. The data tell. You will notice that data quality does not fall under one specific step in the process. It is instead something that is important throughout the entire data journey.

Quality

(Diagram of the six dimensions of data: Relevance; Accuracy, Timeliness, Interpretability, Coherence, Accessibility)

The six dimensions of data quality are: Relevance, Accuracy, Timeliness, Interpretability, Coherence, Accessibility. Each dimension will be examined separately over the next few slides.

Relevance

The relevance of data or statistical information reflects the degree to which it meets the needs of data, users and stakeholders to test a data product for relevance, you should ask yourself, does this information matter? At Statistics Canada, it is our responsibility to provide Canadians with information that matters. In other words, is it useful in building policy? Does it aid in long term planning? Does it fill an existing data gap? Can it promote new initiatives that would benefit Canadians? Does it help improve services? What questions would you ask to test the relevance of your data?

Accuracy

Accurate data give a true reflection of reality. Ask yourself if what is being measured is in line with what is actually true.

Timeliness Timeliness is the delay between the time when the data are meaningful and when they are available. For example, school bus authorities need UpToDate weather forecasts very early in the morning to make good decisions about whether to cancel school buses. Likewise, parents need to know about school bus cancellations before they had to work. Timeliness is closely related to accuracy and relevance.

Interpretability

Information people can't understand or can easily misunderstand has no value and could even be misleading. To avoid such misunderstandings, supplementary information, or documentation, called metadata should always be provided with any data set as it allows users to interpret the data properly.

Coherence

Coherence can be split into two concepts, consistency and commonality. Consistency means using the same concepts, definitions and methods overtime. Commonality means using the same or similar concepts, definitions and methods across different statistical programs. If there's good consistency and good commonality, than it is easier to compare results from different studies or track how they stayed the same, or change overtime with regards to data quality. Coherence is the ability to make comparisons across cities, regions, time periods, etc.

Accessibility

The final dimension of quality is Accessibility, which means that people are aware of and have access to the data. When determining whether data are iaccessible, make sure they are organized a system or a catalog is in place to allow the users to locate all available data available. Once the location of a data source has been determined, a consistent means of accessing these data must also be provided.

Accountable, a data producer must be accountable for assisting users experiencing difficulty or dissatisfaction with any aspect of data access affordable. What good are the most reliable data? If you can't afford to use them?

Applying the dimensions of quality

Imagine that you own a pizza shop and you are considering expanding your business by opening a second location in the Toronto area. What kind of data could help you make your decision, and where might you find such information?

(Text on screen: The types of questions that could help you expand in Toronto are: What kind of data could help you to decide whether to open a second location? Where might you find such information at a relatively low cost? How could you ensure that the data are accurate, timely, interpretable and coherent?

Relevant data

Opening a second location in Toronto would require social and economic data about the city, including neighborhood profiles, business expansion, and location assistance, employee data in household spending habits. Grants, incentives, and rebates, festivals, events, parks and beaches, municipal development plans.

Accessible data

Being able to access reliable data helps inform your decision of whether to open a second location and to assess its potential growth overtime. Ideally this information would be well organized and readily available at little to no cost from reliable open data sources such as: the federal government's open data site, the city of Toronto's open data portal, the Ontario Ministry of Finance and newspapers.

What makes these sites so accessible? They have many features, including open by default, menu driven apps Gallery, open government licenses, open data inventory, application programming interface API and content in both official languages, federal and provincial sites.

(Text on screen: Access to the aforementioned portals: the federal government's Open Data site: https://open.canada.ca/; the Ontario Ministry of Finance: https://www.fin.gov.ca/; the City of Toronto's Open Data Portal: https://www.toronto.ca/

Accurate data Accurate data allow you to make precise calculations about your expected costs. An earnings as well as about the potential success of any new restaurant operation. The success of your new restaurant operation will depend on U preparing accurate financial projections based on solid research, planning and good quality data.

Timeless data (Table titled: Historical and projected population by census division, selected years - reference scenario)

Data tend to be of greater value when they are released at a consistent, favorable or useful time. The release of projected population data by region gives restaurant tours a sense of which areas are likely to experience population growth.

Interpretable data

There are several ways in which these open data sites make it easier to understand and interpret their data. They apply a structured, standardized format or user friendly interfaces. They provide the user with a consistent way to access, view, and understand the data. They incorporate a variety of data into a single visualization tool to make them easy to interpret.

Documentation and supplementary information are readily available to help provide context around the datasets, notes, Footnotes, and sources appear within the table. The site makes use of data visualization tools, tables, info, graphics charts which make it easier to interpret the data.

Coherent data (Image of the Socioeconomic highlights from the 2016 census for the Scarborough centre (Toronto Ward 21))

Comparative measures of employment rates, income levels, and education are important indicators of economic outlook and the potential success of any new restaurant operation. The city of Toronto's open data portal has predefined views with built in coherence analysis. Each view allows the user to compare the data for each Ward with those for the entire city as well as those for other wards in a single visualization tool.

Summary of key points

Data can be a very powerful decision-making tool, but when used improperly they can be misleading by applying the six dimensions of quality, you can choose a high quality data source that's right for your needs.

An acceptable level of quality can be achieved by ensuring that there is a good balance among all six dimensions relevance, accuracy, timeliness, interpretability, coherence and accessibility.


Consider

Relevance - The relevance of data or statistical information reflects the degree to which it meets the needs of data, users and stakeholders to test a data product for relevance. Ie. Does it matter?

Accuracy - Accurate data give a true reflection of reality. Ask yourself if what is being measured is in line with what is actually true.

Timeliness - is the delay between the time when the data are meaningful and when they are available. For example, school bus authorities need up to date weather forecasts very early in the morning to make good decisions about whether to cancel school buses.

Interpretability- Information people can't understand or can easily misunderstand has no value and could even be misleading. To avoid such misunderstandings, supplementary information, or documentation, called metadata should always be provided with any data set as it allows users to interpret the data properly.

Coherence - can be split into two concepts, consistency and commonality. Consistency means using the same concepts, definitions and methods overtime. Commonality means using the same or similar concepts, definitions and methods across different statistical programs. If there's good consistency and good commonality, than it is easier to compare results from different studies or track how they stayed the same, or change overtime with regards to data quality. Coherence is the ability to make comparisons across cities, regions, time periods, etc.

Accessibility - which means that people are aware of and have access to the data.

Conclusion

Ultimately, data makes up much of our evidence. It is our responsibility to interpret and evaluate the data with a critical lens. With each passing year, the Government of Canada increases our data holdings and rapidly improving technology allows us to do more with it than ever before. In considering how to use evidence, we need to understand the data to unlock the value and create evidence informed decision making.


Resources:

Apolitical | To understand evidence use, understand the goals of decision makers

Statistics Canada | Data Accuracy and Validation: Methods to ensure the quality of data

Statistics Canada |Data Learning Catalogue

Learning Path: Discover Data
Get familiar with data concepts, terminology and practices, and gain an understanding of the changing citizen expectations of government.
A Closer Look at Women in Data
Data is (or should be) how we assess needs and should inform how we subsequently allocate resources for women. That isn’t always the case.
Data Partnerships: Unleashing the value of IRCC Data to inform public policy
At IRCC, we leverage our data assets through building strategic data partnerships, which significantly increase our policy relevance evidence base and fill important data gaps.
Mackenzie Kitchen

Mackenzie Kitchen

Product Owner with Digital Academy working on data and AI learning. | Responsable de produit avec l'Académie numérique travaillant sur les données et l'apprentissage de l'IA.

Recommended for you

Topic: Discover Series

Learning Path: Discover Cloud

Get familiar with the opportunities, applications and limitations of cloud computing.

8 months ago2 min read

Topic: Discover Series

Learning Path: Discover Human-Centred Design

Get familiar with human-centered design principles and as a result, speak confidently to your colleagues about it and become curious to learn more.

8 months ago2 min read

Topic: Discover Series

Learning Path: Discover Digital Leadership

Develop some shared understanding as to what "a digital government" means and the leadership that is required to enable it in the Public Service.

8 months ago2 min read