Analysis 101, part 2: Implementing the analytical plan - Transcript
(The Statistics Canada symbol and Canada wordmark appear on screen with the title: "Analysis 101, part 2: Implementing the analytical plan")
(The Statistics Canada symbol and Canada wordmark appear on screen with the title: "Analysis 101, part 2")
Implementating the analytical plan (Analysis 101: Part 2)
Hi, welcome to analysis 101 video 2. Make sure you've watched video one before you start because we're diving right back in. Now that we've learned how to plan, an analytical project will discuss best practices for implementing your plan.
In this video you will learn how to implement your analytical plan. The key steps in implementing your plan include preparing and checking your data. Performing your analysis. And documenting your analytical decisions.
Steps in the analytical process
(Diagram of 6 images representing the steps involved in the analyze phase of the data journey where the first steps represent the making of an analytical plan, the middle steps represent the implementation of said plan and the final steps are the sharing of your findings.)
In the first video we went through how to plan your analysis. In this video will go through how to implement your plan. If you've been watching the data literacy videos by Statistics Canada, you'll recognize that this work is part of the third step, which is the analyze phase of the data journey.
Step 3: Prepare and check your data
The first step in implementing your plan is to prepare and check your data. Preparing and checking your data will make your analysis more straightforward and rigorous.
Define your concepts
Start by defining your concepts in our previous example that examined the economic status of millennials, we needed to determine how we would define millennials. In the literature, we found no official definition for that generation, but many different recommendations. It's important to make an analytical decision that's meaningful and defendable, and to apply it consistently and documents your decision. In this paper, Millennials were defined as those age 25 to 34 in 2016 in age group that aligns with our typical definition for young workers.
Clean up the variables and the dataset
Now that the concepts are clear, will start digging into the data. Start by cleaning and preparing your data set. You'll want to rename the variables so that they are meaningful an formatted in a consistent manner. For example, rather than using the name Var 3, which is confusing, we rename the variable highest degree earned, which is much clearer. The effort you invest at this step will serve to make your life easier as you proceed with your analysis, especially if you document your decisiones well.
Check your data
(Table of presenting the economic well-being research by generation where the left column represents the generational groups. The middle columns and right column represent the average age in 1999 (Gen-Xers = 26 years old & Millennials = 14 years old) and 2016 (Gen-Xers = 43 years old & Millennials = 66 years old), respectively.)
At this stage, check your data to ensure that it's of the highest quality. For our example, we should check the average age by generational group to make sure there is no issue with how age is calculated. The average age for Generation X is 26 years old in 1999 and in 2016 their average age is 43. This makes sense, however. Well Millennials are 14 years old on average in 1999. They are 66 on average in 2016. In this case we should check our program code, examine the day to fix the error, and document why this error occurred.
Data checks throughout your analysis
To add rigor to your analysis, there are data checks that you should perform at different stages. In the early stages you can check the raw data to ensure that it's clean and ready for analysis. You can also check the frequency distributions of the variables to ensure that the data are consistent with past datasets. Then as you are checking the results of your analysis, you can verify whether your findings are consistent with the literature. All of this work should be done in well documented code that is saved for future reference.
Step 4: Perform the analysis
The second step in implementing your plan is to perform the analysis. As discussed in video one, your analysis should be planned out when creating your analytical plan. So once your data are clean and prepared, you're ready to perform the analysis.
Implementing your plan
Performing the analysis should be straightforward. If you created a clear analytical plan and cleaned and prepared your data appropriately. You should conduct your analysis as planned and as discussed previously, check your results as you go to ensure that the data and methods you are using are producing valid results. Another benefit of checking your results as you go is that you can flag unexpected findings.
If you have unexpected results, this may be due to an error in the data, or it might be some unexpected research finding. Be flexible and adjust your analytical plan to further investigate results that are not in line with your expectations or do not match up with theory. We will see an example of this in the case study video where additional analysis was necessary to disentangle a complex relationship.
Summary of key points
And that is a quick overview of how to implement your analytical plan. This involves preparing and checking your data. And then performing the analysis. Throughout this work, make sure to document your decisions. In the next video you'll be learning about interpreting and sharing your work.
(The Canada Wordmark appears.)