Skip to content

Lesson 15: Tangible Data Merging

Lesson 15: Tangible Data Merging

Objective:

Students will learn how to merge two datasets and ask statistical investigative questions about the merged data.

Materials:

  1. Tangible Data Merging file (LMR_2.14_Tangible Data Merging)

    Advanced preparation required (see Step 4 of lesson)

  2. Copy paper in two colors

    Advanced preparation required (see Step 4 of lesson)

Vocabulary:

merge

Essential Concepts:

Essential Concepts:

We can enhance the context of a statistical problem by merging related datasets together. To merge data, each dataset must have a "unique identifier" that tells us how to match up the lines of the data.

Lesson:

  1. Inform students that they are going to examine the research question "Does the personality color test really work?" To answer this, we're going to examine whether the different color groups actually differ on particular beliefs or attitudes, or if these differences might just be due to chance. In particular, we are going to use the Stress/Chill data to see if there is evidence that the "colors" actually differ.

  2. Show students the variables in each of these datasets. Give students time to brainstorm statistical investigative questions of interest with their teams and record their questions in their DS journals. Encourage them to think of two- and three-variable questions.

  3. Conduct a share out of some of the questions students came up with. Examples include: (1) Do people whose predominant color is Gold tend to stress more than people whose predominant color is Blue? (2) Is there a difference between the sorts of things that stress out the different personality colors?

  4. In order to answer the above questions, we will need to merge our class’s 2 datasets together (Personality Color and Stress/Chill). In order to do this, we will be practicing how to merge datasets today.

  5. Print out the material from the Tangible Data Merging file (LMR_2.14). Use a different color of paper for each of the two datasets. For example, Data Set 1 could be on plain white paper and Data Set 2 could be on blue paper. Cut the paper by creating horizontal strips of each observation of data. For example, from the screenshot below, of the first page of Data Set 1, you would create 12 different strips of paper, one for each observation.

  6. Hand each student in the class a strip of paper. Ask them to try to find someone with the other dataset (i.e., a person with a different colored strip of paper) that they can “match up,” or merge, with.

  7. For example, a student with the first row of data listed below from Data Set 1 might want to match up with the second row of data listed below from Data Set 2 because a person who is 21 has probably graduated high school.

    Birth Month Zip Code Age ID Number Favorite Movie
    January 90064 21 1742 The Notebook
    Zip Code ID Number Birth Month Siblings Education
    91331 1352 August 2 High School
  8. However, they should notice that they cannot just make guesses about a person’s characteristics in order to match up the data. They should realize that only 3 of the variables are the same in both datasets: Birth Month, Zip Code, and ID Number.

    1. Since multiple people have the same Birth Month, discuss why this may not be the best variable to merge with. Multiple people are born in January, so we would have no way of differentiating between those people.

    2. The same is true for the Zip Codes variable. Although there are less repeats with Zip Codes, we still see some overlap between observations.

    3. So, the only UNIQUE identifier in both data sets is ID Number. So the students should end up in pairs at the end of the exercise – a student from Data Set 1 is matched with the student from Data Set 2 that has the same ID Number.

  9. Have the students write about the experience of tangible data merging in their DS journals and ask:

    1. Why is it important to have at least one unique identifier for both datasets? It is the only way to know which information belongs to which person. We want to make sure we do not match up observations (in this case, people) incorrectly because that will compromise any analysis we do later.
  10. Inform students that they will learn to merge datasets using RStudio during the next lab.

Class Scribes:

One team of students will give a brief talk to discuss what they think the 3 most important topics of the day were.

Homework & Next Day

Students will collect data for one more day for the Stress/Chill campaign either through the UCLA IDS UCLA App or via web browser at https://portal.idsucla.org

LAB 2G: Getting it Together

Complete Lab 2G prior to the Practicum.