Unit 7 Flashcards
Use lines to:
Examine trends in deviations across interval units
Bullet graph background fill
Typically darker means worse performance in what graph
Bullet Graph in Tableau For Process Data
Shaded Regions Represent 60% and 80% of Target
Color Coding Performance bars is ok or not?
OK
Do not color code regions because why?
Too many colors means nothing stands out!
Few Recommends what with bullet graphs??
Black Bars with Colored Dots to Call out Critical Values
Use ____________for Multiple Entities
Bullet Graphs
DO NOT DO When everything is in color, nothing stands out
Dont do this with the colors on graphs for different units
Try comparing the three sites using the gauges. Ask the following questions:
How much better is the first and the third (in an intuitive sense and numerical sense).
Visually, the gauge does not convey this information well, especially when compared to the bullet/bar graph below.
Use ________graphs for direct comparison
Plot deviations can be used for what
Use __________ or ____________to Show Raw Numbers
Tool tips or labels
Use ____________ to Display the Same Data in Different Ways
Dashboards
use ______,______________,and________- to indicate that lower is better
SORTING, COLOR, and TEXT
But note that sorting ordinal or interval data may confuse the viewer.
Here we sort nominal data, so we are ok.
Ordinal and Interval can still be sorted, but be sure to communicate this to the viewer.
Indicating_________is Better in Bullet Graphs
LOWER
bullet graphs
[reverse background fill (from light to dark)]
[reverse background fill (from light to dark) and the X axis]
Different Measures with Reversed and Normal Scales
For the first two metrics lower is better, so the scales are reversed along with reference shading. Here, I’ve used the darker gray palette from Tableau to highlight good and bad reversal.
Evaluate Visualizations in Context: Note effect of reducing colors
Few Suggests Colored Dots to Highlight Critical Values
Tables: Add Deviation Information and Rank
We can display the same information in a table with the actual deviation shown. Good/bad direction is still not clear here
Use Labels to Clarify Meaning of Negative and Positive Numbers
Consider Reframing Metric to Higher is Better
*30 day Readmission Rate -> 30 day No Readmission Rate
*Or: Percentage of Patients Not Readmitted Within 30 days
Doing this is tricky, because certain metrics are commonly known and reframing can confuse people. Some metrics cannot be reframed, such as Length of Stay.
Deviations of Lower is Better Targets Are Confusing
Visualizing Deviations
*More difficult than it seems at first glance.
*Often no perfect solution.
*As usual, testing with users (even informally) is critical.
Which of the following statements are true about the color coding used in Graph A? Note that for Graph A all Target bars are green and all Units Using Process bars are Red. Select all that apply.
- Color is a redundant code in graph A, because the bars are also labeled with the name of the variable
- Graph A does not effectively use color to call attention to units who have not met their target, because the graph includes both green and red bars regardless of whether the target is met
- Graph A may be somewhat misleading because green typically means good and red bad, so at a glance a user might assume that all the green bars mean good performance and the red bars bad performance
You have a worksheet with four variables: Hospital, Units Using Process (the actual number of units converted to a new process), Target (the number of units that was set as a goal for converting to the new process), and Missed Target (True if a hospital missed–fell below–its target, otherwise False).
In Graph A above all Target bars are green and all Units Using Process bars are red. To create graph A, above, you must
drag Hospital to Columns
Measure Names to Columns
Measure Values to Rows
then color the graph by dragging Measure Names to Color
and finally filter Measure Names so that bars are shown only for Units Using Process and Target.
You have 10 years of data showing the number of children (in thousands) who live in a home with only their mother. The data is grouped by the status of their mother as shown below:
Of the graphs above, which graph or graphs are best for the task of examining which status increased or decreased the most in a given year? In other words, which is best to answer questions like: In 2007 which status saw the largest decrease over the previous year?
Grouped bar chart of annual deviation
Which of the following questions best represents a deviation analysis?
How does patient waiting time for each clinic compare to the national average waiting time?
Right! You want to know how each clinic’s waiting times differ from the national average. The main information of interest here is the deviation—not the raw value.
Select the most accurate statement(s) regarding bullet graphs
- Few recommends using a red dot to the left of the text label to indicate performance measures that did not meet their target.
- To color code the qualitative performance regions, Few recommends using the lighter color intensity for favorable states and the darker color intensity for poor states.
You have a number of units using a new quality improvement process for one year in three different hospitals. Each hospital has a different target number of units that were supposed to adopt the new process during the year. What is one way to normalize values in order to compare the hospitals’ relative performance toward meeting the goal?
Calculate the percentage of goal met by each hospital
You have a number of units using a new quality improvement process for one year in three different hospitals. Each hospital had a different target number of units that were supposed to adopt the new process during the year. Which of the visualizations below are best for comparing 1) the difference in number of units using the process among the hospitals and 2) how many units each hospital is over or under its target? Select all that apply.
C and E
If we want to know which region has the lowest proportion of patients in an age range, what is the most effective graph? For example, consider the question: which region has the lowest proportion of patients in the 41-54 age range?
Grouped Bar Chart
You have a number of units using a new quality improvement process for one year in three different hospitals. Each hospital had a different target number of units that were supposed to adopt the new process during the year. Which of the visualizations below are best for comparing the normalized performance of the hospitals toward meeting their targets? Select all that apply.
D
and C?
You have 10 years of data showing the number of children (in thousands) who live in a home with only their mother. The data is grouped by the status of their mother as shown below:
Consider the following graphs of year-to-year deviation by each status. This means that each bar or point in the graph shows the difference between the previous and the current year. For example, if you look at the data for Divorced mothers in 2013, it is -133, which means that the number of children living in a mother-only home with a divorced mother dropped by 133,000 from 2012 to 2013.
Of the graphs above, which graph or graphs are best for the task of examining trends in year-to-year deviations among a single status? In other words, which is best to answer questions like: is the change in children living with divorced mothers going up, down, or staying the same over the 10 year period? In this case, we are NOT interested in comparing across status, but instead we want to look at a single status at a time to examine trends in change for that status.
NOT?
Small multiples line chart of annual deviation
Grouped bar chart of annual deviation
In graphs B, C, and E below, hospitals that missed their targets are colored red. Which of the following methods will work to ensure the correct color coding even if the underlying data changes (either the performance or comparative measure)?
Define a calculated field, Missed Target, using the formula: [Units Using Process] < [Target]
Correct, this will be true for hospitals that failed to meet their individual targets.
You have a number of units using a new quality improvement process for one year in three different hospitals. Each hospital has a different target number of units that were supposed to adopt the new process during the year. What is one way to normalize values in order to compare the hospitals’ relative performance toward meeting the goal?
Calculate the percentage of goal met by each hospital
Yes, this normalizing all hospital performance metrics to a single comparable number.
Suppose you have connected Tableau to the following data file:Before you can create a Bullet Graph in Tableau that displays both a current performance metric and projected future (year end) performance, such as you must define a calculated field, Expected Remaining, with the formula:
(([Units Using Process]/[Months])*12)-[Units Using Process]
The most direct way to display deviations is by calculating and plotting (such as on a bar chart) the difference between the comparative measure and the performance measure. For example, we could calculate and plot the target number of units - actual number of units. However, when we do this, the underlying raw data is not visible on the display. Which of the following are effective ways to supply this data? Select all that apply.
*For interactive visualizations, provide a tooltip on each performance measure and target that shows the raw performance and comparative measures.
- Use labels to overlay the raw data on the visualization
You have data showing the rate of hospital acquired infections for each of 20 units for the past 12 months. You would like to show how each unit ranks (with lowest rate being best) each month so that you can see how the ranking changes over time. What is the best chart to show change in rank by month?
A bump chart with month on the X axis and rank on the Y axis
When you have a single bar chart showing percent of total for a number of items (say percent of patients for 10 different hospitals) you can best help the viewer determine their relative ranking by
Sorting the bars from lowest to highest (or highest to lowest) percent of patients
You create a dynamic dashboard in Tableau that uses a bar graph to show a hospital’s readmission rate for each month. The data is subject to revision up to several months later as clinical records are updated. You would like the months in which the readmission rate goal of 2% was exceeded to use red bars and the other months to be in blue. What is the best way to do this in Tableau so that if the data changes the coloring will reflect these changes?
Create a calculated set that includes months where the readmission rate is above 2%
At the most fundamental level, analytical navigation can be divided between two approaches: directed and exploratory. Which statement about these two approaches is true?
All of above:
Exploratory analysis begins by looking at the data without predetermining what you expect to find, proceeds to noticing things in the data that are interesting and asking a question about it, then proceeds in a directed fashion in search of an answer to that question
Directed analysis begins with a specific question that you wish to answer, proceeds to a search specifically for the data that will answer that question, such as a particular pattern, and hopefully results in finding the answer.
Comprehensive analysis requires that you sometimes start with a blank slate and let the data itself direct you to items worth examining.
Why do we need to provide an overview first?
All of above
It helps to give an overall sense of patterns in the data
It helps the user decide the next move in information retrieval
It helps in deciding where the important details lie in a graph
You have a data set in Excel with two columns: Hospital and LOS. Each row shows the average length of stay (LOS) for a given hospital for one year. The target LOS for all hospitals was 3, but that information is not in the Excel file. Which of the following could you do so that you could display the target in a Tableau Bullet Graph? Select all that would result in a target at the correct location.
- Add a column in the Excel file, called Target and fill it with 3.
- Create a calculated field, called Target, with the formula: 3
Select all of the statements below that are true of bullet graphs
- Bullet graphs were designed by Stephen Few to replace meters and gauges that are often used on dashboards.
- Bullet graphs use a symbol marker to code a quantitative measure and indicate its distance from a comparative measure.
What do you need to do in Tableau to create the line portion on the Pareto chart above for Insurer Cost, such that it shows a running sum of % of Insurer Cost?
Apply a quick table calculation, Running Total, to Insurer Cost and then Edit the table calculation and perform a secondary calculation: Percent of Total summarizing values from Disease. Finally, right click the rightmost axis and select Synchronize Axis.
What is the most useful, frequently used process in interaction?
Comparing and Contrasting
