The Do s and Don ts of Data Visualization Meghal Parikh and Sandra Archer University of Central Florida University Analysis and Planning Support Southern Association of Institutional Research Atlanta, GA Oct. 10, 2011
How to NOT present your data. Example 1 Showing off improvement in your work Source: The Maddow Blog (http://maddowblog.msnbc.msn.com) Kent Wells, the senior vice president of BP, in a technical briefing video on BP s website, shows how their group is continually improving and increasing their performance in oil collection efforts in the seas. What s wrong? The chart is cumulative number of barrels of oil collected by date Oct. 10, 2011 The Do's and Don'ts of Data Visualization 2
Blogger Stephen Frew (www.perceptualedge.com) shows how BP s increasing bars does not mean improving performance. Source: Visual Business Intelligence: A blog by Stephen Few (http://www.perceptualedge.com) He plotted the same data by daily oil collection rate; the story is much different than shown by Kent in the video. Oct. 10, 2011 The Do's and Don'ts of Data Visualization 3
How to NOT present your data. Example 2: Perceived versus Actual substance use at U. of California, Santa Barbara The problem here does not need much explanation. Clearly, 0.7% column cannot be half as tall as 29.4% column because at least mathematically 0.7 is not half of 29.4 Oct. 10, 2011 The Do's and Don'ts of Data Visualization 4
Below is the geometrically improved version of this charting disaster And a better way of explaining the same data Source: Oops, but how to best present these data: A blog by Alex Kerin (http://www.datadrivenconsulting.com) Oct. 10, 2011 The Do's and Don'ts of Data Visualization 5
How to NOT present your data. Example 3: FOX news Chicago makes a GOP candidate support pie chart Source: All 193% of Republicans Support Palin, Romney and Huckabee (http://wonkette.com) No explanations needed for the problem here 63 + 70 + 60 100 Oct. 10, 2011 The Do's and Don'ts of Data Visualization 6
Lessons Learned Data Visualizations are simplified representations of a complex reality used to influence others. We are not here to convince the decisionmakers. Institutional researchers are here to guide decision-makers to better, data-driven decisions. GEOMETRY MATTERS; same scale, same diameter, correct Y-axes etc. LEAVE SPECIAL EFFECTS, PRETTY COLORS & 3-D GRAPHICS to HOLLYWOOD. 3D looks cool but doesn t change the perspective of the decision maker. USE TWO-DIMENSIONAL CHARTS. Find the message you want to show from the data. Consider your audience. Important note: 1 of 10 men are color blind Prepare your data. And don t have blind faith on your data visualization software. Oct. 10, 2011 The Do's and Don'ts of Data Visualization 7
1995-96 1996-97 1997-98 1998-99 1999-00 2000-01 2001-02 2002-03 2003-04 2004-05 2005-06 2006-07 2007-08 2008-09 2009-10 Look at data in a different way Example : Trend data for Baccalaureate Degrees Awarded in the State of Florida Story to tell: Growth of UCF in terms of Baccalaureate degree production is much higher than the rest of state 90,000 Growth in Baccalaureate Degrees Awarded in Florida 80,000 70,000 60,000 50,000 40,000 30,000 20,000 10,000 0 Number of Baccalaureate Degrees Awarded Source:IPEDS Oct. 10, 2011 The Do's and Don'ts of Data Visualization 8
Look at data in a different way 90,000 80,000 Growth in Baccalaureate Degrees Awarded in Florida 83,499 9,969 70,000 60,000 50,000 40,000 UCF 46,327 4,930 43,419 30,000 Other SUS 27,280 20,000 30,111 10,000 Other 14,117 0 1995-96 2009-10 First majors only. Other includes private, for-profit, and state colleges. Source:IPEDS Oct. 10, 2011 The Do's and Don'ts of Data Visualization 9
Look at data in a different way Example 1: Comparing Retention Rates among different student segments Story to tell: How first year retention rates for transfer students compared to freshmen? Oct. 10, 2011 The Do's and Don'ts of Data Visualization 10
Look at data in a different way Looking at first year retention rates did not provide an adequate comparison, since we are comparing Freshmen retention to Junior retention. 6,000 FTIC Retention (Summer-Fall 2004 Entering Cohort) - Retention of FTICs from 5,000 their third to fourth year 4,000 3,000 2,000 100% 94.9% 92.2% 92.2% 91.1% Grad Enroll is 94.9%, while retention of transfers from their 1,000 - first to second year is (Enter UCF) Second Third Fourth Fifth Sixth Seventh 79.6%. 3,000 2,500 AA and AS Transfer Retention (Summer-Fall 2006 Entering Cohort) - Not all transfers enter upon their third year of 2,000 1,500 100% 1,000 79.6% 74.4% 72.6% 71.7% 500 0 First Second (Enter UCF) Fourth Fifth Sixth Seventh Source: UCF IR Retention Reports; Summer and Fall All Cohorts, FTIC and CCT with AA or AS Grad Enroll post-secondary enrollment. - The year labeling is used here for alignment only. Oct. 10, 2011 The Do's and Don'ts of Data Visualization 11
Contact Information Dr. Sandra Archer Director (Sandra.Archer@ucf.edu) Mr. Meghal Parikh Analyst/Programmer (meghal.parikh@ucf.edu) University Analysis and Planning Support University of Central Florida 12424 Research Pkwy, Ste 215 Orlando FL 32826-3207 Phone: (407) 882-0285 The mission of the office of University Analysis and Planning Support (UAPS) is to enhance the management capability within the University of Central Florida (UCF) by providing models and information to support and empower academic units, administrative units, and external stakeholders to utilize analysis and research results as the cornerstone for informed decision-making. Oct. 10, 2011 The Do's and Don'ts of Data Visualization 12