EFFECTIVE SCIENTIFIC PRESENTATIONS

Transcription

1 EFFECTIVE SCIENTIFIC PRESENTATIONS Dan Seaton Royal Observatory of Belgium Space Science Training Week KULeuven 2013 September 16 Why does most scientific communication go wrong? Often, the problem is that PowerPoint templates reduce nuance and cause speakers to try to fit their information into inappropriately narrow constraints.

2 Slideware often reduces the analytical quality of presentations. In particular, the popular PowerPoint templates (ready-made designs) usually weaken verbal and spatial reasoning, and almost always corrupt statistical analysis. What is the problem with PowerPoint? And how can we improve our presentations? Edward Tufte, The Cognitive Style of PowerPoint (2006) Slideware can help speakers to outline their talks, show visual materials, and to communicate slides in talks, printed reports, and internet. But careless use can obscure meaning and context, inhibit deep analysis, or worse, distort the truth. How can we avoid this?

3 A CASE STUDY: SPACE SHUTTLE COLUMBIA

4 NASA (2003) COLUMBIA Broke up on re-entry 1 February 2003 The Columbia broke up because insulating foam broke off its external fuel tank and struck left wing during launch, damaging thermal tiles needed to withstand extremely hot conditions during re-entry. Mission managers were aware of the risk during the flight, and Boeing engineers made a presentation on their risk analysis.

5 Review Of Test Data Indicates Conservatism for Tile Penetration The existing SOFI on tile test data used to create Crater was reviewed along with STS-107 Southwest Research data Crater overpredicted penetration of tile coating significantly Initial penetration to described by normal velocity Varies with volume/mass of projectile(e.g., 200ft/sec for 3cu. In) Significant energy is required for the softer SOFI particle to penetrate the relatively hard tile coating Test results do show that it is possible at sufficient mass and velocity Conversely, once tile is penetrated SOFI can cause significant damage Minor variations in total energy (above penetration level) can cause significant tile damage Flight condition is significantly outside of test database Volume of ramp is 1920cu in vs 3 cu in for test 2/21/03 6 Columbia Accident Investigation Board, Report Vol. 1, p. 191 (August 2003) This is a critical slide from the Boeing presentation. Note that Crater is the impact model used to evaluate the damage and SOFI means Spray-On Foam Insulation.

6 What is the most significant statement in the slide?

7 Review Of Test Data Indicates Conservatism for Tile Penetration The existing SOFI on tile test data used to create Crater was reviewed along with STS-107 Southwest Research data Crater overpredicted penetration of tile coating significantly Initial penetration to described by normal velocity Varies with volume/mass of projectile(e.g., 200ft/sec for 3cu. In) Significant energy is required for the softer SOFI particle to penetrate the relatively hard tile coating Test results do show that it is possible at sufficient mass and velocity Conversely, once tile is penetrated SOFI can cause significant damage Minor variations in total energy (above penetration level) can cause significant tile damage Flight condition is significantly outside of test database Volume of ramp is 1920cu in vs 3 cu in for test 2/21/03 6 Columbia Accident Investigation Board, Report Vol. 1, p. 191 (August 2003) How do you know what information is important when there are So many levels and so much information, not all of it critical?

8 Review Of Test Data Indicates Conservatism for Tile Penetration The existing SOFI on tile test data used to create Crater was reviewed along with STS-107 Southwest Research data Crater overpredicted penetration of tile coating significantly Initial penetration to described by normal velocity Varies with volume/mass of projectile(e.g., 200ft/sec for 3cu. In) Significant energy is required for the softer SOFI particle to penetrate the relatively hard tile coating Test results do show that it is possible at sufficient mass and velocity Conversely, once tile is penetrated SOFI can cause significant damage Minor variations in total energy (above penetration level) can cause significant tile damage Flight condition is significantly outside of test database Volume of ramp is 1920cu in vs 3 cu in for test 2/21/03 6 Columbia Accident Investigation Board, Report Vol. 1, p. 191 (August 2003) The most important point is that the model tested an impact by a projectile 640 times smaller than the piece of foam that struck the wing during launch. The post-accident report itself acknowledged the problems with this slide later on.

9 As information gets passed up an organizational hierarchy, from people who do analysis to mid-level managers to high-level leadership, key explanations and supporting information are filtered out. In this context it is easy to understand how a senior manager might read this PowerPoint slide and not realize that it addresses a life threatening situation. Columbia Accident Investigation Board, Report, vol. 1 (August 2003)

10 Could this slide be improved?

11 Review Of Test Data Indicates Conservatism for Tile Penetration The existing SOFI on tile test data used to create Crater was reviewed along with STS-107 Southwest Research data Crater overpredicted penetration of tile coating significantly Initial penetration to described by normal velocity Varies with volume/mass of projectile(e.g., 200ft/sec for 3cu. In) Significant energy is required for the softer SOFI particle to penetrate the relatively hard tile coating Test results do show that it is possible at sufficient mass and velocity Conversely, once tile is penetrated SOFI can cause significant damage Minor variations in total energy (above penetration level) can cause significant tile damage Flight condition is significantly outside of test database Volume of ramp is 1920cu in vs 3 cu in for test 2/21/03 6 Columbia Accident Investigation Board, Report Vol. 1, p. 191 (August 2003) What is the pertinent information here? There was a real risk of penetration Extensive damage was possible The simulations looked at a scenario that was nowhere close to reality

12 Test data show damage is possible, but test models are not applicable A SOFI particle can penetrate tile coating at high energy cause major damage after penetration Columbia flight is way out of tested range (one fragment estimated at 1920 in 3 vs. 3 in 3 for tests) Adapted from Doumont, Tech. Comm. (February 2005) One proposal for a better version of this slide might look like this.

13 What is the lesson here?

14 What comes out of a talk depends on what goes into it. What I mean here is that what the audience takes away depends strongly on how much effort goes into preparing the talk and the slides. If you make little effort the audience will probably take away very little meaning.

15 This talk is mainly about making better slides. During this talk, I ll assume you know most of the fundamentals of public speaking: Face the audience as much as humanly possible Avoid distractions (don t wave pointer all over, for example) Speak clearly and loudly enough Make notes about what you need to remember your slides should distill information, not serve as notes

16 FOUR CRITICAL ERRORS THAT CAN RUIN A TALK

17 If your content disorganized, the audience will not be able to follow the talk or determine what is new, what is context, and what the problem and solution are. The same goes for your slides. Remember also that PowerPoint is a visual tool, use it effectively: present complex info or complement/emphasize spoken presentation, which should be the focus of your slides. Finally, note that your audience gains nothing from a talk that is not appropriate for the level, setting, etc. 1. Failure to organize the information presented 2. Unreadable or poorly organized slides 3. Unclear, unreadable, or inefficient visuals 4. Failure to understand your audience

18 1. Failure to organize the information presented 2. Unreadable or poorly organized slides 3. Unclear, unreadable, or inefficient visuals 4. Failure to understand your audience We have already examined the first two points in our case study on the Columbia.

19 1. Failure to organize the information presented 2. Unreadable or poorly organized slides 3. Unclear, unreadable, or inefficient visuals 4. Failure to understand your audience Let s look at what can go wrong with a slide.

20 Here s a real slide from a talk I once attended.

21 What is being conveyed in this figure?

22 It s hard to know what s being conveyed. Why? There are several reasons: Both space and color are used confusingly. An audience might ask why there are bullets on right and left and are they related? They might ask what is meant by the colored text? The biggest problem is that the author has tried to shoehorn too much information onto a single slide, rendering the figures unreadable.

23 Is there a better way to convey complex information?

24 90N SUNSPOT AREA IN EQUAL AREA LATITUDE STRIPS (% OF STRIP AREA) > 0.0% > 0.1% > 1.0% 30N EQ 30S 90S DATE 0.5 AVERAGE DAILY SUNSPOT AREA (% OF VISIBLE HEMISPHERE) DATE Figure 8: Sunspot area as a function of latitude and time. The average daily sunspot area for each solar rotation since May 1874 is plotted as a function of time inhathaway, the lowerliving panel. Reviews The relative Solar Phys., area7 in (2010) equal area latitude strips is illustrated with a color code in the upper panel. Sunspots form in two bands, one in This eachis hemisphere, one example that of start a figure at about I think 25 from is very the equator successful. at theit start combines of a cycle and and relates migrate multiple toward the data equator in a as way thethat cycleadds progresses. information to the data cm solar flux

25 The figure presents sunspot data in a condensed and concise way.

26 90N SUNSPOT AREA IN EQUAL AREA LATITUDE STRIPS (% OF STRIP AREA) > 0.0% > 0.1% > 1.0% 30N EQ 30S 90S DATE 0.5 AVERAGE DAILY SUNSPOT AREA (% OF VISIBLE HEMISPHERE) DATE Figure 8: Sunspot area as a function of latitude and time. The average daily sunspot area for each solar rotation since May 1874 is plotted as a function of time inhathaway, the lowerliving panel. Reviews The relative Solar Phys., area7 in (2010) equal area latitude strips is illustrated with a color code in the upper panel. Sunspots form in two bands, one in This eachis hemisphere, called the that butterfly start at diagram, about 25 a from figure thefirst equator designed at the start by Edward of a cycle Maunder and migrate in towardit the shows equator how as the the cycle solar progresses. cycle starts with activity around that then drifts to near (but not quite to) equator. The figure digests a complex dataset to show clear, but complex behavior, far beyond what is encoded in the sunspot number or sunspot observations alone cm solar flux

27 What makes this figure successful?

28 1. Legible 90N SUNSPOT AREA IN EQUAL AREA LATITUDE STRIPS (% OF STRIP AREA) > 0.0% > 0.1% > 1.0% 30N EQ 30S 90S DATE 0.5 AVERAGE DAILY SUNSPOT AREA (% OF VISIBLE HEMISPHERE) DATE Hathaway, Living Reviews Solar Phys., 7 (2010) Figure 8: Sunspot area as a function of latitude and time. The average daily sunspot area for each solar rotation since May 1874 is plotted as a function of time in the lower panel. The relative area in equal area latitude strips is illustrated with a color code in the upper panel. Sunspots form in two bands, one in each hemisphere, that start at about 25 from the equator at the start of a cycle and migrate toward the equator as the cycle progresses. The figure labels, scales, and legend are all readable cm solar flux

29 2. Uses color effectively 90N SUNSPOT AREA IN EQUAL AREA LATITUDE STRIPS (% OF STRIP AREA) > 0.0% > 0.1% > 1.0% 30N EQ 30S 90S DATE 0.5 AVERAGE DAILY SUNSPOT AREA (% OF VISIBLE HEMISPHERE) DATE Figure 8: Sunspot area as a function of latitude and time. The average daily sunspot area for each solar rotation since May 1874 is plotted as a function of time in the lower panel. The relative area in equal area latitude strips is illustrated with a color code in the upper panel. Sunspots form in two bands, one in each hemisphere, that start at about 25 from the equator at the start of a cycle and migrate toward the cm solar flux Hathaway, Living Reviews Solar Phys., 7 (2010) There is little wasted ink; color is used to augment data comprehension. It is not used decoratively, equator as the cycle progresses. Note that wasted color serves only to distract, confuse, or obscures the real meaning you are trying to convey.

30 3. Relates complex data clearly 90N SUNSPOT AREA IN EQUAL AREA LATITUDE STRIPS (% OF STRIP AREA) > 0.0% > 0.1% > 1.0% 30N EQ 30S 90S DATE 0.5 AVERAGE DAILY SUNSPOT AREA (% OF VISIBLE HEMISPHERE) DATE Hathaway, Living Reviews Solar Phys., 7 (2010) Figure 8: Sunspot area as a function of latitude and time. The average daily sunspot area for each solar rotation since May 1874 is plotted as a function of time in the lower panel. The relative area in equal area latitude strips is illustrated with a color code in the upper panel. Sunspots form in two bands, one in each hemisphere, that start at about 25 from the equator at the start of a cycle and migrate toward the equator as the cycle progresses. The figure relates data in space and time in a meaningful way cm solar flux

31 4. Well organized 90N SUNSPOT AREA IN EQUAL AREA LATITUDE STRIPS (% OF STRIP AREA) > 0.0% > 0.1% > 1.0% 30N EQ 30S 90S DATE 0.5 AVERAGE DAILY SUNSPOT AREA (% OF VISIBLE HEMISPHERE) DATE Figure 8: Sunspot area as a function of latitude and time. The average daily sunspot area for each solar rotation since May 1874 is plotted as a function of time in the lower panel. The relative area in equal area latitude strips is illustrated with a color code in the upper panel. Sunspots form in two bands, one in each hemisphere, that start at about 25 from the equator at the start of a cycle and migrate toward the cm solar flux Hathaway, Living Reviews Solar Phys., 7 (2010) The figure s two panels are relatable directly, visually. Each panel augments our understanding of the other equator as the cycle progresses. One takeaway lesson: Striving for simplicity in figures and presentations usually yields very good results.

32 1. Failure to organize the information presented 2. Unreadable or poorly organized slides 3. Unclear, unreadable, or inefficient visuals 4. Failure to understand your audience Finally, let s examine point four by looking at a figure that I think is very good.

33 2004 AL EAST This figure is clear, color is used well, and lots of related information is presented in a distilled format. However, those who do not follow baseball will never recognize that this is the result of a baseball season, nor will they understand why it might be significant. Context would be required for the audience to follow this talk, even though the figure is a good one.

34 1. Failure to organize the information presented 2. Unreadable or poorly organized slides 3. Unclear, unreadable, or inefficient visuals 4. Failure to understand your audience Here s a quick review of the four things that could go wrong. But avoiding these errors is not enough. How can we make our talks great?

35 FROM GOOD TO GREAT

36 1. Start with the problem 2. Analyze the details, master the details 3. Practice, practice, practice 4. Never fear repetition 5. Finish early

37 QUESTIONS?