Transcript Title

Giving Talks
Seminars are important
• They provide a way to communicate about your
research
– They are a key element to getting jobs
– As your career progresses, seminars help establish
your reputation
– As your career progresses, requests to give
seminars are a signal of your accomplishment
• They provide a way to receive feedback on your work
– It is “quick”, sometimes useful
– Reach a wide variety of specialties
• So an ability to give good seminars is important
Some Mechanics
• Put together computer-based presentation slides; it is
the norm now for professional presentations.
– But be prepared, a few venues might not have
computers or projectors.
– Some people think what software you use matters,
but most people don’t care.
• Have back-up copies, and make sure the software is on
the computer – one advantage of PDF presentations
– On your computer, if you are using your own
– On a thumb drive
– Email yourself a copy to an accessible account, so you
can always access it
• Look at the audience, not the screen, computer or
projector
Preparing for the seminar
• You should be prepared
• Practice on your own, and before a friendly audience
– An “informal” seminar to other grad students
– A “more formal” seminar for your department
• Don’t just practice the words
• Practice and adjust the organization and tempo
• Be ready to change the presentation from what you
learn by giving it. Things that might change
– Slides
– Words
– Style
Your slides
• Keep them professional
– No “cute” pictures
– Norm now seems to be mostly plain, black text,
white background
– Keep the font and content readable
• No unreadable tables (I’ll give some examples)
• Font appropriate for the size of the room and
the screen. Try to use a clear font, like Calibri
• Have an appropriate number for the time frame
• Time will usually go fast, unless the seminar is going
poorly – then time will go slowly
Your Slides (continued)
• Every bullet point on the slide should have a point
– Think about what you want the audience to learn
from each slide
– Try to keep one idea per slide (obviously I don’t do
this when teaching). Making them readable with
a large font helps with this.
• Don’t use too fancy animations
– Things “flying in” quickly becomes distracting, and
(I think) is inappropriate for professional
presentations
– Fade in or simple animations like wipe tend to
work better
Graphs
• MUST be readable
• Often graphs made for papers don’t work in
presentations; you may need to redo them
• Use text boxes for labels
• Make sure axis labels are readable
• If you will not use the graph, and not talk about it,
don’t include it.
• Rarely should there be more than one graph per slide
as they are too hard to read
• Colors are important to differentiate lines (it is easier
to refer to a color) but make sure they are visible. No
pale yellow (yellow)
6
5
4
White collar UE
3
Blue collar UE
Inflation
2
1
0
2007
2008
2009
2010
6
5
Blue collar UE
4.4
4
3
2.8
White collar UE
2.4
2
Inflation
1.8
1
0
2007
2008
2009
2010
Tables
• Tables made for papers are rarely appropriate for
presentations
• Redo your tables, including only what you want to
talk about
• The font in tables should be at least 20 point (this is
28 point; this is 20 point)
• Use bright colors to bring attention to key elements
of a tables
10
Table 1: Determinants of Pr(y=1) with covariate dependant misclassification
Probit
HAS1
GHAS
Variable
a1=a2=0.05
Intercept
beta1
beta2
True Value
Est.
Hit Rate
Est.
Hit Rate
Est.
Hit Rate
-1
0.2
1.5
-0.791
0.163
1.335
0.01
0.19
0.00
-0.759
0.187
1.454
0.13
0.84
0.81
-1.095
0.213
1.583
0.83
0.90
0.89
beta3
a1=a2=0.1
Intercept
Beta1
Beta2
-0.6
-0.681
0.71
-0.765
0.58
-0.594
0.93
-1
0.2
1.5
-0.705
0.126
1.033
0.00
0.00
0.00
-0.624
0.178
1.285
0.23
0.72
0.55
-0.975
0.209
1.525
0.86
0.92
0.89
Beta3
a1=a2=0.2
Intercept
beta1
beta2
-0.6
-0.447
0.29
-0.589
0.94
-0.655
0.95
-1
0.2
1.5
-0.561
0.086
0.763
0.00
0.00
0.00
-1.394
0.262
1.777
0.76
0.86
0.89
-1.152
0.225
1.618
0.85
0.89
0.88
beta3
-0.6
-0.149
0.00
-0.518
0.81
-0.494
0.88
Intercept
-1
-0.691
0.00
-0.731
0.40
-1.030
0.92
beta1
0.2
0.146
0.01
0.173
0.62
0.209
0.87
beta2
1.5
1.229
0.00
1.357
0.57
1.538
0.90
beta3
-0.6
-0.520
0.81
-0.594
0.96
-0.573
0.92
Intercept
-1
-0.601
0.00
-1.320
0.81
-1.061
0.81
beta1
0.2
0.122
0.00
0.215
0.84
0.213
0.87
beta2
1.5
1.036
0.00
1.613
0.88
1.534
0.85
beta3
-0.6
-0.070
0.00
-0.177
0.19
-0.515
0.87
Intercept
-1
-0.624
0.00
-1.312
0.78
-1.146
0.78
beta1
0.2
0.105
0.00
0.220
0.88
0.222
0.90
beta2
1.5
0.857
0.00
1.508
0.90
1.623
0.86
beta3
-0.6
-0.006
0.00
-0.036
0.07
-0.531
0.90
=0.1 , =0.05
=0.2 , =0.05
=0.2 , =0.1
10
Table 1: Determinants of Pr(y=1) with covariate dependant misclassification
Probit
HAS1
GHAS
True
Hit
Hit
Hit
Variable
Value
Est.
Rate
Est.
Rate
Est.
Rate
a1=a2 =0.05
Intercept
-1
-0.791 0.01
-0.759 0.13
-1.095 0.83
Beta1
0.2
0.163 0.19
0.187 0.84
0.213 0.90
Beta2
1.5
1.335 0.00
1.454 0.81
1.583 0.89
Beta3
-0.6
-0.681 0.71
-0.765 0.58
-0.594 0.93
a1=a2 =0.1
Intercept
-1
-0.705 0.00
-0.624 0.23
-0.975 0.86
Beta1
0.2
0.126 0.00
0.178 0.72
0.209 0.92
Beta2
1.5
1.033 0.00
1.285 0.55
1.525 0.89
Beta3
-0.6
-0.447 0.29
-0.589 0.94
-0.655 0.95
a1=a2 =0.2
Intercept
-1
-0.561 0.00
-1.394 0.76
-1.152 0.85
beta1
0.2
0.086 0.00
0.262 0.86
0.225 0.89
beta2
1.5
0.763 0.00
1.777 0.89
1.618 0.88
beta3
-0.6
-0.149 0.00
-0.518 0.81
-0.494 0.88
During the talk
• Pay attention to the time
– Don’t spend too much time up front, so you have
to rush the conclusions and message
– Leave plenty of time for questions and suggestions
from the audience
– For a 60 minute presentation, plan on 45 minutes
• Let the audience know if you want questions during
the talk
– Clarifying questions are always in order
– But content questions, suggestions and challenges
can wait to the end
During the talk (continued)
• Be flexible
– Be prepared to skip some slides if things are
progressing too slowly
– Have a couple of extra to lengthen the talk if
needed
– End a few minutes early rather than going too long
• 60 minute presentations cannot be condensed to a 20
minute talk at a meeting. For short presentations, the
point is to convince the audience to read your paper.
• For 90 minute presentations (norm for job market
talks) add more detail, and present key elements of
derivation, empirics or whatever to go about minutes.
During the talk (continued)
• Control the presentation, it is your talk
• If questions take it off track, move it back on
• Be prepared to ask people to hold questions to later
• Be prepared to suggest that you would like to discuss
the point with the questioner after the talk
Content
• Long introductions are almost always a bad idea
– Save the time for substance
– Literature reviews are not needed, except perhaps
a key citation or two for context
– The focus of the introduction is your research
question, why the question and answer are
important, and what your answer will be
• Don’t be mysterious – let the audience know what
you research, what you find, and why, early on in the
talk
• Then, give the substance of how you do it.
Dos
• Identify your main point (finding, importance) and
state it up front and succinctly.
• Repeat your main point, and summarize your
findings, at the end.
• Speak clearly and loudly.
• Know your audience.
• Don’t show your back, don’t talk to the screen.
• Use a laser pointer if you want to refer to a specific
place on a slide.
• Stick to your time limit.
• Practice several times before the presentation.
Dos (continued)
• Put an appropriate amount of information on a slide.
• Use bullet points, not full sentences.
– Don’t crowd slides.
– Don’t read slides (see how I violate these rules all
the time).
• Make sure you know how to use the equipment.
• Get to the #1 important contribution as quickly as
possible.
• Give people time to digest your slides.
• Listen carefully to questions, but think before
answering.
• Keep presentations and answers simple.
Don’ts
• Make the motivation too long
• Have a long literature review
• Give extensive previews of the results
• Give useless context
– If your paper is primarily empirical, skip the theory
– Do not discuss preliminary or interim results, get
to the final results
• Give an answer to a question if you don’t know it.
• Speak softly and tentatively.
• Go over your time limit.