Professional Documents
Culture Documents
Writing Papers and Going To Conferences: Gita Subrahmanyam
Writing Papers and Going To Conferences: Gita Subrahmanyam
AND GOING TO
CONFERENCES
Gita Subrahmanyam
Authoring a PhD and Developing
as a Researcher
OUTLINE OF WORKSHOP
Why go to conferences and seminars?
Hierarchy of conferences
Writing, structuring and proposing
papers
Delivering papers
Transmitting
ideas is a key
step in getting
feedback and
upgrading
your
knowledge.
Credit:www.imageafter.com
HIERARCHY OF CONFERENCES
Seminars in home institution - known
audience
Postgraduate conferences
External seminars, specialist groups in
your profession (wider audience)
UK national conference choice of panels
European-level international conferences
workshops, panels, specialist groups
US/global conferences huge attendance
but often tiny audiences at individual
panels real action in bars, book fairs,
receptions
FOCUS ON THE
NEED TO KNOW CRITERION
Normal (written) form is:
What do readers really need to
know?
Conference (presentation) form is:
What does the audience really need
to see on screen?
What do listeners really need to have
explained to them?
NEED TO KNOW
IMPLICATIONS 1
However literary your normal style, plan
the talk as a sequence of exhibits
Put all that you want to say/show on
screen, in a user-friendly manner
Practice timings for your talk
Aim for a fast, well-paced start do not
warm up the audience to your subject
Sell the paper dont be diffident
NEED TO KNOW
IMPLICATIONS 2
Organise your talk into 3 minute chunks,
planning for one display per chunk
Use PowerPoint (not Word) to compose
your displays and have OHP backups!
Text should be free-standing and readily
understandable without you speaking
(audience will deconstruct it like that)
Try to avoid a build-up of slides or too
many flying bullets delays exposition
and too controlling
NEED TO KNOW
IMPLICATIONS 3
Pick a font that is visible to someone in
the back row - like this one
Put equations and quantitative tables
into separate image screens, magnified
so that the smallest subscript is visible
Preferably use summary data tables,
rather than detailed ones
Pick the best feasible fonts for display
HAVE A GO
Write a proposal/abstract for the
conference of your choice
Follow the Call for Papers guidelines in
the example you brought in, EXCEPT
stick to a maximum of 200 words
If you havent brought a Call for
Papers, then try using one of the spare
copies at the front of the room
A GOOD PROPOSAL/ABSTRACT
Sentence 1 a hook, indication of
motivation (for you and reader)
Sentences 2 3 formulation of research
problem/question
Sentences 3 4 outline of core finding
(maybe a sideways glance at method)
Sentences 5 6 - implications
http://www.mcdonald.cam.ac.uk/McD/Seminar.jpg
Credit: http://www.finearts.uvic.ca/visualarts/facilities/images/seminar/seminar-1.jpg
CREDFIT: http://www.eastwood.asn.au/images/hall15_b.jpg
Credit: http://www.brc.ubc.ca/vtour/images/cell/L3_seminar1.jpg
http://www.ccc.ox.ac.uk/conference/images/semnarrm2.jpg
Credit: http://www.ruwpa.st-and.ac.uk/workshop2002/seminar%2520room3.jpg
http://www.sunyit.edu/news/academic/pictures/main.jpg
Credit: http://www.reidkerr.ac.uk/conference/images/ante2B.jpg
PRESENTING DATA
poorly
START
BADLY
Ive printed
my cover
page in tiny
font and
slapped it
on the OHP
slide
Abstract: Pioneering work by Laver and Benoit (LB) argues that a drive by individual
legislators to maximize their per capita Shapley-Shubik power scores could explain the
evolution of party systems in legislatures. But LBs analysis exhibits several problems.
Theoretically their utility premises are incompletely specified and would lead to
systematically irrational and short-termist behaviour by members of vote blocs.
Methodologically LB focus on a complex ratio variable, whose patterning essentially depends
on another largely unanalysed variable, the power index scores of whole vote blocs. LB have
no framework for economically analysing variations in power index scores across very
numerous and diverse voting situations. Empirically LBs account radically mis-specifies the
factors conditioning blocs incentives or actors incentives. We show that: (i) they offer an
exaggerated picture of the scope for defection; and (ii) their emphasis on the importance of
dominant bloc status for the largest bloc is incorrect - dominance is often empirically trivial
in shaping bloc scores when there are more than five blocs. Instead, the factors that do
influence blocs scores are predictable, (if complex), patterns, which repeat in recognizable
ways across weighted voting situations, for any given threshold level. We demonstrate a
method for mapping these scores comprehensively and economically, and for analysing
influences on the scores precisely.
Paper to the panel on New Perspectives on Rights, Freedoms, and Powers at the European
Consortium of Political Research, Annual Workshops 2003, University of Edinburgh, 28
March 2 April 2003.
analysis, and his lonely faith in the value of other effective number indices, for which there
has been little or no take-up in the existing literature. By contrast we believe that the wider
effective number family has little to offer, and that continuing to use unmodified N 2 in
MAINTAIN
CONSISTENCY:
Some of
you may
not be able
to see the
subscripts
here too
well
particular in quantitative applications cannot be defended because of the defects set out
here.
In our view averaging N2 scores with the 1/V1 score creates a simple but useful
variant of the effective number index, Nb:
(3)
The data demands of equation (3) are no greater than for the N 2 index, and Nb and N2 are
highly correlated with each other. Yet this straightforward modification has useful effects.
Figure 6 shows the minimum and maximum fragmentation lines for Nb with between 2 and
8 parties, and also includes the 1/V1 line and the overall maximum fragmentation line for
Nb (with a 1 per cent floor for party sizes, as before). The averaging of N 2 and 1/V1
creates much less curved minimum fragmentation lines. And although there are still
transitions in their slopes around the anchor points, they are much less sharp than with N 2.
The maximum fragmentation lines for different relevant numbers of parties are also
considerably straightened out under Nb, without strongly visible curves close to their
terminal anchor points. The overall maximum fragmentation line for Nb is appreciably
lower than the 1/V12 line under N2. In fact the Nb maximum fragmentation line runs quite
close to but slightly above the N3 maximum line shown in Figure 1. For instance, with V1
at 60 per cent, the maximum Nb score is more than half a party less than with N2 ; and at
50 per cent support the Nb upper limit is 3 parties, instead of 4 for N2. Thus the Nb index
delivers many of the same benefits in terms of more realistically denominated scores as N 3,
but it avoids N3s severe kinks around anchor points (which is evident in Figure 4).
Table 2 shows how the N2, Nb and Molinar measures behave empirically across the
TABLES
complex,
difficult to
read, weak
heading/title,
unnecessary
abbreviations,
space wasted
between data
points
CHARTS
3D design,
small and thin,
weak heading,
no logic to
arrangement
of bars, labels
in a legend,
key details in
micro font
70000
60000
50000
40000
30 000
20000
1000 0
0
1
10
11
12
13
14
15
T rtm nt ra te s /po p
Key: The health boards are as follows: 1 Ayre & Clyde; 2 Ayrshire & Arran; 3 Border; 4 Dumfries &
Galloway; 5 Fife; 6 Forth Valley; 7 Grampian; 8 Greater Glasgow; 9 Highland; 10 Lanarkshire; 11 Lothian;
12 Orkney; 13 Shetland; 14 Tayside; 15 Western Isles.
Table 5: The extreme bloc sizes and per capita SS values in the triads, quinns and sevens
areas
i. Triads area
VERY
LARGE
TABLES
multiple
smudges of
micro font are
not ideal for
presenting full
regression
results to a
crowded room
Description
Bottom left cell
Bottom right cell
Blocs
All
4
8
14
20
24
26
4
8
14
20
24
26
V1
26
48
44
38
32
28
26
48
44
38
32
28
26
Bloc sizes
V2
V3
26
25
26
25
26
48
44
38
32
28
26
25
3
7
13
19
23
25
V1
1.28
0.69
0.76
0.88
1.0
1.2
1.28
0.69
0.76
0.88
1.0
1.2
1.28
Blocs
All
6
8
14
20
6
8
14
20
Top cell
V1
17
31
29
23
17
24
23
20
17
V2-V4
17
V5
17
17
17
17
24
23
20
17
17
3
5
11
17
V1
1.18
0.65
0.69
0.87
1.18
0.69
0.76
1.0
1.18
V2-V4
1.18
V5
1.18
1.18
1.18
1.18
0.69
0.76
1.0
1.18
1.18
6.67
4.0
1.82
1.18
Dif
0
0.53
0.49
0.45
0
5.98
3.24
0.18
0
Blocs
All
6
8
14
6
8
14
V1
13
21
15
13
16
14
13
Bloc sizes
V2-V4 V5-V6
13
13
13
13
13
13
13
13
16
13
14
13
13
13
V7
13
13
13
13
9
11
13
V1
1.10
0.68
0.95
1.10
0.89
1.02
1.10
Dif.
0
0.42
0.15
0
0.70
0.28
0
PRESENTING DATA
properly
Credit: http://www.pi1.physik.uni-stuttgart.de/Soellerhaus2002/Bilder/Soellerhaus2002-12.jpg