delphi tutu2 ptapio 050218
DESCRIPTION
Petri Tapion tietoiskuTRANSCRIPT
P.Tapio TUTU 2, 18.2.2005
What to do with Delphi data?
Feedback and scenarios
Case: Traffic CO2 policy in Finland
Petri Tapio
P.Tapio TUTU 2, 18.2.2005
What will you do with the gathered data?
• Think about it before you gather it!– Concentrate on statements or arguments?– Gathering of authentic statements/arguments
or content analysis?– Quantitative data? Think about scales!
• do not use 5-Likert (or other ordinal scale) scale unless you have to
• use rather interval scale or relative scale
• use rather only one or two scale types
P.Tapio TUTU 2, 18.2.2005
Feedback from the first round may include
• Answers from the previous round + same questions again
• Separate report and better questions• Each participant’s own answers
related to others
P.Tapio TUTU 2, 18.2.2005
Feedback in the case
The preferable futures of road traffic volume- Organisation X and other responses
0
10000
20000
30000
40000
50000
60000
70000
80000
1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025
Time/yr
Road TrafficVolume/
106 vehicle km
Historia
org x VOL
P.Tapio TUTU 2, 18.2.2005
Arguments in the case…
• Arguments for the upper curves– public transport should be supported without
restricting passenger car traffic– the freedom for private car use should not be
restricted
– the growth of CO2 emissions can be stopped with
technical development
P.Tapio TUTU 2, 18.2.2005
…Arguments 2…
• Arguments for the middle curves– the growth potential of passenger car traffic
should be guided towards soft modes, public transport and telecommunications
– urban infill is preferable to reduce the need for traffic
– traffic is a mean, not an end itself, economy and communication should be handled with low need for traffic
– economic growth should be achieved by electronic industry and services, which would reduce the growth rate of freight transport in relation to GDP
P.Tapio TUTU 2, 18.2.2005
…Arguments 3
• Arguments for the lower curves– telecommunications will and should substitute
physical traffic– local more non-material economy is preferable– railroads should be emphasised in both
passenger traffic and freight transport
– CO2 emissions should be reduced 60-80% from
today’s level to stop climate change, which is not possible without also reducing road traffic volume
P.Tapio TUTU 2, 18.2.2005
Cluster analysis
• grouping the responses to clusters• preferable as well as probable futures of
the three key variables– GDP, road traffic volume, CO2 emissions from
road traffic
Qualitative content analysis • are the arguments for the responses within
a cluster similar?
P.Tapio TUTU 2, 18.2.2005
Rescaled Distance Cluster Combine 0 5 10 15 20 25Organisation +---------+---------+--------+--------+--------+
pre=preferable sty pro -+-----+ pro=probable sty pre -+ | akt pro -------+-------+ dodo pro -----+-+ | rhk pro -----+ | ym pro ---+---+ +---------+ rhk pre ---+ | | | ytv* pro -+-+ | | | tl pro -+ +-+ +-------+ | lm pro -+-+ | | | lm pre -+ | | | lal pro ---+-+-+ +---------------+ al pro ---+ | | | akt pre -----+ | | lili pro ---+-+ | | al pre ---+ +---+ | | ene pro -----+ | | | lal pre ---+-+ +---------------+ | tl pre ---+ +-+ | | ym pre -----+ +-+ | ytv* pre -------+ | dodo pre ---+-------+ | ene pre ---+ +-----------------------------+ lili pre -----------+lili pre -----------+
P.Tapio TUTU 2, 18.2.2005
0
5
10
15
20
1970 1996 C 1 C 2 C 3 C 4 C 5 C 6 AlternativeClusters for 2025
GDP index/1926=1,00;
CO2 Emissions
from Road Traffic
/106 tn
0
20
40
60
80 Road TrafficVolume /
109 vehicle km
GDP CO2 Emissions Road Traffic Volume
• Cluster 1– BAU plus
• Cluster 2– Ecological modernisation
• Cluster 3– Modest structural change
• Cluster 4– Strong structural change
• Cluster 5– Deep ecology
• Cluster 6Cluster 6– Steady state economySteady state economy
Probable: STY, AKT, DODO, RHKPreferable: STY
Probable: YM, YTV*, TL, LM, LAL, ALPreferable: LM, RHK, AKT
Probable: LILI, ENEPreferable: AL
Probable: -Preferable: LAL, TL, YM, YTV*
Probable: -Preferable: DODO, ENE
Probable: -Probable: -Preferable: LILIPreferable: LILI
P.Tapio TUTU 2, 18.2.2005
Should you really buy our method?
Qualitative arguments
Quantitative statements
Similar Different
Similar Consistent Inconsistent
Contradictory Inconsistent Consistent
Different Non-systematic Non-systematic
P.Tapio TUTU 2, 18.2.2005
Argumentative interviews
• The role of the interviewer may be
– Neutral
– Sympathetic
– Argumentative
• Argumentative interviews
– The researcher gives counterarguments and further
questions to the interviewee’s statements
– The aim is to produce deeper high quality arguments to
make the interviewees learn from each other
– …and to make the scenarios more precise and consistent
• What kind of problems does this approach generate?
• What are the ways to ameliorate the problems?
P.Tapio TUTU 2, 18.2.2005
Argumentative interviews in round 2: Four ways to avoid bias
• Tell them your argumentative role
• Externalise yourself from the arguments– Rhetorically ”a counterargument has been posed that”
– Think hypothetically - do not try to prove anything
– Respect them and be curious - they may think differently but they do think
• Use systematically first round arguments– Present arguments for upper curves and lower curves
• Concentrate on rational arguments– Dismiss jokes or emotional statements
– ”Dig into their heads”
P.Tapio TUTU 2, 18.2.2005
Examples of personal bias
• Some participants knew the interviewer’s
real opinions – ...or thought they knew
• Two interviewees were not dealt with severely enough– familiarity with the person– the interviewer did not manage to interrupt the
interviewee• Sometimes the interviewer did participate in
non-rational debate– jokes– being too enthusiasted in some arguments
• Concentration is essential!
P.Tapio TUTU 2, 18.2.2005
Conclusions
• Cluster analysis surprises
– Who give similar and different responses?
• Similar quantitative statements sometimes have different qualitative arguments
– Ameliorate with logical analysis, empirical
knowledge and common sence
• Argumentative role of the interviewer induces more in-depth arguments
– Not applicabile to lay people?
• The method does not produce consistency from non-consistency…
– …but it reduces oversimplification and helps in
interpretation
P.Tapio TUTU 2, 18.2.2005
Reference
• Tapio, P. 2003. Disaggregative Policy Delphi: Using cluster analysis as a tool for systematic scenario formation,
• Technological Forecasting and Social Change 70(1): 83-101.