Online Appendix for “Electoral Confidence, Overconfidence, and
Risky Behavior: Evidence From a Study With Elected Politicians”
Lior Sheffer∗
Peter Loewen†
Contents
1 Instrument 2
2 Partisanship Proportionality Descriptives 6
3 MP Sample Gender and Age Descriptives 7
4 Robustness Check: Estimation with Country Interactions 8
5 Robustness Check: Estimation with Treatment Effects 8
6 Robustness Check: Model 2 Estimation with Interaction Term Included
8
7 Robustness Check: Estimation with Objective Electoral Safety as
Subsequent Election Performance 12
8 Raw Plots 12
9 Country-Specific Plots 16
10 Plots with Treatment Assignment Coloring 18
11 Do Subjective Evaluations Predict Subsequent Electoral Performance?
20
12 Questionnaire 22
∗
Department of Political Science, University of Toronto.
lior.sheffer@utoronto.ca
†
Department of Political Science, University of Toronto.
peter.loewen@utoronto.ca
1
This appendix contains supplementary information for “Electoral Confidence,
Overconfidence, and Risky Behavior: Evidence From a Study With Elected Politicians”.
We first provide the instrument used to assess risk-taking and include descriptive
statistics on our MP sample’s partisan and demographic makeup. We then
present results of several robustness checks for the main estimation models reported
in the paper. We then present raw (rather than predicted probability) versions for
two of the three plots used in the paper. We further include by-country panel plots
of the same plots used in the paper, and repeat the original plots with color-coding
for MP gain/loss treatment assignment. Lastly, we include a short discussion and an
empirical analysis of the added predictive power of in-office subjective evaluations
of re-election likelihood over ‘objective’ electoral safety. We also append the full
questionnaire used in this study.
1 Instrument
We use the original Asian Disease vignette first introduced by Tversky and Kahneman
(1981). On top of the original gain/loss framing manipulation, we introduce a
low/high accountability manipulation, resulting in four conditions into which participants
were randomly assigned. In the current study we are only interested in the
overall rates of risk-seeking, across these experimental conditions.
We describe here the texts of each of the four conditions in the Canadian English
version. Translations from this version were made to French, Dutch, and Hebrew,
and are available upon request.
Condition 1: Gain Frame, High Accountability
Canada is planning for the outbreak of an unusual disease, which is expected
to kill 600 people. Two alternative programs to combat the disease
have been proposed and are brought up for vote in the Health Committee,
and you are a member of the committee. The exact scientific estimate
of the consequences of the program is as follows:
If program A is adopted, 200 people will be saved.
If program B is adopted, there is a 33% probability that 600 people will
be saved, and 66% probability that no people will be saved.
Which of the two programs would you support?
2
Program A
Program B
Condition 2: Loss Frame, High Accountability
Canada is planning for the outbreak of an unusual disease, which is
expected to kill 600 people. Two alternative programs to combat the
disease have been proposed and are brought up for vote in the Health
Committee, and you are a member of the committee. The exact scientific
estimate of the consequences of the program is as follows:
If program A is adopted, 400 people will die.
If program B is adopted, there is a 33% probability that nobody will die,
and 66% probability that 600 people will die.
Which of the two programs would you support?
Program A
Program B
Condition 3: Gain Frame, Low Accountability
Imagine that Germany is planning for the outbreak of an unusual Asian
disease, which is expected to kill 600 people. Two alternative programs
to combat the disease have been proposed. The exact scientific estimate
of the consequences of the program is as follows:
If program A is adopted, 200 people will be saved.
If program B is adopted, there is a 33% probability that 600 people will
be saved, and 66% probability that no people will be saved.
Which of the two programs would you support?
3
Program A
Program B
Condition 4: Loss Frame, Low Accountability
Imagine that Germany is planning for the outbreak of an unusual Asian
disease, which is expected to kill 600 people. Two alternative programs
to combat the disease have been proposed. The exact scientific estimate
of the consequences of the program is as follows:
If program A is adopted, 400 people will die.
If program B is adopted, there is a 33% probability that nobody will die,
and 66% probability that 600 people will die.
Which of the two programs would you support?
Program A
Program B
1.1 Instrument-Related Results
For the sake of completion, we report the outcomes of the experimental treatments
in the risk-taking module for the 139 MPs in our sample. Table 1 reports both
raw frequencies and predicted probabilities, obtained using Clarify (Tomz et al.,
2003). For the latter procedure, we estimate a simple logit model of risk-taking, with
the treatment variables (gain/loss frame and accountability) as predictors. Using
Clarify, we derive the predicted probability and confidence intervals for the overall
sample by setting treatment likelihoods for 0.5. To derive the predicted probability
for each treatment condition, we set the orthogonal treatment likelihood to 0.5 and
the relevant treatment to either 0 or 1, depending on the condition evaluated. In
either method the treatment effects are substantively similar, although the predicted
probability for overall risk-taking when treatment assignment is fully balanced is
higher than when observing the raw distribution, likely owing to small imbalances
in treatment assignment in our sample.
4
Proportion Risk-Taking
Raw Predicted Prob. 95% CI
Gains Frame 0.37 0.37 [0.27 - 0.48]
Loss Frame 0.79 0.79 [0.68 - 0.88]
Low Acc. 0.56 0.62 [0.48 - 0.74]
High Acc. 0.53 0.58 [0.44 - 0.71]
Overall 0.54 0.60 [0.51 - 0.68]
Table 1: Risk taking, by treatment assignment.
1.2 Correlation Between Treatment Variables and Confidence
in Re-Election
To validate that our results are not skewed by treatment assignment, we estimate the
effect of gains/loss frame condition and high/low accountability condition on each
of our variables of interest. In all cases, treatment assignment is not significantly
associated with the outcome. Table 2 reports the full estimation results. We also
conduct a robustness check of the models reported in the paper that include the
treatment variables. These are reported below in Table 6.
Dependent Variable:
Confidence in Re-Election Overconfidence Electoral Safety
Loss Frame 0.003 0.009 −0.006
(0.032) (0.055) (0.051)
High Accountability −0.010 0.071 −0.081
(0.032) (0.054) (0.050)
Constant 0.739∗∗∗
0.202∗∗∗
0.537∗∗∗
(0.026) (0.043) (0.040)
Observations 139 139 139
R2
0.001 0.013 0.019
F Statistic (df = 2; 136) 0.047 0.884 1.300
Note: ∗
p<0.1; ∗∗
p<0.05; ∗∗∗
p<0.01
Table 2: Estimation Results: Relationship between treatment assignment and three
variables of interest. Models reported are standard linear regressions.
5
2 Partisanship Proportionality Descriptives
The following table provides a comparison of the by-party proportions in the parliaments
we sampled at the time of the study, and in the sample itself.
Parliament Party Proportion of Proportion in
Sample Parliament
(N interviewed) (N Parliament)
Chamber of
Representatives,
Belgium
CD&V 22.3% 22%
19 22
Groen 7.1% 6%
6 6
LDD 1.2% 1%
1 1
N-VA 28.2% 28%
24 28
Open VLD 12.9% 16%
11 16
sp.a 17.6% 16%
15 16
Vlaams Belang 10.6% 11%
9 11
Total 100% 100%
85 100
House of % %
Commons, N N
Canada
Bloc Quebecois 0% 1.3%
0 4
Conservative 32.5% 53.9%
14 166
Green 0% 0.3%
0 1
Liberal 16.2% 11%
7 34
NDP 51.1% 33.4%
22 103
Total 100% 100%
43 308
The Knesset, % %
Israel N N
Balad 2.8% 2.5%
1 3
Hadash 8.3% 3.3%
3 4
6
Hatnua 8.3% 5%
3 6
Israel Beitenu 2.8% 10.8%
1 13
Jewish Home 8.3% 10%
3 12
Kadima 5.6% 1.6%
2 2
Labor 19.4% 12.5%
7 15
Likud 11.1% 15%
4 18
Meretz 11.1% 5%
4 6
Ra’am-Ta’al 2.8% 3.3%
1 4
Shas 2.8% 9.2%
1 11
Yehadut Hatorah 0% 5.8%
0 7
Yesh Atid 16.7% 15.8%
6 19
Total 100% 100%
36 120
3 MP Sample Gender and Age Descriptives
Belgium Canada Israel
Mean Range Mean Range Mean Range
Year of birth 1967 [1951 - 1987] 1961 [1943 - 1990] 1959 [1945 - 1978]
Gender (% Female) 37.3% 29.7% 18.5%
Overall 75 37 27
Table 4: Elected politicians participating in the study, by age and gender.
7
4 Robustness Check: Estimation with Country
Interactions
We re-estimate Model 1 by adding an interaction term of confidence in re-election
with country fixed effects. For Model 2, we alternately interact overconfidence
(2A) and electoral safety (2B) with country fixed effects. We do this using both
the election-based electoral safety measure and the polling-based electoral safety
measure. In all cases the results remains substantively similar.
5 Robustness Check: Estimation with Treatment
Effects
We re-estimate Models 1 and 2 by adding the treatment variables used in the Asian
Disease experiment - the gain/loss treatment, and the low/high accountability treatment
- to which respondents were randomly assigned. We do this using both the
election-based electoral safety measure and the polling-based electoral safety measure.
In all cases the inclusion of these treatments in the models does not impact
our results.
6 Robustness Check: Model 2 Estimation with
Interaction Term Included
We report a version of Model 2 in which an interaction term between overconfidence
and electoral safety is included, to evaluate the potential existence of compound
relationship between these two constructs and preference for risk-taking. We do
8
Electoral Safety: Election-Based Electoral Safety: Polling-Based
Model 1 Model 2A Model 2B Model 2A Model 2B
Confidence in Re-Election 2.321∗
(1.260)
Overconfidence 2.719∗∗
2.247∗∗
2.605∗∗
2.099∗∗
(1.328) (1.072) (1.311) (1.057)
Electoral Safety −1.035 −0.480 −0.792 −0.352
(0.642) (0.822) (0.640) (0.829)
Canada (FE) 1.703 1.384 0.522 1.449 0.644
(2.008) (2.040) (0.882) (2.034) (0.891)
Israel (FE) 0.468 0.173 0.198 0.204 −0.200
(2.086) (2.127) (0.880) (2.123) (0.862)
Male 0.205 0.197 0.188 0.184 0.170
(0.409) (0.414) (0.405) (0.413) (0.405)
Year of Birth −0.029 −0.029 −0.028 −0.029 −0.028
(0.018) (0.018) (0.019) (0.018) (0.018)
Confidence X Canada (FE) −2.476
(2.743)
Confidence X Israel (FE) −1.106
(2.586)
Overconfidence X Canada (FE) −2.078 −2.170
(2.785) (2.776)
Overonfidence X Israel (FE) −0.993 −0.991
(2.630) (2.619)
Electoral Safety X Canada (FE) −1.191 −1.423
(1.475) (1.482)
Electoral Safety X Israel (FE) −1.787 −0.706
(1.716) (1.682)
Intercept 54.794 56.446 54.486 55.776 54.619
(36.171) (36.374) (36.510) (36.262) (36.380)
Observations 139 139 139 139 139
Log Likelihood −92.062 −90.737 −90.330 −91.289 −91.132
Note: ∗
p<0.1; ∗∗
p<0.05; ∗∗∗
p<0.01
Table 5: Estimation results for a robustness check on country interactions. The
dependent variable is the choice type made in the Asian Disease experiment, coded
0 for certain and 1 for risky. FE - fixed effects; base rate is Belgium. Models are logit
regressions. Standard errors reported in parentheses. * = p ≤ 0.1, ** = p ≤ 0.05,
*** = p ≤ 0.01.
9
Model 1 Model 2 Model 2
(ES: Election-Based) (ES: Polling-Based)
Confidence in Re-Election 1.837∗
(1.083)
Overconfidence 2.324∗∗
2.246∗∗
(1.131) (1.128)
−1.223∗
−1.058
(0.709) (0.707)
Loss Frame 2.005∗∗∗
2.024∗∗∗
2.040∗∗∗
(0.434) (0.442) (0.442)
Accountability (High) −0.233 −0.347 −0.314
(0.395) (0.406) (0.403)
Male 0.449 0.485 0.463
(0.446) (0.452) (0.450)
Year of Birth −0.012 −0.015 −0.014
(0.021) (0.021) (0.021)
Canada (FE) −0.390 −0.438 −0.448
(0.489) (0.494) (0.494)
Israel (FE) −0.102 −0.344 −0.336
(0.539) (0.569) (0.568)
Intercept 21.940 27.051 25.339
(41.022) (40.951) (40.864)
Observations 139 139 139
Log Likelihood −79.653 −78.120 −78.503
Note: ∗
p<0.1; ∗∗
p<0.05; ∗∗∗
p<0.01
Table 6: Estimation results for a robustness check when including Asian Disease
treatment variables. The dependent variable is the choice type made in the experiment,
coded 0 for certain and 1 for risky. Loss Frame is 0 for the gains treatment, 1
for losses. Accountability is 0 for low accountability, 1 for high. FE - fixed effects;
base rate is Belgium. Models are logit regressions. Standard errors reported in
parentheses.
10
this using both the election-based electoral safety measure and the polling-based
electoral safety measure. As the results in Table 7 demonstrate, we do not find an
interaction effect on the likelihood of risk-seeking, and the original results remain
substantively similar, although with reduced significance.
Model 2 Model 2
(ES: Election-Based) Model 2 (ES: Polling-Based)
Overconfidence 2.556 2.392
(1.985) (2.005)
Electoral Safety −0.578 −0.410
(2.489) (2.473)
Overconfidence X Electoral Safety −0.682 −0.590
(3.286) (3.291)
Male 0.144 0.124
(0.403) (0.401)
Year of Birth −0.027 −0.027
(0.018) (0.018)
Canada (FE) −0.096 −0.098
(0.433) (0.432)
Israel (FE) −0.563 −0.530
(0.519) (0.521)
Intercept 52.579 51.931
(36.469) (36.317)
Observations 139 139
Log Likelihood −91.015 −91.598
Note: ∗
p<0.1; ∗∗
p<0.05; ∗∗∗
p<0.01
Table 7: Estimation results: impact of overconfidence and re-election in the subsequent
election on the likelihood of risk-taking in the Asian Disease module. FE
- fixed effects. Reported models are logit regressions. Standard errors reported in
parentheses.
11
7 Robustness Check: Estimation with Objective
Electoral Safety as Subsequent Election Perfor-
mance
We conduct a robustness check of the main Model 2 results in which we substitute
our original measure of election-based electoral safety by whether or not the MP
was re-elected in the subsequent election. That is, electoral performance / safety
is evaluated here not based on recent attainment / polling-based performance, but
based on eventual success. We code this re-election variable as 1 if the MP was reelected
(including election to similar office - for example, in Belgium some MPs chose
to run for the Flemish Parliament or the European Parliament in the subsequent
election). We code it 0 if the MP ran again but was not re-elected. MPs who did
not run again (e.g. retired or passed away) were coded as missing on this measure,
resulting in ten MPs being excluded from the analysis. As can be seen in Table
8, overconfidence is a stronger predictor of risk-taking than re-election, but this
relationship does not meet standard levels of statistical significance (p = 0.18).
Re-election is not meaningfully correlated with risk-taking, and the relationship is
insignificant (p = 0.85).
8 Raw Plots
These plots use the raw risk-seeking data, rather than the models’ predicted probabilities
that are used for the in-paper plots. For these plots we use the election-based
electoral safety measure. Results are visually (and substantively) very similar when
using the polling-based electoral safety measure.
12
Model 2
Overconfidence 1.466
(1.103)
Re-Election 0.088
(0.454)
Male 0.162
(0.413)
Year of Birth −0.026
(0.020)
Canada (FE) 0.011
(0.489)
Israel (FE) −0.338
(0.510)
Intercept 49.771
(38.494)
Observations 129
Log Likelihood −85.925
Note: ∗
p<0.1; ∗∗
p<0.05; ∗∗∗
p<0.01
Table 8: Estimation results: impact of overconfidence and re-election in the subsequent
election on the likelihood of risk-taking in the Asian Disease module. FE
- fixed effects. Reported model is a logit regression. Standard errors reported in
parentheses.
13
0.00
0.25
0.50
0.75
1.00
0.0 0.3 0.6 0.9
Self−Reported Confidence in Re−Election
RiskTaking
Fig. 1: Observed risk-seeking by politicians’ self-reported re-election rates. Lines
are linear predictions
14
0.00
0.25
0.50
0.75
1.00
0.00 0.25 0.50 0.75 1.00
Electoral Safety
RiskTaking
Fig. 2: Observed risk-seeking by politicians’ imputed electoral safety. Lines are
linear predictions.
0.00
0.25
0.50
0.75
1.00
0.00 0.25 0.50 0.75
Overconfidence
RiskTaking
Fig. 3: Observed risk-seeking by politicians’ imputed overconfidence. Lines are
linear predictions.
15
9 Country-Specific Plots
We report by-country versions of the plots reported in the paper. We include plots
for both types of electoral safety measures we employ.
Belgium Canada Israel
0.00 0.25 0.50 0.75 1.000.00 0.25 0.50 0.75 1.000.00 0.25 0.50 0.75 1.00
0.2
0.4
0.6
0.8
1.0
Confidence in Re−Election
RiskTaking(ModelIPredictedLikelihood)
Fig. 4: Predicted risk-seeking by politicians’ self-reported re-election rates.
Belgium Canada Israel
0.00 0.25 0.50 0.75 0.00 0.25 0.50 0.75 0.00 0.25 0.50 0.75
0.2
0.4
0.6
0.8
1.0
Overconfidence
RiskTaking(ModelIIPredictedLikelihood)
Fig. 5: Predicted risk-seeking by politicians’ imputed overconfidence. Electoral
safety measure used is election-based.
16
Belgium Canada Israel
0.00 0.25 0.50 0.75 1.000.00 0.25 0.50 0.75 1.000.00 0.25 0.50 0.75 1.00
0.2
0.4
0.6
0.8
1.0
Electoral Safety
RiskTaking(ModelIIPredictedLikelihood)
Fig. 6: Predicted risk-seeking by politicians’ election-based electoral safety.
Belgium Canada Israel
0.00 0.25 0.50 0.75 1.000.00 0.25 0.50 0.75 1.000.00 0.25 0.50 0.75 1.00
0.2
0.4
0.6
0.8
1.0
Overconfidence
RiskTaking(ModelIIPredictedLikelihood)
Fig. 7: Predicted risk-seeking by politicians’ imputed overconfidence. Electoral
safety measure used is polling-based.
17
Belgium Canada Israel
0.00 0.25 0.50 0.75 1.000.00 0.25 0.50 0.75 1.000.00 0.25 0.50 0.75 1.00
0.2
0.4
0.6
0.8
1.0
Electoral Safety
RiskTaking(ModelIIPredictedLikelihood)
Fig. 8: Predicted risk-seeking by politicians’ polling-based electoral safety.
10 Plots with Treatment Assignment Coloring
We present the main plots reported in the paper, with color-coding for Asian Disease
treatment assignment.
18
0.2
0.4
0.6
0.8
0.00 0.25 0.50 0.75 1.00
Self−Reported Confidence in Re−Election
RiskTaking(ModelIPredictedLikelihood)
Fig. 9: Predicted risk-seeking by politicians’ self-reported re-election rates. Red:
respondents assigned to the gain frame. Turquoise: loss frame.
0.2
0.4
0.6
0.8
−0.5 0.0 0.5
Overconfidence
RiskTaking(ModelIIPredictedLikelihood)
0.2
0.4
0.6
0.8
0.00 0.25 0.50 0.75 1.00
Electoral Safety
Fig. 10: Predicted risk-seeking by politicians’ imputed overconfidence (left panel),
and electoral safety (right panel). Lines are linear predictions. Plots reported are for
the election-based electoral safety measure and results. Red: respondents assigned
to the gain frame. Turquoise: loss frame.
19
0.2
0.4
0.6
0.8
−0.5 0.0 0.5
Overconfidence
RiskTaking(ModelIIPredictedLikelihood)
0.2
0.4
0.6
0.8
0.00 0.25 0.50 0.75 1.00
Electoral Safety
Fig. 11: Predicted risk-seeking by politicians’ imputed overconfidence (left panel),
and electoral safety (right panel). Lines are linear predictions. Plots reported are for
the polling-based electoral safety measure and results. Red: respondents assigned
to the gain frame. Turquoise: loss frame.
11 Do Subjective Evaluations Predict Subsequent
Electoral Performance?
We conduct two separate tests. First, we estimate two logit models of whether
the MPs we interviewed were re-elected in the election subsequent to the time of
interview. Second, we estimate two OLS models where the dependent variable is
our polling-based adjusted measure of electoral safety at the time of the interviews.
In both cases, we examine whether overconfidence and our measure of ‘objective’
electoral safety predict re-election / polling-based electoral attainment. We also
estimate the same models but include country fixed effects.
For re-election, we code the dependent variable as 1 if the MP was re-elected
(including election to similar office - for example, in Belgium some MPs chose to run
20
for the Flemish Parliament or the European Parliament in the subsequent election).
We code it 0 if the MP ran again but was not re-elected. MPs who did not run again
(e.g. retired or passed away) were coded as missing on this measure, resulting in
ten MPs being excluded from this analysis. The polling-based measure is scaled to
[0,1]. Results are reported in Table 9. As the results show, electoral safety, which is
based on performance in the previous election, is a significant positive predictor of
re-election in the subsequent election, while overconfidence is not. We observe the
same with respect to our polling-based measure of electoral safety - past-electionbased
safety is a significant predictor, while overconfidence is not.
Dependent variable:
(Model: logit) (Model: OLS)
Re-Elected (1) Re-Elected (2) Polling-Based (1) Polling-Based (2)
(1) (2) (3) (4)
Overconfidence 1.153 1.618 −0.020 −0.004
(1.122) (1.350) (0.034) (0.035)
Electoral Safety 2.316∗∗
2.874∗∗
0.951∗∗∗
0.957∗∗∗
(1.175) (1.358) (0.037) (0.037)
Canada (FE) −2.511∗∗∗
−0.013
(0.520) (0.015)
Israel (FE) −1.156∗
−0.031∗
(0.590) (0.017)
Intercept −0.635 0.100 0.034 0.036
(0.830) (0.950) (0.026) (0.026)
Observations 129 129 139 139
R2
0.938 0.940
Log Likelihood −78.065 −64.200
Note: ∗
p<0.1; ∗∗
p<0.05; ∗∗∗
p<0.01
Table 9: Estimation results: impact of confidence in re-election and overconfidence
on the likelihood of MPs being re-elected in the election that follows the interview.
FE - fixed effects. Models are logit regressions. Standard errors reported in paren-
theses.
21
12 Questionnaire
(Begins next page.)
22
2013-2014 MP Survey Questionnaire
1) Do you read most of your information on paper (e.g. printed reports, notes) or through digital
means (e.g. email, online)? [SLIDER]
Everything on paper – Everything through digital media
2) To what extent do you feel overwhelmed by the information you receive on a daily basis?
[SLIDER]
Not overwhelmed – Overwhelmed
3) To what extent do you think that other politicians feel overwhelmed? [SLIDER]
Not overwhelmed – Overwhelmed
4) What features of information make you take action (e.g. ask a staff member to follow up on it,
write a press release, and so on)? Please indicate for the following features how much they
matter to you: [SLIDER: DOESN’T MATTER AT ALL – MATTERS VERY MUCH]
______ Deals with an issue that I am specialized in (1)
______ Deals with an issue that my party is specialized in (2)
______ Deals with something that is important for my voters (3)
______ Deals with something that is important for my constituency (4)
______ Can be used to generate negative attention for another party (5)
______ Can be used to generate positive attention for my party (6)
______ Has a lot of potential to gather media attention for me and my party (7)
______ Can help me realize my policy goals (8)
______ Can help me realize reelection or a higher position (9)
5) Information can take different forms, and it can deal with different topics. Please use the bars
to indicate your preference for each feature: [INDIVIDUAL SLIDERS]
Information that is useful in the short term – Information that is useful in the long term
Information that contains only facts – Information that contains only opinions
Information that is very specialized – Information that is very general
Information about an issue that is already on the agenda – Information about an issue that is not
yet on the agenda
6) Please rank the following types of information by order of their importance to your work. Drag
the answers to change their ranking. [RANK BY DRAGGING]
______ Information on political developments and events (1)
______ Information on new policy issues (2)
______ Information on ongoing policy issues you are interested in (3)
______ Information on your constituency (4)
______ Information related to your responsibilities as a minister (5)
7) How frequently do you check the following sources of information?
Multiple
times per
hour (1)
Once an
hour (2)
Multiples
times per
day (3)
Once per
day (4)
Less than
once per
day (5)
Never (6)
Twitter (1)      
News
websites (2)
     
Email (3)      
Social
media
(Facebook,
Linked
In,...) (4)
     
News
agency (5)
     
Print
newspapers
(6)
     
TV news
broadcasts
(7)
     
Radio news
(8)
     
8) Of the initiatives you personally raised in Parliament or in government last year (e.g. motions
for the agenda, bills, written and oral questions), roughly what percentage were inspired by the
following: [0 – 100 PERCENT SLIDERS]
______ The media (1)
______ Interest and action groups (2)
______ Within the party (e.g. leadership, research center) (3)
______ Meeting with individual citizens (4)
______ Personal experience (5)
______ Other (6)
9) Of the initiatives you personally raised in caucus in the last year, roughly what percentage
were inspired by the following: [0 – 100 PERCENT SLIDERS]
______ The media (1)
______ Interest and action groups (2)
______ Within the party (e.g. leadership, research center) (3)
______ Meeting with individual citizens (4)
______ Personal experience (5)
______ Other (6)
10) Please rank the following tasks according to their importance to you as a politician. Drag the
tasks to change their order: [RANK BY DRAGGING]
______ Influencing government policy (1)
______ Providing assistance to individual voters in their dealings with public authorities (2)
______ Looking after the collective social and economic needs of my local area (3)
______ Liaising between members of the parliamentary party and the party leadership and
managing Parliament’s business (4)
11) Some politicians specialize in one or two policy areas, while others prefer to speak and act
upon a wide range of issues from different policy areas. Where would you place yourself on the
following scale? [SLIDER:]
I focus on one issue – I focus on a wide range of issues
12) How important is it to you, personally, to promote the views and interests of the following
groups of people?
Very unimportant
(1)
unimportant (2) Important (3) Very important
(4)
All the people
who voted for
me (1)
   
All the people
who voted for my
party (2)
   
All the people in
my constituency
(3)
   
My party (4)    
A specific group
in society (5)
   
All people in my
region (6)
   
All people in the
country (7)
   
13) If you were to decide to stand at the next general election, how confident do you feel that
you would be renominated by your party?
 I would surely win the nomination contest (1)
 I would probably win the nomination contest (2)
 It could go either way (3)
 I would probably lose the nomination contest (4)
 I would surely lose the nomination contest (5)
14) If you were to decide to stand at the next general elections, how confident do you feel that
you would be re-elected?
 I would surely be elected (7)
 I would probably be elected (2)
 It could go either way (3)
 I would probably not be elected (4)
 I would surely not be elected (5)
15) If it was up to you, where would you like to be five years from now? Tick as many as
applies.
 A Member of Parliament (1)
 A member of a provincial legislature (2)
 Senator (3)
 A cabinet member in the Federal government (4)
 A cabinet member in a provincial government (5)
 Mayor (7)
 Retired from public life (8)
 Working in business (9)
 Working for a public agency (10)
 Working in the non-profit sector (11)
 Other (12)
 (Not in use) (13)
16) Politicians need to balance between different goals. What do you think is the right balance a
politician should have on the following issues: [SLIDER]
Represent the views of citizens and transform them into policy as accurately as possible Represent
their party platform and gain the public's support for it
Resist the demands of other interests and keep to the party line - Take into account other
interests
17) How do you get informed about current affairs? Please tell us the three information sources
that are most important to you: [TEXT BOXES]
1.
2.
3.
18) Please tell us the three most important issues for you as a politician: [TEXT BOXES]
1.
2.
3.
19) How do you get informed about the three top issues you have indicated? Please mark the
three channels most important to these issues.
 Politicians from other parties (1)
 Politicians from my own party (2)
 Mass media (3)
 Industry associations (4)
 My political party (5)
 Social media (6)
 Interest groups (7)
 Individual citizens (8)
 Federal bureaucrats (9)
 Scientific institutions (10)
 Personal contacts (11)
 Parliamentary officers, e.g. Library of Parliament, PBO (12)
For the following questions, we'd like to know about how you make decisions and the about the
kinds of decisions that you make.
[RANDOMLY PRESENT ONE OF THE FOLLOWING FOUR TREATMENT CONDITIONS]
20a) Canada is planning for the outbreak of an unusual disease, which is expected to kill 600
people. Two alternative programs to combat the disease have been proposed and are brought
up for vote in the Health Committee, and you are a member of the committee. The exact
scientific estimate of the consequences of the program is as follows:If program A is adopted,
200 people will be saved. If program B is adopted, there is a 33% probability that 600 people
will be saved, and 66% probability that no people will be saved.
Which of the two programs would you support?
 Program A (1)
 Program B (2)
20b) Canada is planning for the outbreak of an unusual disease, which is expected to kill 600
people. Two alternative programs to combat the disease have been proposed and are brought
up for vote in the Health Committee, and you are a member of the committee. The exact
scientific estimate of the consequences of the program is as follows:If program A is adopted,
400 people will die. If program B is adopted, there is a 33% probability that nobody will die, and
66% probability that 600 people will die.
Which of the two programs would you support?
 Program A (1)
 Program B (2)
20c) Imagine that Germany is planning for the outbreak of an unusual Asian disease, which is
expected to kill 600 people. Two alternative programs to combat the disease have been
proposed. The exact scientific estimate of the consequences of the program is as follows:If
program A is adopted, 200 people will be saved.If program B is adopted, there is a 33%
probability that 600 people will be saved, and 66% probability that no people will be saved.
Which of the two programs would you support?
 Program A (1)
 Program B (2)
20d) Imagine that Germany is planning for the outbreak of an unusual Asian disease, which is
expected to kill 600 people. Two alternative programs to combat the disease have been
proposed. The exact scientific estimate of the consequences of the program is as follows:If
program A is adopted, 400 people will die.If program B is adopted, there is a 33% probability
that nobody will die, and 66% probability that 600 people will die.
Which of the two programs would you support?
 Program A (1)
 Program B (2)
Thank you for your participation. Please hand the device back to the interviewer.
Interviewer transition text
Thank you for completing the survey. If this is ok for you, we will first start with some general
questions on the way you deal with information. Then, we can talk about the questions you
indicated for follow-up on the iPad.
Semi-structured Interview Section - Question List
1) The survey included a few questions on your goals and motivations. Can you explain to me
what the most satisfying aspect of your job as a politician is?
In the survey a lot of information was devoted to how you get informed about society. In the
following part of the interview I would like to ask you some follow-up questions on the
importance of information for your work.
2) We already listed some things that may cause you to pay more or less attention to
information, but can you explain to me what it is about information that makes it of interest to
you?
a. How important is the source of the information?
3) Can you describe the role of the party in your work/what you do/what you pay attention to?
4) How many personal staff members do you have?
5) Can you describe their role and what they do?
References
Tomz, M., J. Wittenberg, and G. King (2003). Clarify: Software for interpreting
and presenting statistical results. Journal of Statistical Software 8(1), 1–30.
Tversky, A. and D. Kahneman (1981). The framing of decisions and the psychology
of choice. Science 211(4481), 453–458.
30