Content uploaded by Adriano Bisello
Author content
All content in this area was uploaded by Adriano Bisello on Sep 21, 2016
Content may be subject to copyright.
1
XXXVII CONFERENZA ITALIANA DI SCIENZE REGIONALI
THE BENEFIT OF AN INNOVATIVE URBAN INFORMATION INFRASTRUCTURE: THE
CHOICE EXPERIMENT METHOD APPLIED TO THE “TOTEMS” OF EU PROJECT
SINFONIA
Adriano BISELLO1,2, Gianluca GRILLI1,3
ABSTRACT
The diffusion and integration of the information and communication technologies (ICT) in the urban
environment is a pillar of the current smart city development approach. Sensors, monitors, and portable
devices allow acquisition, exchanging and querying of data (or even big data) in real time. However, the
technological and digital innovation in our cities should be considered not the final goal, as the application
per se of all the available technologies. On the contrary, this should be done as the most efficient and cost-
effective way to improve citizens’ quality of life. In designing and construction of an innovative urban
information infrastructure, enabling the users interaction and communication, it plays a prominent role to
understand how and if the project meets users expectations.
By recording the stated preferences of users is possible to determine the value of a service offered for the
first time by the Municipality. Thus, a Choice Experiment valuation method (CE) is here applied to estimate
the economic value of an innovative infrastructure, as a non-market good. According to the CE rules, the
interviewer presents in a face to face surveys specific hypothetical scenarios, describing the offered services
and costs, asking for the best and worst option. The survey involves a sample representative of the whole
potential users, including citizens, commuters, and tourists. The case study analysis is developed within the
European smart city project SINFONIA. In this context, CE allows researchers to estimate the perceived
benefits of forthcoming smart points, called “totems”, in the Italian city of Bolzano, and to provide designers
with insights on additional attributes.
The suggested method allows designers and decision makers to clear understand citizens’ priorities and
expectation toward an innovative urban infrastructure. The purpose of this paper is also to contribute to a
better understanding of the socio-economic aspects connected to the implementation of smart devices in the
urban environment. It shows a high replicability potential in other similar contexts, undertaking the smart
city development approach.
1 EURAC Research, Institute for Renewable Energy, Via G. Di Vittorio 16, 39100 Bolzano, e-mail:
adriano.bisello@eurac.edu (corresponding author).
2 University of Padua, Doctoral School in Management Engineering and Real Estate Economics.
3 University of Trento, Department of Civil, Environmental and Mechanical Engineering.
2
1. Introduction
All over the word, urban settlements are embracing the so-called “smart city” development approach.
Basically, it means ensuring a better quality of life for citizens, by making a full use of information and
communication technologies (ICT), in compliance with the sustainable development principles
(Mosannenzadeh et al., 2016). Currently, there is a sort of environmental-technological nexus, where ever
more people is confident with the internet and mobile devices (see the high market penetration of
smartphones) and at the same time aims to “live greener”. Consumers preferring organic products to large
scale agri-food industry goods, or drinking tap water instead of bottled, are asking to urban public authorities
and service companies to support local agri-food markets, to install drinking water facilities in public spaces
and to promote eco-friendly happenings. Similarly, environmental issues as local air pollution, call for
tangible measures aiming at reducing emissions; for example, by encouraging people to use electric vehicles
in urban commuting instead of traditional cars. In this context wireless connections, interactive
communication tools and innovative devices are essential components to involve citizens in tackling urban
challenges and to make useful the huge amount of information, also called “big data”, continuously
harvested by sensors and stored in servers. New public infrastructures are expected to increase the quality of
life of the residents, meeting their needs and satisfying expectations. In a context of scarcity of public
budgets, investments should be done carefully, without following the “smart city” fascination uncritically. In
some cases, a merge between innovation and traditional solutions, well designed for the specific context,
allows the best result.
Starting from these premises, and in view of a public funding for an innovative urban information
infrastructure in the city of Bolzano, some specific questions rise: (i) do people prefer to have only
information or do they need additional services? (ii) What services mostly increase the perceived benefits?
(iii) Is there any kind of willingness to pay for them? We decided to investigate these points by using the
Choice Experiment method, as face to face questionnaire surveys are usually adopted as a major instrument
for eliciting citizens or tourists’ preferences for non-market goods, as changes in landscape quality,
environmental resources quantity, and availability of new green features in buildings (Chau, Tse and Chung,
2010).
In this paper, a brief description of the study area and data collection method is presented, together with a
general overview ion the methodology and assumptions. Then, results from the testing phase (involving
citizens, urban commuters, and tourists in Bolzano) are presented and insights gained commented, in order to
fix bugs and improve the statistical efficiency of the forthcoming major survey.
2. Methods
2. 1. Study area and data collection
The study has been carried out in the city of Bolzano, which is a city in the northern part of Italy.
Bolzano, with approximately 100,000 inhabitants, is the main city of South Tyrol, a mountainous
province in the Alps. The city itself and the surrounding alpine territory are a well-known holiday
destination, appreciated for a high environmental quality and hospitality. Efficient public services
(especially transportation) and a green tourist-oriented marketing also contribute to promoting the image
of Bolzano as well organized territory inland and abroad. At the country level, the city often ranks on the
top of yearly charts comparing the quality of life in different contexts. In 2015 it was first-ever and under
the “free time” category for presences at cultural events, sports, and expenditures from foreign tourists
(Sole24Ore, 2015).For historical and cultural reasons, Bolzano is characterized by strong liaisons to the
neighboring Austria, it is, in fact, bilingual and both Italian and German are official languages. The
economy of the region, and similarly of the city, is based on services and a good industrial sector, which
allowed Bolzano to be one of the richest Italian city. At the same time, the municipality committed to
making out of Bolzano a smart and technological city, in order to increase inhabitants’ well-being.
3
In mid-2014 the municipality of Bolzano joined a five years smart cities and communities project,
funded by the European Union. The so-called “SINFONIA project”, whose acronym stands for “Smart
INitiative of cities Fully cOmmitted to iNvest In Advanced large-scaled energy solutions”, is going to
develop some different smart measures within the city. The main benefit for single users will be a deep
energy retrofit of selected dwellings in public owned residential buildings, while the most relevant result
for the community will be the implementation of an innovative urban information infrastructure. Such
infrastructure will interact with users through some smart points, called “totems”, to be located in the
most strategic areas of the city.
In this framework, the present research undertakes the issue of what could be the most appreciated
services for citizens, urban commuters, and tourists in Bolzano. Data collection was made by means of a
semi-structured questionnaire, structured as the typical CE survey (Carson et al., 1994; Adamowicz et al.,
1998), personally administrated to the local inhabitants during month July 2016. Two interviewers were
recruited for the data collection. This paper discusses the result of the pre-test, which is conducted with
the aim of receiving feedbacks from the present version of the questionnaire and collect priors for
increasing the design efficiency for the major survey (Johnson et al., 2013). The present version of the
questionnaire contains 4 sections, for a total of 29 questions. The first section contains introductory and
“warm-up” questions, with the main objective of presenting the topic and let respondent be familiar with
it. The second section includes attitudinal questions, to assess the degree of knowledge and feeling of the
respondent with technology in general and informative points in particular. The third section contains the
choice tasks and lastly, the fourth section contained socio-demographic questions to collect information
about the sample. The CE exercise was designed according to the guidelines available in the literature
(Hoyos, 2010; Riera et al., 2012). Relevant attributes and attribute levels included in the choice tasks are
summarized in table 1. Attributes were chosen from a list of candidate services that was discussed with
representatives of the municipality of Bolzano. In particular, possible solutions involve the inclusion of
SOS points, water supply, hotspot for Wi-Fi connections, some level of information about the city (in
particular information about weather conditions, tourist attractions and for residents), electricity supply
(for tablet and smartphones, electric bicycles and electric cars) and information about mobility (parking
area availability, free charging areas and traffic information). Finally, the last attribute is the cost
associated with each alternative, in the form of a monthly ticket to be paid to access those services. The
status quo situation (SQ) is a situation with a basic level of services, still hypothetical because these
infrastructures have to be built. In particular, the SQ foresees a basic level of totems with no SOS, no
water supply, and no Wi-Fi connection. The only available services are: charging points for smartphones,
information about weather conditions and, concerning mobility, information about available parking
spaces.
Table 1 – Attributes and levels for the experiment
Attribute Description Levels
SOS Emergency call service Yes (SOS)
No (SQ)
WATER
Drinking water
Yes (WATER)
No (SQ)
WI-FI
Hotspot for Wi-Fi connection
Yes (WI-FI)
No (SQ)
ELECTRICITY Charging station Devices (tablet + smartphone) (SQ)
Electric bicycle (BICYCLE)
Electric vehicle (AUTO)
INFO Information about the city Weather and environmental conditions (SQ)
Touristic and cultural (TOURISTS)
For residents (RESIDENTS)
4
MOBILITY Information about mobility
Availability of charging points for vehicles
(SQ)
Traffic conditions and public transports
(TRAFFIC)
Availability of parking spaces (PARKING)
COST Monthly cost of the option (€) 0 (SQ),
0.50
1.00
1.50
2.00
2.50
3.00
Source: Author's own elaboration
In order to allocate attributes in the alternatives, a rotation orthogonal design with 36 choice tasks was
created, subsequently arranged in 3 blocks, using the software “R” and in particular the function called
“rotation.design” in the package support “CEs” (Vecchiato and Tempesta, 2015). Each respondent was
required to complete 12 choice tasks, composed by 2 alternatives and the SQ option (which is the current
situation with the infrastructure designed to provide a basic level of information services). An example of a
choice task is given in figure 1. After the pre-test, prior values for the parameters will be estimated, so that
the experimental design could be modified in order to increase its statistical efficiency, following a
procedure described by Huber and Zwerina (Huber and Zwerina, 1996). The efficiency of the new design
will be evaluated based on the D-error criterion, as explained by Rose et al. (2008).
Figure 1 – Example of choice task
Source: Author's own elaboration (translated from the Italian questionnaire)
5
3. 1. Econometric modeling
The CE methodology is embedded in McFadden´s Random Utility theory (RUM), stating that individual
choices are observable with a certain degree of uncertainty (Manski, 1977). According to RUM, the utility
that people derive from a certain purchasing option is given by:
Where is the global utility obtained by individual I, from alternative n in the choice situation t, is
the observable (deterministic) component of the utility, while is a random unobservable disturbance. The
analyst is able to evaluate the deterministic component of the utility that, according to Lancaster´s attribute
theory (Lancaster, 1966), is given by the sum of the utility provided by each attribute of the option:
Where represents the vector of attribute of the option n in the choice situation t and a vector of
parameters, indicating the effect of each attribute in the composition of the observed utility. In order to
evaluate with a statistical model, it is necessary to make assumptions about the distribution of
(Henser, Rose and Greene, 2005). The most common way to do that is to assume a Generalized Extreme
Value distribution type I, allowing the computation of choice probabilities through a Multinomial Logit
Model (MNL). Under the assumption of Independently and Identically Distributed (IID) random terms and
the Independent from Irrelevant Alternatives (IIA), the MNL model restitutes one point estimate for each
parameter. For this reason, is not capable to capture preference heterogeneity across respondents. When
preferences across respondents is supposed to matter, a common statistical specification to relax the
assumption of preference homogeneity is to implement a mixed logit model (MXL). MXL assume a random
distribution of the parameters, thus it is possible to compute individual parameters. In the MXL model,
choice probabilities take the form (Train, 2003):
In which is the logit formula, is the density function of the distribution of the
coefficients. A common practice in empirical evaluations is to assume normally distributed parameters, to a
lesser extent analysists make use of triangular or uniform distributions. The cost attribute is assumed to be
constant; this is not likely to be verified in reality because people may react differently to price changes, but
it is done for computation reasons. The calculation of willingness to pay (WTP) for each single attribute is
given by the negative of the ratio between coefficients, formally:
From this equation, it is possible to see that, assuming a normally distributed cost coefficient, the
computation of WTP would be given by a ratio between two normal distributions. Such an operation
restitutes a Cauchy distribution that doesn´t have central finite moments and it is not desirable (Giergiczny et
al., 2012). Thus, a constant price attribute is preferred. In this paper we make use of the standard MNL and
of the MXL to model information point choices (Greene, Hensher and Rose, 2006), all the computations
were done in R statistical software (R Core Team, 2013).
3. Results and Discussions of the testing phase
As already mentioned, this paper discusses the preliminary result of the study, i.e. the pre-test that was
completed with the main objective of collecting priors for the parameters. Priors are useful to increase the
efficiency of the design, which will be modified for the main survey. During the pre-test it was possible to
6
collect 37 questionnaires, out of which only 29 were compiled enough to be useful for the analyses. Given
that we had 12 choice tasks per respondents and each choice task had to be filled with the best and worst
alternative, it was possible to collect 24 observations per individual, thus leading to 696 total observations. In
Table 2 results of the MNL and MXL models are provided. It is possible to see that, even if the sample is
rather small, several variables are significant at least at 10% of confidence level. This might be considered a
good result, because from the literature it is well-known that orthogonal designs require much bigger sample
sizes for the variables to be significant, compared to efficient designs (Rose et al., 2008; Rose and Bliemer,
2009; Greiner, Bliemer and Ballweg, 2014).
Table 2 – Results of the MLN and MXL models
MNL
MXL
Attribute β St. err. β St. err.
Fixed Parameter
COST - 0.3197 0.1057 ** - 0.535 0.125 ***
Random Parameter
SOS 0.2547 0.1291 * 0.495 0.209 *
WATER 0.3909 0.1316 ** 0.429 0.201 *
WI-FI 0.8167 0.1317 *** 2.362 0.356 ***
BICYCLE 0.2883 0.1944
0.490 0.278 .
AUTO 0.4452 0.1721 ** 0.785 0.223 ***
TOURISTS 0.4638 0.1789 ** 0.438 0.227 .
RESIDENTS 0.5556 0.1835 ** 0.511 0.241 *
TRAFFIC - 0.1052 0.1863
- 0.144 0.228
PARKING 0.2341 0.1749
0.590 0.235 *
SQ - 0.5446 0.3243 . - 0.556 0.408
SCALE 0 0
2.761 0.663 ***
Sd_SOS 1.623 0.252 ***
Sd_WATER 0.877 0.211 ***
Sd_WI-FI 1.850 0.279 ***
Sd_BICYCLE 0.000 0.248
Sd_AUTO 0.119 0.291
Sd_TOURISTS 0.277 0.237
Sd_RESIDENTS 0.828 0.222 ***
Sd_TRAFFIC 0.571 0.274 *
Sd_PARKING 0.209 0.180
Sd_SQ 2.561 0.297 ***
LL
- 533.63
- 428.22
AIC
1090.78
898.43
BIC
1145.33
993.88
N
696
696
Respondents
29
29
.,*,** and *** indicate significance levels of 10%, 5%, 1% and 0.1%, respectively
Source: Author's own elaboration
Concerning attributes, the coefficient for the COST of the alternatives is negative and statistically
significant at 1% of confidence level. This is usually the most important result in a CE application, because it
proves that paying money reduces the perceived utility of any alternative. With regard to the other attributes,
7
it is be shown that people are willing to pay positive figures for many of them. In particular, the coefficient
for SOS is positive and statistically significant, as well as coefficients for WATER and WI-FI. These are
dummy variables, so including such features in the final design of the totems contributes to a higher
appreciation of this foreseen infrastructure. On the other hand, BICYCLE is still positive but not statistically
different from zero, which means that people are indifferent to having electricity points for bicycles,
compared to the basic level that provides power for tablet and smartphone. This might be for several reasons.
For example, not many people have electric bicycles, thus such attribute might be less important for many.
Although in fact these achieved in the last years a quite satisfactory local market penetration and are used by
daily commuting within the city by students and workers. Moreover, electric bicycles are currently offered to
guests by several hotels or rented to tourists. Probably the result relates to the fact that electric bicycles do
not require constant charging, and their autonomy in most of the cases matches the current daily needs of
users. Thus it could be irrelevant for many to have such an additional feature outside their home or
accommodation place. The coefficient for AUTO is positive and statistically significant. This result is in a
way surprising if jointly read with BICYCLE, because people seems not willing to pay for electric bicycles
only, but they are for having the possibility to charge both bikes and cars, the latter being not much common
in Italy. This result could be seen may be as anticipating a further development in the electric vehicles (EV)
sector. Availability and wide territorial coverage of charging stations are fundamental requirements for an
increase in public acceptance and trust toward EVs (Zubaryeva et al., 2012), those are beneficial for
decreasing polluting emissions in transport and reducing dependence on fossil fuels. Concerning INFO, the
level is positive for both coefficients: additional levels of information provided (for tourists and local
inhabitants) contribute to increasing the perceived benefit. Conversely, the attributes for MOBILITY are not
significant, thus such attribute levels seem not to be relevant in the computation of individual preferences.
Finally, the SQ is negative and significant. A negative coefficient for the SQ is an indication that people tend
to prefer alternative solutions, compared to the present situation (in this case a basic solution of an
informative panel). For this reason, additional characteristics of the totems seem to be reasonable to increase
the benefit of the users. The last parameter called SCALE, is the scale parameter connected with the second-
best choice and it is usually included to assess whether the variance of answers vary when the number of
alternatives decreases (Goodman, 2005; Scarpa and Notaro, 2009). In this case, the scale is very close to 0,
meaning that variance is constant between the best and worst choice.
Looking at the MXL model results it is possible to notice that parameters are slightly different in the
magnitude, while almost all of them are coherent in the sign. The SQ becomes non-significant. This result,
jointly read with the significance of the standard deviation of the SQ, indicates that, when taking into account
preference heterogeneity, the current situation is on average nonimportant for respondents, but with many
differences across the sample. In general, the MXL model considerably increases the level of log-likelihood,
meaning that it is probably a better model for the data. In particular, having the log-likelihood closest to zero
(it goes up to -428 from -533, as in MNL), indicates a considerable improvement in a fitting curve to
observations. The significance of many parameters variance is an indication that preference heterogeneity
matters. Concerning point estimates of the parameters, it has to be noticed that the WI-FI parameter sharply
grows from around 0.8 up to about 2.3, while the others remain very similar. In this model, the scale
parameter is also important and statistically significant (with a value of about 2.7).
Table 3 shows the computation of WTP for each attribute. From this table it can be seen that the higher
level of WTP is for having a WI-FI connection (2.55 €/month). This figure increases up to 4.4 €/month in the
MXL model. WTP for information is high as well, assessed to be 1.45 €/month for including tourist
information and 1.74 €/month for including also information to citizens. These figures reduce if calculated
with a MXL model to 0.82 and 0.96 €/month, respectively.
8
Table 3 – WTP for each attribute
WTP (€/month)
Basic level (SQ) MNL MXL
Digital information panel and charging station only for devices (tablet + smartphone) - 1.70 - 1.04
Additional attributes MNL MXL
Emergency call service (SOS) 0.80 0.93
Drinking water 1.22 0.80
Hotspot for Wi-Fi connection 2.55 4.41
Charging station for electric bicycles 0.90 0.92
Charging station for electric bicycles and vehicles 1.39 1.47
Cultural and tourists information 1.45 0.82
Cultural and tourists information, and interaction with municipal information office 1.74 0.96
Information about traffic conditions and public transports - 0.33 - 0.27
Information about traffic conditions, public transports and availability of parking spaces 0.73 1.10
Source: Author's own elaboration
Subsequently, WTP for electricity supply for vehicles and emergency call are positive as well but with a
lower magnitude, while WTP for mobility information is quite low compared to other attributes. The WTP
for SQ is negative, indicating disutility coming from having a basic level of information points, without
innovative services.
This study is just a pre-test, implemented for a better tailoring of the major survey, so results can change
after the administration of more questionnaires and the fine tuning of the experimental design. Still, it is
important to notice that these data, although coming from a small sample, provide interesting insights about
people´s preferences. For example, it is possible to see that probably, the inclusion of Wi-Fi and several
types of information within the final version of the totems might be preferred by citizens, because of their
high coefficients and significance levels. In addition, it is possible to notice that preference heterogeneity
seems to be an important issue in this case. The superiority of the MXL model is not only given by the
performance of the parameters, which are on average more significant, but also by the standard measures of
goodness of fit. In particular, the AIC and BIC statistics are lower in the MXL model, indicating a better
performance of this model. One might think that MXL model is, therefore, more appropriate to investigate
the heterogeneity of totems’ catchment area. Moreover, a further investigation of answers to attitudinal or
socio-demographic questions will help to understand how different categories of users perceive the benefit of
an innovative urban infrastructure with additional services.
4. Conclusions
The present contribution has shown the results of a pretest for a CE application, aimed at estimating
individual´s preferences for smart information points in the city of Bolzano. The main objective of this pre-
test was to obtain estimations of the prior distribution for the coefficients for each attribute included in the
study, in order to improve statistical efficiency for the major survey. Results highlighted that most of the
coefficients for policy attributes are positive, meaning that those features for totems are perceived as
beneficial for the city. Conversely, the SQ situation has a negative WTP, thus indicating that the present
situation (a basic level of the infrastructure) is perceived as not optimal for the city.
In stated preference studies, the main positive aspects are related to the possibility to investigate
preferences of social actors that are affected by public choices, in order to obtain insights about their needs
and, possibly, tailor policies or measures based on their tastes. Drawbacks of this approach are connected to
the high costs of doing the survey, in terms of economic, time and human resources effort. In addition, it is
well-known that stated preferences applications suffer from some bias, which is common to all questionnaire
survey. Next steps for this research will include the modification of the design for the major field survey,
9
which will be administered face-to-face by means of recruited interviewer and translated in German.
Moreover, the pre-test highlighted the need of minor modification of layout and the wordings of some
questions, which will be considered for increasing the quality and clarity of the questionnaire. This step is
very important in surveys, because it is rarely possible to assess a priori the adequacy of a questionnaire as a
tool for collecting information. Pre-tests give the possibility to modify the questionnaire based on
respondents reaction and, consequently, increase the quality of the data in the major survey.
5. Acknowledgements
The research leading to these results has received funding from the European Union’s Seventh
Programme for research, technological development and demonstration under grant agreement No. 609019.
The European Union is not liable for any use that may be made of the information contained in this
document, which is merely representing the authors view. Author contributions: A. Bisello and G. Grilli
designed the research and wrote the paper; A Bisello design the questionnaire and collect the data; G. Grilli
held the literature review, design the choice tasks and analyzed the data. Many thanks to the Municipality of
Bolzano for helping us in collecting data and providing useful comments.
10
6. References
Adamowicz, W., Boxall, P., Williams, M. and Louviere, J. (1998) ‘Stated Preference Approaches for
Measuring Passive Use Values: Choice Experiments and Contingent Valuation’, American Journal of
Agricultural Economics, 80(1), pp. 64–75. doi: 10.2307/3180269.
Carson, R. T., Louviere, J. J., Anderson, D. a., Arabie, P., Bunch, D. S., Hensher, D. a., Johnson, R. M.,
Kuhfeld, W. F., Steinberg, D., Swait, J., Timmermans, H. and Wiley, J. B. (1994) ‘Experimental analysis
of choice’, Marketing Letters, 5(4), pp. 351–367. doi: 10.1007/BF00999210.
Chau, C. K., Tse, M. S. and Chung, K. Y. (2010) ‘A choice experiment to estimate the effect of green
experience on preferences and willingness-to-pay for green building attributes’, Building and
Environment, 45(11), pp. 2553–2561. doi: 10.1016/j.buildenv.2010.05.017.
Giergiczny, M., Valasiuk, S., Czajkowski, M., De Salvo, M. and Signorello, G. (2012) ‘Including cost
income ratio into utility function as a way of dealing with “exploding” implicit prices in mixed logit
models’, Journal of Forest Economics, 18(4), pp. 370–380. doi: 10.1016/j.jfe.2012.07.002.
Goodman, S. (2005) ‘Best-worst scaling: a simple method to determine drinks and wine style preferences’,
International Wine …, 8(972), pp. 1–16.
Greene, W. H., Hensher, D. A. and Rose, J. (2006) ‘Accounting for heterogeneity in the variance of
unobserved effects in mixed logit models’, Transportation Research Part B: Methodological, 40(1), pp.
75–92. doi: 10.1016/j.trb.2005.01.005.
Greiner, R., Bliemer, M. and Ballweg, J. (2014) ‘Design considerations of a choice experiment to estimate
likely participation by north Australian pastoralists in contractual biodiversity conservation.’, Journal of
Choice Modelling, 10, pp. 34–45. doi: 10.1016/j.jocm.2014.01.002.
Henser, D. A., Rose, J. and Greene, W. (2005) Applied Choice Analysis. Cambridge: Cambridge University
Press.
Hoyos, D. (2010) ‘The state of the art of environmental valuation with discrete choice experiments’,
Ecological Economics, 69(8), pp. 1595–1603. doi: 10.1016/j.ecolecon.2010.04.011.
Huber, J. and Zwerina, K. (1996) ‘The importance of Utility Balance in Efficient Choice Design’, Journal of
Marketing Research, 23(8), pp. 307–317.
Johnson, F. R., Lancsar, E., Marshall, D., Kilambi, V., Mühlbacher, A., Regier, D. a., Bresnahan, B. W.,
Kanninen, B. and Bridges, J. F. P. (2013) ‘Constructing experimental designs for discrete-choice
experiments: Report of the ISPOR conjoint analysis experimental design good research practices task
force’, Value in Health, 16(1), pp. 3–13. doi: 10.1016/j.jval.2012.08.2223.
Lancaster, K. (1966) ‘A new approach to consumer theory’, Journal of political Economy, 74(2), pp. 132–
157.
Manski, C. F. (1977) ‘The structure of random utility models’, Theory and decision. Springer, 8(3), pp. 229–
254.
Mosannenzadeh, F., Bisello, A., Pezzutto, S. and Vettorato, D. (2016) ‘Smart energy city definition’,
Manuscript submitted for pubblication.
R Core Team, R. C. (2013) ‘R: A language and environment for statistical computing’. Vienna, Austria.
Riera, P., Signorello, G., Thiene, M., Mahieu, P.-A., Navrud, S., Kaval, P., Rulleau, B., Mavsar, R.,
Madureira, L., Meyerhoff, J., Elsasser, P., Notaro, S., De Salvo, M., Giergiczny, M. and Dragoi, S. (2012)
‘Non-market valuation of forest goods and services: Good practice guidelines’, Journal of Forest
Economics, 18(4), pp. 259–270. doi: 10.1016/j.jfe.2012.07.001.
11
Rose, J. M. and Bliemer, M. C. J. (2009) ‘Constructing Efficient Stated Choice Experimental Designs’,
Transport Reviews. Routledge , 29(5), pp. 587–617. doi: 10.1080/01441640902827623.
Rose, J. M., Bliemer, M. C. J., Hensher, D. A. and Collins, A. T. (2008) ‘Designing efficient stated choice
experiments in the presence of reference alternatives’, Transportation Research Part B: Methodological,
42(4), pp. 395–406. doi: 10.1016/j.trb.2007.09.002.
Scarpa, R. and Notaro, S. (2009) ‘Exploring scale effects of best-worst rank ordered choice data to estimate
visitor’s benefits from alpine transhumance’, in International Choice Modeling Conference. Leeds, pp. 1–
30.
Sole24Ore (2015) No Title. Available at: http://www.ilsole24ore.com/art/notizie/2015-12-20/qualita-vita-
sole-24-ore-bolzano-prima-milano-seconda-sorpresa-reggio-calabria-ultima-
201038.shtml?uuid=ACDibAxB&refresh_ce=1 (Accessed: 27 June 2016).
Train, K. (2003) Discrete choice methods with simulation. Berkeley: Cambridge University Press.
Vecchiato, D. and Tempesta, T. (2015) ‘Public preferences for electricity contracts including renewable
energy: A marketing analysis with choice experiments’, Energy, 88(JUNE), pp. 168–179. doi:
10.1016/j.energy.2015.04.036.
Zubaryeva, A., Thiel, C., Barbone, E. and Mercier, A. (2012) ‘Assessing factors for the identification of
potential lead markets for electrified vehicles in Europe: expert opinion elicitation’, Technological
Forecasting and Social Change, 79(9), pp. 1622–1637. doi: 10.1016/j.techfore.2012.06.004.