Title of manuscript:

“Using Decision Analysis to Explore Cable Television Delivery: A Case Study in University Technology Adoption”

Authors of manuscript:

Keith A. Willoughby Department of Finance and Management Science

Edwards School of Business University of Saskatchewan

Saskatoon, SK Canada S7N 5A7 Phone: (306) 966-2128 Fax: (306) 966-2515

E-mail: willoughby@edwards.usask.ca

Christopher J. Zappe Provost’s Office

Gettysburg College Gettysburg, PA 17325-1400

Structured Abstract:

Purpose:

To demonstrate the efficacy of decision analysis in determining the most efficient strategy for

installing cable television in the residence halls of Bucknell University.

Design/methodology/approach:

The decision analysis model compared five distinct approaches for achieving and maintaining a

successful delivery of cable television service to students enrolled in this private, residential

institution. For each alternative, the model incorporated installation costs, likelihood of

installation failure, installation failure costs, likelihood of obsolescence, and obsolescence-

related costs. In addition to considering the tradeoffs between cost, timing, and riskiness of the

various alternatives, a thorough set of sensitivity analyses was performed to gain insight into the

parameters that most strongly influence this decision-making process.

Findings:

The quantitative model advocated the adoption of the university’s data network as the mode for

cable delivery. Sensitivity analysis further supported this notion.

Practical Implications:

The analysis of this problem incorporated the knowledge and judgments of senior administrators

and staff members, thus demonstrating the critical contributions offered by subject matter experts

in advising, informing and launching successful decision analysis projects. Incorporating

stakeholder viewpoints enhances model understanding and eventually model implementation.

Decision analysis represents a powerful approach in communicating uncertainties and advising

on the benefits of particular alternatives.

Originality/Value:

To the best of the researchers’ knowledge, this paper represents an initial attempt to investigate

cable delivery options within a decision analysis framework.

Keywords: Decision analysis; Telecommunications; University operations; Case study

1.0 Introduction

Decision analysis is a useful approach to quantitatively assess various alternatives in the

face of future uncertainties. These uncertainties, often called “states of nature”, are characterized

by specific probability distributions. By incorporating the associated payoffs of different

alternative-state of nature combinations, a decision-maker can determine a preferred alternative.

To illustrate how the preferred alternative is affected by changes in underlying parameters, a

variety of sensitivity analyses may be performed.

This paper is an account of an actual operations research project. To wit, the case study

demonstrates the efficacy of decision analysis in determining the most efficient (i.e. expected

cost-minimizing) strategy for installing cable television in the residence halls of Bucknell

University. Founded in 1846, this private, residential institution educates about 3,500 (nearly all

undergraduate) students. The model investigates five distinct alternatives and a variety of costs

and future outcomes. Critical tradeoffs involve the cost and riskiness of separate strategies. To

augment the likelihood of model acceptance, a concerted effort is made to evoke the viewpoints

and judgments of senior administrators and staff members in the development of the decision

analysis model.

Researchers have applied decision analysis in a rich variety of real-world settings. Cable

television delivery, the focus of this paper, may be considered within the class of technology-

adoption applications. Although this model - to the best of the researchers’ knowledge -

represents an initial attempt to investigate cable delivery options, decision analysis approaches

within a technology-adoption environment have enjoyed a productive history. For example,

Brooks and Kirkwood (1988) deployed a decision analysis model with a multi-attribute utility

function to determine preferred strategies for microcomputer networking. Basing their analysis

on important decisions encountered by a public utility, the authors conjoined various evaluation

measures to determine an index for each strategy’s value and used probabilities to represent

critical uncertainties regarding network performance. Borski et al. (2008) developed a multi-

criteria model to assess how specific technology could manage environmental risk and improve

water management for a treatment plant in Switzerland. Jackson et al. (2007) implemented

decision analysis for waste site remediation. In their model, the decision-maker needed to

optimally determine the processes for a site, as well as the appropriate technology for each

process. To determine a particular timing of software upgrades that would amplify productivity

gains, Ngwenyama, Guergachi and McLaren (2007) combined a decision analysis approach with

a learning curve model. Browne, O’Regan and Moles (2010) explored various energy

technologies within a multi-criteria framework, and applied their approach to an Irish city-

region.

Other technology-based applications include a decision analysis model to investigate the

purchase or conversion of specific postal automation equipment for reading addresses on

packages (Ulvila 1987), a method to scrutinize how uncertainties concerning weather and soil

fertility impact specific technologies within agricultural decision-making (Isik and Khanna 2003)

and an approach to identify software development practices (Machado, Pinheiro and Tamanini

2015). Cortes et al. (2001) built a decision support system to guide the planning of fiber optic

telecommunication networks for a particular region in Spain. Their model explored key

decisions such as hub location and the particular urban nodes deemed appropriate for receiving

telecommunication contents.

Besides technology adoption, researchers have explored a number of other topic areas

through the successful application of decision analysis approaches. These include health care

(Kocher et al., 2002; Keren et al., 2002; Taris 1998; Fleming et al., 1993; Stonebraker 2002),

financial option analysis (Miller and Park 2002) and environmental management (Khalili and

Duecker 2013; Huang, Keisler and Linkov 2011; Mosadeghi et al., 2013). Other situations

involve capital budgeting (Ghosh et al., 2009), information systems (Johnson et al., 2002) and

supply-chain management and analysis (Kirkwood, Slaven and Maltz 2005; Chai, Liu and Ngai

2013). Additional topics comprise airport passenger screening procedures (Martonosi and

Barnett 2006) and the formation of civilian response teams (Linkov et al., 2012).

The paper proceeds as follows. The next section presents a background for this

managerial modeling project, highlighting proposed benefits and costs of cable installation, as

well as critical events that occurred as a university-wide committee sought to identify a suitable

cable television technology. This followed by a discussion of model development. The model

results are then provided, including the insights obtained from sensitivity analyses. Some

concluding remarks are offered in the final section.

2.0 Background

At the outset of the project, cable television service was not provided (with but few

exceptions) in the residence halls and dormitories of the Bucknell University campus.

Admittedly, the process of securing cable television access had been somewhat grueling and

arduous. For several years, various student leaders had espoused the benefits of cable access and

advocated its installation to college administrators.

The current modeling project began during a decision sciences course taught by one of

the co-authors. Concurrently, the notion of providing cable television access in individual

student dormitory rooms was being informally discussed in various committees that included

both students and administrators. A particular student in the co-author’s course, a member of

Bucknell Student Government (BSG), approached the instructor with the suggestion of using

decision analysis to explore the cable television adoption decision. Apparently, despite the fact

that considerable discussion occurred regarding distinct alternatives in these collaborative

meetings, no type of quantitative methodology had been applied to this important real-world

problem.

Eventually, these committee meetings spurred the creation of the “Task Force for Cable

Television on Campus”. This university-wide committee was charged with formally and

thoroughly investigating the feasibility of providing cable access in campus dormitories, thus

demonstrating the importance of this issue for this particular educational institution This task

force included representatives from the student body, faculty (one of the co-authors was the

faculty representative), staff and administration (from such areas as Physical Plant, Information

Services and Resources, Finance Office, Housing and Residential Life, and Student Services).

Through its deliberations, the Task Force determined some significant reasons for

supporting cable television delivery. First, various educational benefits could materialize by

providing cable access to individual students. Potentially, the University could use cable

television to deliver class films (e.g., the first-year Economics course shows weekly films as part

of its pedagogy). Thus, students could watch the material at a specific time of their choosing,

rather than requiring the instructor or his/her teaching assistant to display a film to a large group

of students. Occasionally, professors want students to report on various news, documentary or

current events programs available on such networks as CNN, MSNBC, and the History Channel.

Individual cable television access would provide students with the opportunity to participate in

such activities. Admittedly, students with cable access may not watch news or current events

programs exclusively. Despite the fact that television may provide a distraction, Task Force

members felt that college-aged students should be given the responsibility to balance academics

with their social life.

Secondly, cable television could prompt the development of a Bucknell television station

to broadcast collegiate cultural and athletic events. A third benefit for dormitory room cable

access is that it would eliminate student inconvenience. Currently in some residence halls, there

exists a single big-screen television serving the viewing desires of a few hundred students.

Obviously, not all needs can be met with just one television.

To provide an enhanced, external context for these critical sets of decisions, the Task

Force surveyed seven local area colleges and universities (Gettysburg, Lycoming, Franklin &

Marshall, Colgate, Lehigh, Susquehanna and Bloomsburg) with respect to their delivery of cable

television. Each of these seven schools offers cable television on campus, with all but one of

them making it a mandatory service. For what it is worth, none of these schools reported any

negative effects (such as lower grade-point average) associated with cable delivery.

3.0 Model Development

Five specific delivery alternatives were considered in the decision analysis model. Staff

members from Information Services and Resources, as well as Physical Plant, were quite helpful

in eliciting and explaining these separate options. All features of our decision analysis model

(including costs of each alternative and probabilities for various uncertainties) were determined

during a series of Task Force meetings. These formal consultations occurred on a regular basis

(roughly each month) over an eight-month period. Informal discussions were arranged outside

of the regular Task Force meetings to clarify particular aspects of the analytical model.

The first alternative entailed building a brand-new “cable plant”. This would involve

installing a main coax cable “highway” from a central location to each campus dormitory. From

the equipment in each dormitory, separate coax runs would be installed, eventually connecting to

individual televisions in all dormitory rooms. Although this represented the traditional way to

provide cable television service, Task Force members determined that this was the costliest of all

the alternatives.

Previously, it was indicated that Bucknell University provides cable television service in

a limited number of student rooms. The second option took this into account by using some of

the existing fiberoptics already in place for cable delivery on campus, rather than building a

brand-new cable plant as required in the first alternative. For those buildings not currently

wired for cable service, coax highways would be installed, with coax runs established for

individual rooms. This alternative is less costly than the first one since a completely new

infrastructure is not required.

The third alternative resembled option two, except that instead of requiring separate coax

runs to individual rooms, one would use Lucent Giga Speed Unshielded Twisted Pair (UTP)

cable for student room access. Despite the fact that it is less costly than establishing separate

coax runs, Task Force members learned that it could only support a maximum of 77 channels of

cable television. It was not known if this would fully accommodate the current (and future)

world of cable channel possibilities.

A fourth alternative provided cable television service directly through the existing

Bucknell data network. This would eliminate the need for a coax highway between buildings, as

well as coax runs to separate student televisions. Students would be required to purchase a set-

top box (a device that would connect into the existing data port in each room in order to receive

cable signals). Although this alternative could deliver up to 270 channels and picture quality

would supposedly be excellent, Task Force members realized that this represented more of a

“cutting-edge” technology. Regrettably, newer technologies may be prone to unforeseen

technical issues at the outset of their launch.

The final option entailed delivering cable television content directly to a student’s

computer (e.g. Web TV), thus eliminating the need for a set-top box (in addition to the coax

highways required in the first three options). Although this alternative had the least installation

costs, decision-makers felt that it was so cutting edge that its eventual utilization may not be

fully practical until several years into the future. Further, the user’s interface was thought to be

less than desirable. For example, the size of the television viewing area was limited to the PC

monitor size, and one could not use a remote control to change channels.

Table 1 provides the installation costs for the five delivery alternatives. Due to reasons of

confidentiality, all costs have been scaled. One can interpret this table by noting that the

installation costs of Web TV (alternative #5) were 10% of the costs of UTP cable (option #3) and

that replacing the entire cable plant (alternative #1) was 1.84 times the costs of the UTP cable.

Stakeholders from the Finance Office, Information Services and Resources, and Physical Plant

were confident that these costs accurately characterized the given alternatives.

===== insert Table 1 about here =====

During the Task Force meetings, the group uncovered two relevant uncertainties with

respect to selecting the appropriate cable television delivery option. The first uncertainty

concerned the probability of success of the specific choice; in other words, will the technology

work as anticipated? Given that the particular option was successful, a second uncertainty

involved the likelihood that it would become obsolete. Decision-makers did not attempt to

tackle the (rather complex) problem of determining when a specific delivery mode would

become obsolete; rather, they were simply concerned with establishing the likelihood of the

technology eventually becoming outdated.

The probabilities of these various uncertainties were determined in direct consultation

with various college stakeholders. Initially, they had qualitatively assessed different risks using

terms like “very low”, “medium” and “high”. Obviously, one needed to quantify these different

uncertainties by providing a probability estimate. The researchers discussed ranges of

probability values with the Task Force members before zeroing in on a specific estimate. A

more objective approach to quantify risk could have involved the use of event trees. Although

event trees may have provided a scientific method to elicit risk values, it was felt that the

iterative approach of generating specific probability estimates by initiating thorough discussions

with Task Force members yielded accurate probability values. In the end, it was crucial that the

model development not become a “black box”, thus potentially limiting its eventual use as an aid

in actual decision making.

Table 2 offers the respective probabilities used in our decision analysis model. More

attractive alternatives have higher estimates for successful installation and lower values for

obsolescence risk.

===== insert Table 2 about here =====

Tradeoffs were readily apparent with these alternatives. “Older” approaches such as

replacing the entire cable plant (alternative #1) are almost guaranteed to be successful; the

problem arises due to their likelihood of becoming obsolete. For example, personnel from

Information Services and Resources pointed out that “tried and true” technologies may be unable

to handle emerging video services such as two-way interactive television. Newer technologies

(alternatives #4 and #5) may be able to manage the “cable television world of tomorrow”, but it

was not certain that they would operate successfully within the confines of the actual collegiate

application explored in this paper. Note that the data network alternative (option #4) possesses

rather attractive values for either form of uncertainty. The fact that this technology was already

being sold to various Internet Service Providers certainly was beneficial. Web TV (alternative

#5) had a relatively healthy obsolescence risk since - according to various Task Force members -

it stood a reasonably good chance of being replaced by some other “futuristic” delivery mode.

Once the uncertainties were outlined, it was important to assess costs for future

“unfortunate outcomes.” In other words, one needed to quantify the costs of unsuccessful

installations and obsolete technologies. After considerable discussion between Task Force

members, the group determined that installation failure would cost $2.5 million while

obsolescence cost was quantified at $2.0 million. The group decided to not scale these costs

since they were not terribly confidential. The one exception to these specific estimates involved

Web TV with an installation failure set at $3.5 million. Admittedly, these costs were not derived

from an analytical model; nonetheless, Task Force members were confident that they were a

good approximation for these outcomes. Besides, sensitivity analysis was performed on these

values to judge how any changes could impact the decision-making results.

A copy of the decision tree is included as Figure 1. Note that each alternative is subject

to the same set of uncertainties, although probability values (and in some cases, penalty costs)

differ. The preferred outcome is for the delivery mode to be successfully installed and not to

become obsolete within a reasonable time frame. The alternative with the lowest expected

monetary value (EMV) would be deemed to represent the best option. The EMV was calculated

using the following expression:

Installation costs + [(Probability of successful installation) x (Probability of becoming

obsolete) x Obsolescence cost] + [(1 - Probability of successful installation) x (Failure

cost)]

The calculation of the specific EMVs involve a straight-forward computation. However,

the determination of specific alternatives, concomitant costs and resulting uncertainties are by no

means a trivial exercise. This represents the point at which analytical theory merges with real-

world practice.

===== insert Figure 1 about here =====

4.0 Model Results

Having identified the specific alternatives in addition to their relevant costs and the

likelihood of various uncertainties, attention can now be turned to solving the decision analysis

model. A spreadsheet add-in PrecisionTree (available from Palisade Decision Tools) was

deployed to solve the model and perform all sensitivity analyses. Albright and Winston (2015)

provide a detailed guide to using PrecisionTree.

Table 3 lists each alternative’s EMV. As with other confidential data, the results have

been scaled, using the lowest EMV as a base value. Based on the decision analysis with the

given probabilities and costs, the preferred alternative is to adopt the data network as the cable

television delivery mode. Although its installation is somewhat riskier than tried and true

technologies (such as replacing the entire cable plant), this option has relatively lower

installation costs. Further, it possesses the least amount of obsolescence risk.

===== insert Table 3 about here =====

A rather large difference is observed between the results of the best option and the EMVs

of competing technologies. Adopting UTP cable, the second best alternative, has an EMV 1.83

times the EMV of using the data network. From an EMV perspective, the least preferred

delivery mode would be to replace the entire cable plant.

Determining the preferred alternative is an important part of real-world decision analysis.

However, it was deemed crucial to show Task Force members the sensitivity of the model to

changes in key parameter values. By performing a wide range of sensitivity analyses, decision

makers can observe the degree to which model results are affected by different parameter values.

The sensitivity analysis demonstrated the impact of adjusting the probabilities, thus avoiding the

situation in which probabilities were merely static values. A few of the more interesting findings

are presented in this section. Since the preferred option involved using the data network, Task

Force members were particularly interested in any sensitivity analysis involving this specific

alternative.

A tornado diagram is illustrated in Figure 2. This graphical representation depicts the

sensitivity of the optimal option’s EMV to each of the selected parameters over user-defined

ranges. Specific parameters were incorporated that were felt to characterize the essence of the

decision analysis model. The length of each bar in the tornado diagram indicates the percentage

change in the EMV in either direction; as a result, longer bars imply that changes in the

particular input have a more substantial effect on EMV. In the decision analysis model, the data

network’s likelihood of installation success has a significant effect on EMV. Decision-makers

need to be confident that its current value of 60% is indeed an accurate representation of this

uncertainty.

===== insert Figure 2 about here =====

Figures 3 through 5 illustrate two-way strategy regions. This form of sensitivity analysis

is used to depict the best alternative under various combinations of two important model

parameters. As it turned out, these graphical representations were quite valuable in

communicating the modeling efforts to other Task Force members. Administrators and staff

members could observe the ranges over which particular alternatives were preferred.

Figure 3 compares the installation costs of the data network and UTP cable, the top two

choices for cable delivery. Cost values were removed on each axis to preserve confidentiality.

Clearly, the data network is an attractive choice as it remains the preferred option under nearly

all combinations of these two costs. Delivering signals via UTP cable makes sense only if its

installation costs decrease significantly (actually, the costs would need to drop to about 1/3 of

their current value), combined with a concomitant increase in data network costs (about a 5-fold

escalation).

===== insert Figure 3 about here =====

Figure 4 evaluates the likelihood of successful installation for the two “cutting edge”

technologies, data network and Web TV. As observed with Figure 3, the data network remains

an appealing alternative. Note that the data network is preferred to Web TV, except when the

former has minimal (near zero) chance of installation success while the latter’s probability

climbs appreciably. Figure 5 compares the failure cost of Web TV with data network’s

probability of installation success. Only when Web TV’s failure cost drops substantially is the

data network rejected as the best option.

===== insert Figures 4 and 5 about here =====

5.0 Conclusions

The paper has described the development of a decision analysis model to explore cable

television delivery options within the residence halls of Bucknell University. Working closely

with senior administrators and staff members, a model was created to consider separate

alternatives, relevant costs, and probabilities associated with various uncertainties. The results

show that this institution should adopt the data network as the mode for cable delivery.

Sensitivity analysis further supports this notion. Decision-makers need to be sure that the

probability of successful installation for the data network option is accurate, for this parameter

had the most substantial effect on the EMV of the optimal alternative.

In modeling this particular decision-making problem, it was not necessarily our aim to

advance theory. Rather, we opted to cooperate with real-world managers on developing a model

that could sharpen understanding and promote implementation. The intent was to inform

practice by illustrating the useful consequences that result from accurate analytical modeling.

We demonstrated how models support implementation.

The application of analytical modeling in this specific case offered the following

important insights:

a. In order to garner model acceptance and implementation, the analyst must convey

concepts using language and terminology understood by the non-specialist. We

accomplished this by clearly articulating features of the decision analysis tool during our

initial Task Force meetings.

b. Often, managers describe uncertain aspects of decision-making problems using terms

like "high chance", "moderate probability" or "low likelihood". Initially, they are rather

unwilling to ascribe specific quantitative assessments to these uncertainties. After

thorough discussion with Task Force members, we were able to shape their qualitative

considerations into quantitative estimates.

c. Visual representations of sensitivity analysis provide considerable managerial appeal.

The acceptance of the model's findings was augmented when Task Force members were

able to see how changes in underlying parameter values impacted the optimal solution, or

observe the comparison between two alternatives when one particular variable was

modified.

As it turned out, these findings became an additional piece of evidence in favor of

adopting the data network as the cable television delivery mode. Eventually, a full-scale roll-out

of this technology occurred. As demonstrated through this application, decision analysis

represents a powerful approach in communicating uncertainties and advising on the benefits of

particular alternatives.

Acknowledgements

Without implicating them, the authors wish to thank the following current or former

employees of Bucknell University for their valuable assistance: Brian Hoyt, Director of

Technology Integration; John Jantzi, Systems Integrator; Charles Pollock, Vice-President for

Student Affairs; Gene Spencer, Associate Vice-President for Information Services and

Resources; and Dennis Swank, Associate Vice-President for Finance. They each contributed of

their time and expertise in helping us to appropriately and effectively model this important

decision-making problem.

References

Albright S.C. and Winston, W.L. (2015), Business Analytics: Data Analysis & Decision Making, Cengage Learning, Stamford, CT. Borsuk M.E. et al. (2008), “Charting a path for innovative toilet technology using multicriteria decision analysis”, Environmental Science & Technology, Vol. 42 No. 6, pp. 1855-1862. Brooks D.G. and Kirkwood C.W. (1988), “Decision analysis to select a microcomputer networking strategy: A procedure and a case study”, Journal of the Operational Research Society, Vol. 39 No. 1, pp. 23-32. Browne D., O’Regan B. and Moles R. (2010), “Use of multi-criteria decision analysis to explore alternative domestic energy and electricity policy scenarios in an Irish city-region”, Energy, Vol. 35 No. 2, pp. 518-528. Cortes P. et al. (2001), “Decision support system for planning telecommunication networks: A case study applied to the Andalusian region”. Journal of the Operational Research Society, Vol. 52 No. 3, pp. 283-290. Chai J., Liu J.N.K. and Ngai E.W.T. (2013), “Application of decision-making techniques in supplier selection: A systematic review of literature”, Expert Systems with Applications, Vol. 40 No. 10, pp. 3872-3885. Fleming C. et al. (1993), “A decision analysis of alternative treatment strategies for clinically localized prostate cancer”, The Journal of the American Medical Association, Vol. 269 No. 20, pp. 2650-2658. Ghosh S. et al. (2009), “An adoption decision model of emerging capital budgeting methodologies”, International Journal of Operational Research, Vol. 5 No. 3, pp. 349-362. Huang I.B., Keisler J. and Linkov I. (2011), “Multi-criteria decision analysis in environmental sciences: Ten years of applications and trends”, Science of the Total Environment, Vol. 409 No. 19, pp. 3578-3594. Isik M. and Khanna M. (2003), “Stochastic technology risk preferences and adoption of site-specific technologies”, American Journal of Agricultural Economics, Vol. 85 No. 2, pp. 305-317. Jackson J.A. et al, (2007), “Selecting a portfolio of technologies: An application of decision analysis”, Decision Sciences, Vol. 30 No. 1, pp. 217-238. Johnson D.E. et al. (2002), “A decision-tree-based symbolic rule induction system for text categorization”, IBM Systems Journal, Vol. 41 No. 3, pp. 428-437.

Keren R et al. (2002), “Projected cost-effectiveness of statewide universal newborn hearing screening”, Pediatrics, Vol. 110 No. 5, pp. 855-864. Khalili N.R. and Duecker S. (2013), “Application of multi-criteria decision analysis in design of sustainable environmental management system framework”, Journal of Cleaner Production, Vol. 47, pp. 188-198. Kirkwood C.W., Slaven M.P. and Maltz A. (2005), “Improving supply-chain-reconfiguration decisions at IBM”, Interfaces, Vol. 35 No. 6, pp. 460-473. Kocher M.S. et al. (2002), “Operative versus nonoperative management of acute Achilles tendon rupture: Expected-value decision analysis”, The American Journal of Sports Medicine, Vol. 30 No. 6, pp. 783-790. Linkov I et al. (2012). “Civilian response corps force review: The application of multi-criteria decision analysis to prioritize skills required for future diplomatic missions”, Journal of Multi-Criteria Decision Analysis, Vol. 19 No. 3-4, pp. 155-168. Machado T.C.S., Pinheiro P.R. and Tamanini I. (2015), “Project management aided by verbal decision analysis approaches: A case study for the selection of the best SCRUM practices”, International Transactions in Operations Research, Vol. 22 No. 2, pp. 287-312. Martonosi S.E. and Barnett A. (2006), “How effective is security screening of airline passengers?”, Interfaces, Vol. 36 No. 6, pp. 545-552. Miller L.T. and Park C.S. (2002), “Decision making under uncertainty – Real options to the rescue”, Engineering Economist, Vol. 47 No. 2, pp. 105-150. Mosadeghi R. et al. (2013), “Uncertainty analysis in the application of multi-criteria decision-making methods in Australian strategic environmental decisions”, Journal of Environmental Planning and Management, Vol. 56 No. 8, pp. 1097-1124. Ngwenyama O., Guergachi A. and McLaren T. (2007), “Using the learning curve to maximize IT productivity: A decision analysis model for timing software upgrades”, International Journal of Production Economics, Vol. 105 No. 2, pp. 524-535. Stonebraker J.S. (2002), “How Bayer makes decisions to develop new drugs”, Interfaces, Vol. 32 No. 6, pp. 77-90. Taris T.W. (1998), “Fertility in the Netherlands as an expected value process and developmental readiness”, The Journal of Psychology, Vol. 132 No. 1, pp. 61-77. Ulvila J. (1987), “Postal automation (Zip+4) technology: A decision analysis”, Interfaces, Vol. 17 No. 2, pp. 1-12.

Table 1: Installation costs

Alternative Installation costs

#1: Replace entire cable plant 1.84

#2: Existing fiberoptics coax 1.50

#3: UTP cable 1.00

#4: Data network 0.38

#5: Web TV 0.10

Table 2: Uncertainty estimates

Alternative Probability of

successful installation

Obsolescence risk

#1: Replace entire cable plant 0.97 0.98

#2: Existing fiberoptics coax 0.98 0.98

#3: UTP cable 0.90 0.80

#4: Data network 0.60 0.20

#5: Web TV 0.10 0.40

Figure 1: Cable television technology decision tree

Cable TVTechnology

Failure Cost

ObsolescenceCost

SuccessfulInstallation

?

Becomes Obsolete

?

Failure Cost

ObsolescenceCost

SuccessfulInstallation

?

Becomes Obsolete

?

Failure Cost

ObsolescenceCost

SuccessfulInstallation

?

Becomes Obsolete

?

Failure Cost

ObsolescenceCost

SuccessfulInstallation

?

Becomes Obsolete

?

Failure Cost

ObsolescenceCost

SuccessfulInstallation

?

Becomes Obsolete

?

Replace EntireCable Plant

ExistingFiber-optics, Coax

UTPCable

DataNetwork

WebTV

Table 3: Expected Monetary Value (EMV) results

Alternative Expected Monetary Value

#4: Data network 1.00

#3: UTP cable 1.83

#5: Web TV 2.14

#2: Existing fiberoptics coax 2.47

#1: Replace entire cable plant 2.71

Figure 2: Tornado Diagram

-40.0% -20.0% 0.0% 20.0% 40.0% 60.0% 80.0%

Obsolescence cost

Failure cost Web TV

Data network cost

Prob. success data net.

% Change from Base Value

Tornado Diagram for EMV

Figure 3: Sensitivity analysis (installation costs)

Cos

t of U

TP c

able

Cost of data network

Comparing the installation costs of two alternatives

3 : UTP cable

4 : Data network

Figure 4: Sensitivity analysis (Installation probabilities)

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1

Prob

. suc

cess

Web

TV

Prob. success Data network

Comparing the Probabilities of Successful Installation

4 : Data network

5 : Web TV

Figure 5: Sensitivity analysis (Failure cost and installation probability)

0 0.2 0.4 0.6 0.8 1

Failu

re c

ost W

eb T

V

Prob. success Data network

Comparing a failure cost and a probability of successful installation

4 : Data network

5 : Web TV