Data Envelopment Analysis in Health Care - Thesis Example

DATA ENVELOPMENT ANALYSIS IN HEALTH CARE By Presented to DATA ENVELOPMENT ANALYSIS IN HEALTH CARE I. IntroductionSince data envelopment analysis (DEA) was initially brought in by Charnes, Cooper and Rhodes in 1978, the just authoritative technique has been vastly advanced and utilized to evaluate the comparative efficiencies of multiple-response multiple-output decision-making units (DMU’s). The acceptance of DEA is because of its capacity to measure comparative efficiencies of multiple-input and multiple-output DMU’s devoid of early masses on the contributions and productions. Presently, DEA is still broadly studied and being used in a lot of areas and fields. In this case, the medicine commerce is one of the most prominent industries that make use of this envelopment approaches. Efficiency capacity has been a domain of incredible interest as firms have strived to better efficiency. Causes for this focus were best claimed half a century ago by Farrell in his archetypal document on the measurement of dynamic efficiency. Farrell additionally stated that that chief cause that all trials to resolve the issue and failed, was because of a failure to join the measurements of the numerous inputs into whichever reasonable measure of efficiency (Kuah, Wong & Behrouzi, 2010, p. 3). These insufficient methods creating standard proficiency for one input while disregarding all additional inputs, and building an index of efficiency in which a weighted standard of inputs is compared with efficiency. Reacting to these insufficiencies of distinct indices of labor proficiency and capital profitability were suggested by Farrell an operation scrutiny method that can more efficiently tackle the issue (Cook 7 Seiford, 2009, p. 5). His measures were intended to be applicable to whichever productive firm. From a workshop to an entire financial system, Farrell apparently confined his digital cases and discussion to one output situation, even though Farrell was capable of making a multiple output case (Kuah, Wong & Behrouzi, 2010, p. 3). The provision of a reasonable assessment of every DMU could be argued out that in whichever provided criteria ought to be contrasted simply to other units in similar or less-deprived criteria. A DMU under substantial struggle would be unethically punished if contrasted to units in importantly more promising modest surroundings. II. Literature review A. Definitions of efficiency In the nonexistence of renowned multipliers such as Charnes, Cooper and Rhodes and many others, anticipated deriving suitable multipliers for a provided DMU by resolving a specific non-linear software design issue. Particularly, if DMU is being designed under consideration, the CCR model can be a perfect definition of efficiency in data envelopment scrutiny because it can be the ideal model for measuring the practical efficiency of that DMU is provided by the resolution to the factional programming issue (Kuah, Wong & Behrouzi, 2010, p. 3). 1. Pareto-Koopmans efficiency The Pareto-Koopmans efficiency gives the consistent returns to scale (CRS). It is pragmatic that in their overall 1978 document, all writers just limited the variables to be positive, such that (e = 0); the imposition of a rigorously non-negative lesser restraint, (e > 0) was brought in a supplement document. For convenience, the writers referred to (2.1) as the initial CCR model. When implementing the CCR model in 1962, alongside the hypothesis of factional programming, making the adjustment of variables, lr = tur and ti = tvi, where t ¼ ð P imixioÞ_1, problem (2.1), could be transformed to the linear programming (LP) paradigm. This model is otherwise termed as the additive models are able to modify CCR and BCC paradigms to be additive models. This is usually done with the aim of joining both alignments so as to demonstrate this notion wherein any course in the quadrant created by B-A-C is allowed (Cook 7 Seiford, 2009, p. 3). 2. Technical efficiency Scientists have integrated stochastic input and output dissimilarities into DEA models. Particularly, stochastic DEA models permit the presence of information flaws and give problematic findings. For instance, applications have such consequences, stochastic DEA and opportunity limited programming methods were used to assess methodical competences in the healthcare system. This indicated the usage of stochastic DEA on input relaxation measure of efficiency. The stochastic information to approximate the most efficient scale in size in DEA/ technical efficiency allows the deployment of opportunities limited programming methods to tackle congestion problems in stochastic DEA (Kuah, Wong & Behrouzi, 2010, p. 3). The score of cross efficiency of a provided DMU is attained by computing of that DMU the set of n technical efficiency scores. Utilizing the n groups of optimal weights matching to the n DMU’s, and then standardizing these scores has to be achieved through manual computation of the variables required. Therefore, cross efficiency develops further than pure self-assessment inherent in conventional DEA analysis, and joins with the additional increasing scores from the optimal peer multipliers. This strategy was brought further by sexton, and additionally researched into by Doyle and Green and other philosophers (Cook 7 Seiford, 2009, p. 3). Cross efficiency gives an efficiency ordering amongst each DMU to distinguish between good and pitiable performers. Most recently, Ruggiero has highlighted that a number of cases, the Banker and Morey Model could over estimate technical efficiency by permitting production impossibilities into referent set. Several studies have aimed on the issue of modeling technical efficiency when the formation for the inputs and outputs were indiscriminate variables. The index could be divided into two elements, with a single one with one computing the adjustment in the technology leading edge and the other the change in technical efficiency. The following modification of Mo makes it likely to measure the change of technical efficiency, and the movement of the leading edge in terms of a detailed DMUo. B. CCR model The CCR DEA model was the opening DEA exemplar suggested in the opening research phase and named after the three researchers (Charnes, Cooper and Rhodes) behind its discovery and development. The CCR model is suitably referred to as giving a circular estimate. Particularly every input is decreased by a similar proportionality element. Making use of the CCR model and hypothesis of fractional indoctrination creates the adjustment of variables. One kind of CCR model focuses on reducing inputs whereas filling a minimum of provided outputs. This is known as the input-based model. The additional aims models aim at maximizing output levels with similar levels of inputs. This is known as output-based model. Nevertheless the following paper is going to tackle the input-based model (Kuah, Wong & Behrouzi, 2010, p. 3). 1. The multiplier form Reflect there are nDMUs: DMU1, DMU2, and DMUn. Each DMUj, (j = 1, 2, n) uses m inputs xij (i = 1, m) and produces s outputs yrj (r = 1, s). Let the input masses be vi (i = 1, m) and the output masses ur (r = 1, s) as variables. Let the DMUj to be assessed on whichever sample be labeled as DMU0 (0 = 1, 2, n). The efficiency of every DMU0, e0, is consequently set up by resolving the linear programming, which is recognized as multiplier method in DEA. There are a number of editions of the additive model and the most fundamental part being the provided linear optimization issue indicated in the aforementioned multiplier form (Cook 7 Seiford, 2009, p. 5). The convexity circumstance on the kj variables suggests that the multiplier form makes use of the VRS technology. The leading edge produced by this model is similar to that coming from the matching VRS outline. Therefore, a DMU additive-varying or PK efficient, where all slacks are equivalent to zero at the maximum point occurs if and only if it is the VRS efficient. Evidently, the CRS production likelihood group could be made use of. Po ¼ max Pi s_i þ Pr sþrs:t: Pj kjxij þ s_i ¼ xio; i ¼ 1; . . . ;m Pj kjyrj _ sþr ¼ yro; r ¼ 1; . . . ; s Pj kj ¼ 1 kj; s_i ; sþr P 0; 8j; i; r: ð2:7Þ Since the various inputs and outputs might be measured in non-proportionate units. It might not be technical in definite backgrounds to make use of an easy sum of slacks as the goal presented in the multiplier form above (Kuah, Wong & Behrouzi, 2010, p. 3). Additionally, the CCR paradigm does not give an actual measure of inefficiency as in the case for the BCC and CCR models. To overcome this latter issue, Charnes, Cooper and Rhodes suggested the usage of Qo, where: Qo ¼ d (X is_i =xio þ + Xrsþr) =yro Qo is subject to the limits projected in the multiplier form and an implied value for d was 1/ (m + s). 2. Production possibility set and its properties A production possibility set (PPS) or reference technology could be believed to be a statement of the totality of production operations that may reasonably observed on the proof of the operations observed. In the healthcare industries, record keeping and financial statements of patients, insurance companies and accounting unit of the hospital could discretely apply this reference technology to evade faults in the patients’ confidence and financial contexts. The dealings that the hospital has made with all patients will be systematically arranged over the hospital’s data base, and all partners or competitors can learn and strategize from a single’s hospitals record keeping strategy (Kuah, Wong & Behrouzi, 2010, p. 3). The efficiency of pharmaceuticals and therapy-providing firms can be spelled out by the PPS as well since the properties of the leading edge space settle at 2.3. This way, the healthcare institutions describes its PPS set T as: T ¼ ðX; Y Þ j X P Xj kjXj; Y 6Xj kjY j; kj P 0 So as to attain a geometric appreciation for the CRS model, an individual could stand for an issue. The multiplier form fives an indication of one output single input case. If resolved for every DMU, this amounts to sticking out DMU to the left, to a point on the leading edge. 3. The envelopment form For each inefficient DMU, CCR recognizes a set of matching efficient DMU’s that are able to benchmark for development. The benchmarks could be attained from the dual issue of division. The model could be written as envelopment form, as indicated in the aforementioned multiplier forms. This way, readers, interpreters and implementers can be able to refer to the presented situation on converting a multiplier form linear programming into envelopment form. A subsequent issue has an ideal resolution when θ = 1, λ = 1, and λ ≠ 0. The DMUs with the maximum resolution are referred to as efficient limitations formed by efficient DMU’s by decreasing every input by the proportionality element θ. This element if obtained from the envelopment model aforementioned, whilst sustaining the output intensities. Another means of sticking out the inefficient DMU’s to the efficient leading edge is by maximizing the outputs by a proportionality element of 1/ θ (Kuah, Wong & Behrouzi, 2010, p. 3). 4. Definition of the CCR efficiency, Radial and technical efficiency Whichever CCR efficient DMU is additionally BBC efficient. The profits to scale (RTS) categorization of DMU’s have been the title of the research by many writers, as well as Banker in 1984. Banker made use of the most efficient scale dimension idea and permitting the totality of lambda values dictate the RTS (Cook 7 Seiford, 2009, p. 6). Banker also applied the liberated variable and their scale efficiency catalog technique. An issue categorizing RTS is the presence of numerous goals, implying that the categorization might be a role of the specific resolution chosen by the optimization software. A number of examinations have been conducted to give more certain RTS categorization task for a provided DMU, as well as advancing intervals for the various liberated variables coming from the numerous goals. A solution for CCR techniques has been suggested by other envelopment writers and designers, Zhu and Shen, beneath the numerous objectives in 1995 (Kuah, Wong & Behrouzi, 2010, p. 3). Afterwards, Seiford and Zhu analyzed the various techniques and automated suggestions simplified techniques that illustrate RTS, and evade the necessity the exploration of all options for optimal resolutions. Such models can be used in healthcare commerce since the existing solutions to data envelopment cause flaws that have been diversified to be evadable. Mismanagement of patient funds, security and relationships with doctors can be fully guaranteed with the implementation of the installation of an advanced solution similar to the model defined by Zhu and Shen, and CCR. 5. The orientation set and development in efficiency ( the formula for improvement which called CCR projection) The preceding two models of efficiency in DEA are radial prognosis designs. Particularly, the input-based situation has inputs that are proportionally decreased whilst outputs remain fixed. For the output-based case, outputs are proportionately maximized whereas inputs are supported constantly. CCR brought in the additive or Pareto–Koopmans (PK) exemplar which to a certain level, joins both directions that illustrate the concept. Therefore, any course in the quadrant created by B-A-C is allowed. Several writers have tested the issue of obtaining the minimum distance projection to the efficient leading edge. It should be noted however that this is the conflicting category to that of the additive model that explorations for the greatest distance. In 1991, Frei and Harker suggested the usage of the Euclidean standard to describe the bordering point. Charnes attained the minimum city block distance to the weak efficient leading edge. Gonzalez and Alvarez reduce input contractions whilst the writers of the CCR model approached this by recognizing all efficient elements features (Kuah, Wong & Behrouzi, 2010, p. 8). In a recent document, Aparicio bring forward a set of prototypes for getting a minimum distance prognosis. All stated projections and data envelopment analysis is an input-oriented model. A Non-Archimedean value constructed to enforce rigorous positivity on the variables. It should be highlighted that this model engages the percentage of outputs to inputs is referred to as the input-oriented model. One can overturn this ratio and resolve the matching output-based minimization issue too. The utilization of this model in any healthcare institute will generally cope with input-based model in this. The subsequent issue is entitled the CCR model and gives consistent returns to scale. It is observed that in their initial 1978 document, both authors just limited the variables to be positive (e=0); the obligation of a rigorously decreased constraint (e > 0) was brought in in a continuation document. The content and certification of the document denotes the needs and fills in the necessities of institutions in the healthcare industry that need the assistance of such models to experience maximized input and efficiency (Kuah, Wong & Behrouzi, 2010, p. 8). 6. Inverting the ratio of input oriented model Relating the Charnes, Cooper and Rhodes hypothesis of fractional programming made the adjustments of variables: lr = tur and ti = tvi, where t ¼ ð P imixioÞ_1, problem (2.1) could be changed to the linear programming (LP) exemplar: e0 ¼ max Prlryros:t: (Pi tixio ¼ 1+Pr lryrj _Pi tixij 6 0; 8j lr; ti P e; all r; i: ð2:2Þ By dualism, this issue is similar to the linear programming issue: {Min ho _ ePr sþrþ-Pis_i_ _s:t: (Pjkjxij þ s_i ¼ hoxio; i ¼ 1; . . . ;mPj+kjyrj _ sþr¼ yro; r ¼ 1)} kj; s_i ; sþr P 0; 8i; j; rho unrestrained: ð2:3Þ Problem (2.3) is denoted as the envelopment or primitive issue, and the multiplier or twofold issue. The limit space of such a presented issue defines the production probability set T b (Kuah, Wong & Behrouzi, 2010, p. 8). That is, T ¼ ðX; Y Þ j X-P {(XjkjXj; Y 6Xj+kjY j; kj P)} 0 The multiplier form presented above gives an image of one output one input case. If the case is resolved for every DMU, this ends in the projection of DMU towards the left to a point on the leading edge. In this case, DMU on the third leading edge has its projection to the frontier is represented by the point. Intuitively, one would reasonably measure the efficiency of DMU as the ratio A/B = 4.2/6 = 0.7 or 70%. The answer of (2.3) for this DMU effects in h_3 ¼:70. It is very helpful to denote that the redefinition performance by the reciprocal ratio of inputs and outputs, the final value of that ratio would be 1.43 (Cook 7 Seiford, 2009, p. 4). Healthcare facilities would have to consider the efficiency to be 1/1.43 = .70. This value is similar to the reached value. Pictorially, resolving this output-oriented exemplar engages vertical projection from DMU up to the leading edge, instead of an upright projection towards the left as indicated in the case’s multiplier selected form. 7. Limitations of the CCR model To begin with, an inefficient DMU and its standardizing DMU’s might not be the same as in their activities. The actual cause pointed out in this scenario is because of the composite DMU that does not exist in reality. To surmount this issue, performance-oriented grouping methodologies have been utilized by scientists to group similar DMU’s into groups. The efficient DMU’s in a group used to standardize other unproductive DMU’s in a specific group. The other limitation of CCR exemplar is that it presumes consistent return to scale (CRS) that might be false for a number of applications. To cope with this problem, scientists have executed variable return scale (VRS) into the initial DEA model. The fundamental VRS model is recognized as BCC (Banker, Charnes, Cooper) paradigm that will be explained during the usage of the model in an organization (Cook 7 Seiford, 2009, p. 3). 8. Why we solve the CCR model using the envelopment form? A recent advancement of RM is made publicly official when using the second-order conduit programming (SOCP) method could give a precise RM score (Cook 7 Seiford, 2009, p. 3). Their study is a leading breakthrough in RM measurement. Interested readers and designing programmers are commended to mention from their study document. C. The BCC model In the BCC paradigm, VRS is presumed and the efficient leading edge is created by the U-shaped exterior of the present DMU’s. The envelopment form of BCC and additional limitations point out that BCC varies from CCR in that it has the additional convexity limitation. D. Differences between CCR and BCC model The dissimilarity amid the CCR and BCC model is the sensitivity of DEA and the efficiency score attained from the BCC model. This model can be tested repeatedly sampling from the initial samples (Cook 7 Seiford, 2009, p. 2). On the other hand, the CCR model is suitably referred to as giving a radial projection. Particularly, every input is decreased by similar proportionality elements. E. The additive model The preceding two efficiency models are outspread projection designs. Particularly, in the input-based case, inputs are proportionally deceased whilst outputs remain permanent. For the output-based situation, outputs are proportionally maximized whilst inputs remain consistent. A limitation of the additive model was it does not give an actual measure of efficiency. To tackle this problem, a measure of efficiency for additive model was suggested by the writers of the initial two models (Cook 7 Seiford, 2009, p. 3). F. Slacks-based measures To overcome the weakness in the additive model, a variant of additive model can be suggested in a case similar to the aforementioned multiplier forms and equation representation of the DEA (Cook 7 Seiford, 2009, p. 6). Namely, the Slacks-based model (SRM) brings forward a measure of efficiency when subjecting similar limitations. This insignificant programming issue could be changed into linear programming as indicated in most of the situations. The optimum resolution of SBM is one that could simply be accomplished when every slack is equivalent to zero. This is in line with CCR and BCC models. G. The Russell measure Another DEA exemplar that is significant in the technique advancement of DEA is The Russell measure (RM). The RM was brought in through formulation by Fare and Lovell. RM groups both output and input efficiencies in the framework of a radial measure and treatments amid input and output orientation which was not needed (Kuah, Wong & Behrouzi, 2010, p. 11). RM, in spite of its practicality, it was too complicated in computation wise since it was made as a nonlinear programming issue. RM was then revisited by a figure of scientists. For example, RM was lengthened into a developed exemplar known as Enhanced Russell Graph Measure (EGRM). EGRM has decreased the computational impediment of RM, since it was a compromise; it could simply provide an approximation of RM scores (Cook 7 Seiford, 2009, p. 3). H. Brief of the remaining models of DEA Dissimilarity amid the radial method and non-radial method in DEA exists in the representation on input or output items. Radial inputs and outputs alter proportionality whilst non-radial inputs and outputs adjustment non-proportionality (Cook 7 Seiford, 2009, p. 3). Free disposal Hull (FDH) model is another sample that has received a substantial quantity of scientists’ attention (Cook 7 Seiford, 2009, p. 3). Rather than the convexity presumption that majority of DEA models usage, FDH uses the presumption that the efficient leading guide is designed simply by observed DMU’s. Cross efficiency model that was initiated by Sexton, Silkman and Hogan permits a DMU to be assessed by just with individual masses. Nevertheless, they also had all other DMU’s optimum masses. In cross efficiency model, initially the optimal masses for the inputs and outputs for every DMU are recorded utilizing a normal process (Kuah, Wong & Behrouzi, 2010, p. 16). I. The data Rutledge, R. W. 1995 Assessing hospital efficiency over time: An empirical application of Data Envelopment Analysis. Journal of information technology management, volume VI, number 1 Sherman was the opening noticed application of the DEA technique to hospitals. He made use of DEA to the clinical-surgical region of seven teaching clinics in Massachusetts. The inputs were described as non-surgical equivalents (FTEs), supplied, and bed-days accessible. The outputs were described as patient days for two age groups nurses competent, and medical students and residents’ proficient (Rutledge, 1995, p. 15). Two of the hospitals were recognized as comparatively inefficient. Two hospitals specialists agreed with the DEA evaluation. The management of one of the inefficient hospitals was capable of discovering a number of the likely reason for inefficiencies. Therefore, using the cross-sectional information, Sherman was capable of demonstrating the capacity of DEA in directing management efforts in developing hospital efficiency (Rutledge, 1995, p. 15). Banker, the writer of the BBC model, contrasts both DEA and econometric translog cost purpose methods. The research examined 114 hospitals from North Carolina and inputs were described as nursing services, auxiliary services, overall and managerial services of the highest quality and accessible. Outputs regarded consisted of patient days for three age groups. The two techniques generated similar measurements of efficiencies with a single exclusion. The findings from the translog model would generate a standardizing consequence for economies of scale, therefore implying general consistent returns. The DEA exemplar was capable of recognizing cases of both rising and lowering returns. The lowering was discovered for hospitals with augmented quantities of elderly patients (Rutledge, 1995, p. 15). Borden made use of the DEA technique in 1998 to test the efficiency of 52 New Jersey hospitals. Medicare recompensed for all hospitals was diagnosis-associated sets oriented by 1982. A number of clinical facilities stated making use of DRG’s as a patient publicizing foundation at an untimely date. Borden made use of DEA to test the hospital’s efficiency. The input measures were total FTE’s, nursing FTE’s, a figure of beds and additional non-payroll costs (Rutledge, 1995, p. 16). The number of cases treated in every DRG-oriented criterion was deployed as an output measure. The DEA resolution implied that clinical facilities worked under the DRG-oriented compensation system for an extended period of time were additionally efficient than hospitals that had not functioned under the DRG system as long. No important dissimilarity was noted expending the proportion or regression techniques. Banker demonstrates how DEA could be utilized in setting cost principles for clinical facilities’ use in variance scrutiny. They tested 111 North Carolina clinical facilities. The inputs of the clinical facilities were described as the expense of every five units; nursing, test center and certified, domestic and sustenance, management and nutritional services (Rutledge, 1995, p. 16). Outputs were described as patient days of every eight domain: medicine, surgical treatment, obstetrics, gynecology, psychology, eyes-ears-nose-and-throat, urology, and orthopedics. The research illustrated how making use of DEA to set price principles permits for principles that could be 100% efficient oriented. Alternatively, principles might be at additional relaxed levels like standard productivity levels or other arbitrary quantities (Rutledge, 1995, p. 16). Cooper, P., Seiford, L. and Zhu, T. 2010. Handbook on Data Envelopment Analysis. New York: McHill There is last term on how to outline a structure of a healthcare output model. Every DEA research claims to capture some region of reality. Therefore, the important question in this step the rationality brought about by the production model. There is supposed to be a consistent justification of the inputs and outputs selected (Cooper, Seiford, & Zhu, 2010, p. 482). This could be based on the literature before knowledge and specialized information. The utility of medical specialties is vital if the application is to turn out to be helpful for practice. An interconnected query is the choice of the DEA exemplars. Since the robust intuitive appeal, the CCR (clinical and scale efficiency) and BBC (pure clinical efficiency) paradigms have been used in a majority of healthcare researches (Cooper, Seiford, & Zhu, 2010, p. 482). The multiplicative model for production roles and the additive paradigm barely seem regular n the healthcare industry. Scientists ought to consider the underlying production technology and take a new look at the selection of the DEA exemplars. If the justification for the last choice of a DEA model is sketchy, consider running multiple cases (Cooper, Seiford, & Zhu, 2010, p. 483). There is constantly some difficulty getting the variables for the research records that include the inputs, output, controls, explanatory variables, and similar to other documents produced when implementing these models. Occasionally, a doctor, clinical facility, medical association, and indemnity databanks are supposed to be combined into one research document to connect DEA with additional variables (Cooper, Seiford, & Zhu, 2010, p. 507). DEA presumes that a model is evaluating the efficiency of relative findings, not output variations. Prior to running an efficiency evaluation, if there is a cause to think that outputs are heterogeneous, it is commended that pressure groups be advanced. In the healthcare commerce based on clinical subspecialty (orthopedic surgeons against cardiac surgeons), indicative complication, and additional commodity variations are very significant for the healthcare providing facility (Cooper, Seiford, & Zhu, 2010, p. 508). J. Inputs and outputs that were identified and justified The average number of hours worked by doctors, specialists and consultants per unit time-17 hours Nursing services (the average number of hours worked by nursing service personnel per unit time)-16 hours Ancillary services (the average number of hours worked by ancillary service personnel, which includes medical services for physical therapy, radiology and pharmacy, per unit time)-14 hours Support services (the average number of hours worked by support service personnel, which includes housekeeping, dietary, laundry, business office, medical records and security, per unit time)-19 hours The average amount spent on supplies per unit time-7 hours The average number of bed-days available-49 bed-days Output measures: The percentage of patients with minor injuries who recover satisfactorily-100% The percentage of patients with moderate injuries who recover satisfactorily-40% The percentage of patients with severe injuries who recover satisfactorily-21% The average period of stay per patient-57 hours The average number of nurses trained per unit time-7 nurses The average number of interns trained per unit time-12 interns The average number of minor surgical operations per unit time-3 minor surgical operations The average number of major surgical operations per unit time-2 major surgical operations The average number of treatments provided by emergency services per unit time-26 treatments K. The planning work Patients with severe head trauma can be clinically assessed by clinicians through the application of a number of methods recognized for a group of variables and a number of empirical measures based on a number of vital queries. The organizational outline and managerial elements provide and slow the implementation of the features and case mix variables (Kuah, Wong & Behrouzi, 2010, p. 13). The objective of this DEA technique is to design a conceptual map that recognizes a number of the obstacles, or explanatory variables linked to the highest quality practices from the industry and specialized knowledge. If the theory is not strong enough, the researchers ought to try a number of the easy maps and outlines that increase hypothetical problems. Now the scientist is prepared for the methodical research of the databanks that gather information for the data obtained (Cook 7 Seiford, 2009, p. 3). L. References American Society of Mechanical Engineers. 1959. Resistor Meters: Their Theory and Application. Benninga, S. and Czaczkes, B. 2000. Financial modeling, Volume 1. California: MIT Press Buzacott, J. A. and Shanthikumar, J. G. 1993. Stochastic models of manufacturing systems. Michigan: the University of Michigan Charnes, J. 2011. Financial Modeling with Crystal Ball and Excel. New York: John Wiley and Sons p. 142 Chase, J. Operations management for competitive advantage. Michigan: Springer Cook, W. D. and Seiford, L. M. 2009. Data envelopment analysis (DEA) – Thirty years on. European Journal of Operational Research 192 (2009) 1–17 Cooper, P., Seiford, L. and Zhu, T. 2010. Handbook on Data Envelopment Analysis. New York: McHill Fabozzi, F. J., Focardi, S. M., Focardi, S. and Kolm, P. N. 2006. Financial modeling of the equity market: from CAPM to cointegration. New York: John Wiley and Sons Focardi, S. and Fabozzi, F. J. 2004. The mathematics of financial modeling and investment management. New York: John Wiley and Sons p. 391 Hill, A. V. 2011. The Encyclopedia of Operations Management: A Field Manual and Glossary of Operations Management Terms and Concepts. London: FT Press International Energy Agency (IEA) and Organization for Economic Co-operation and Development (OECD), 2004. Hydrogen and fuel cells. California: OECD Publishing Johnson, R. A., Newell, W. T. and Vergin, C. 1972. Operations management: a systems concept. New York: Houghton Mifflin Jondeau, E., Poon, S. and Rockinger, M. 2007. Financial modeling under non-Gaussian distributions. Boston: Springer Kuah, C. T., Wong, K. Y. and Behrouzi, F. 2010. A Review on Data Envelopment Analysis (DEA). Faculty of Mechanical Engineering. Malaysia: University of Technology Malaysia Lee, T. H., Shiba, S. and Wood, R. C. 1999. Integrated Management Systems: A Practical Approach to Transforming Organizations. London: Wiley Levin, R. I. 1972. Production operations management: contemporary policy for managing operating systems. New York: McGraw-Hill Mackinlay, A. C. 1997. Event Studies in Economics and Finance. Journal of Economic Literature, Vol. XXXV (March), pp. 13–39 Manganelli, S. and Engle, R. F. 2001. Value at risk models in finance. London: European central bank 41 pages Mark, L.S. 1951. Electrical Engineers Handbook. New York: McGraw-Hill Book Co., Inc. McCutcheon, D. M. and Meredith, J. R. 1993. Conducting case study research in operations management. Journal of Operations Management, Volume 11, Issue 3, September, Pages 239–256 Meredith, J. 1998. Building operations management theory through case and field research. Journal of Operations Management, Volume 16, Issue 4, July 1998, Pages 441–454 Meyler, K., Fuller, C., Joyner, J. and Dominey, A. 2008. System Center Operations Manager 2007 Unleashed. California: Sams Publishing New York: American Society of Mechanical Engineers Ower, E. and R.C. Pankhurst. 1966. The Measurement of resistance, 4th Ed. London: Pergamon Press Price, B., Mueller, J. P. and Fenstermacher, S. 2007. Mastering System Center Operations Manager 2007. New York: John Wiley and Sons Pyle, W., Healy, J. and Cortez, R., 1994. Solar Hydrogen Production by Electrolysis. Home Power #39, [online] 15 October. Available at: [Accessed 17 March 2005] Rajeshwar, K., McConnell, D. R. and Licht, S., 2008. Solar hydrogen generation: toward a renewable energy future. New York: Springer Roy, R. N. 2007. A Modern Approach to Operations Management. Michigan: New Age International Rutledge, R. W. Assessing hospital efficiency over time: An empirical application of Data Envelopment Analysis. Journal of information technology management, volume VI, number 1, 1995 Swan, J., Newell, S. and Robertson M. The illusion of ‘best practice’ in information systems for operations management. European Journal of Information Systems, Volume 8, Number 4, 1 December 1999, pp. 284-293(10) Taguchi, G. 1989. Quality engineering in production systems. London: McGraw-Hill College Tran, E., 1999. Verification /Validation/Certification, Dependable Embedded Systems [online] Available at: http://www.ece.cmu.edu/~koopman/des_s99/verification/index.html [Accessed 31 august 2009] Vollaro, R.F. 1976. The Use of Type S resistors for the Measurement of Low resistance and rejection. North Carolina: U.S. Environmental Protection Agency, Emission Measurement Branch, Research Triangle Park Wild, R. 2002. Essentials of Operations Management. California: Cengage Learning EMEA Zelkowitz, V. M. and Wallace, D., 1997. Experimental validation in software engineering. Information and Software Technology, Volume 39, Issue 11, 1997, Pages 735-743 Read More