Racism and ethnicity Essay

Joseph Addison once said that, â€Å"If men would consider not so much where they differ, as wherein they agree, there would be far less of uncharitableness and angry feeling in the world. † This holds true to the sentiments I have when it comes to the numerous injustices people incur due to their skin color, ethnicity, and the like. I have seen and witnessed first hand the harm people can cause to one another due to petty differences between them; and I have also often pondered why such acts occur in our world. I used to think that this may be an isolated case; a freak phenomenon that has occurred solely in the confines of my home land. A social phenomenon that has existed in my home land due to the history our people have gone through. Yet, I have seen that such injustices and anger occur in other lands such as the United States of America and knowing of this makes me wonder if racism can be found in all societies regardless of geographic location and history. To get a better understanding of this social phenomenon I have decided to look into the cases of racism I have seen in my home country of Serbia and compare that to those acts I have seen here in America. Before I do proceed, I think it is necessary to take into consideration what racism really is. Many scholars have given various definitions to this phenomenon and the differences in the definitions can be attributed to the fact that the term covers a broad spectrum of implications of race-based bigotry, prejudice, violence, oppression, stereotyping or discrimination. Since it covers such a wide array of topics and takes into account various social issues we can take into account 2 general definitions of racism; the sociological and the legal. Racism is broadly defined as a form of discrimination based on characteristics of race and existing either as individual racism, which originates in the racist beliefs of a single person, or institutional racism, which occurs when racist ideas and practices are embodied in the folkways, mores and norms (Leeder. 2003). On the other hand, sociologists Noel Cazenave and Darlene Alvarez Maddern define racism as â€Å"†¦ a highly organized system of ‘race’-based group privilege that operates at every level of society and is held together by a sophisticated ideology of color/’race’ supremacy. Racist systems include, but cannot be reduced to, racial bigotry,† (Cazenave and Maddern 1999: 42). Based on these definitions we can see two dominant themes when we speak about racism. The first among the two is the fact that racism takes into account the psyche of an individual. To be more precise, we deal with the construct of beliefs a person or group may hold against others as embodied by the mores and norms they have. The second aspect that we find is the fact that racism is a highly organized group structure as pointed out by Noel Cazenave and Darlene Alvarez Maddern. With regard to the second aspect, we find that racism is a social construct, a privilege of certain groups within a society. Hence, we are led to conclude that the social phenomenon can very well exist in any society. After all, there is no true homogenous society and the fact remain that differences abound between groups. Take the United States of America, though it is a single country the social structure can still clearly delineated between the different races that occupy its geography. As for my home country of Serbia, we also see that racism cuts across groups as I have seen people display acts of racial discrimination on gypsies; a group who since their unexplained appearance in Europe over nine centuries ago, the gypsies have refused to fall in with conventional settled life. They remain a people whose culture and customs are beset with misunderstanding, and who cling to their distinct identity in the teeth of persistent rejection and pressure to conform. This social group has been long been ridiculed and persecuted in Serbia. I have also personally seen people who look like neo-Nazis with their shaved heads launching verbal assaults and beating up gypsies. It is a saddening truth that things like this happen. What’s worse is the fact that gypsies are actually persecuted all around Europe. Other shocking instances of racial acts can also be seen in soccer games in Serbia. An example of this is even cited on a blog/news commentary on the Fox Sports website.

The Status of the Company

Running Head: AVON PRODUCTS, INC. 1 1. Provide a brief description of the status of the company that led to its determination that a change was necessary. Avon Products, Inc. (Avon) is a 122 year old company whose primary focus is on the economic empowerment of women around the world. Basically, the organization is a leader in direct distribution of cosmetics, fragrances and skin care products. Prior to and including the year 2005, the company was considered to be a very successful company operating in over 40 countries with 70% of its revenue from outside of the United States.Its growth rate on profit margin was outstanding. In 2006, the company found itself in a declining state in revenue and profits. The company’s direct-selling business was taking on great costs for a number of reasons including global legal restrictions and some dissatisfaction of the company’s representatives. Since Avon’s reliance is on its direct-selling, the earnings and representative s atisfaction are essential for the success of the business. The underlying factor along with other contributing causes was that Avon had grown faster than portions of its infrastructure and talent could support.The structure, people and processes that support a $5 billion company were not necessarily a good support for the $10 billion company. In the process of reviewing its talent practices the talent management team was able to identify six areas of missing or poorly functioning talent processes. The weaknesses that were found in Avon’s existing talent practices were listed 1) opaque; 2) egalitarian; 3) complex; 4) episodic; 5) emotional; and 6) meaningless. 2. Identify the model for change theory typified in the case study of your choice. Discuss what led you to identify the model that you did.Faced with the challenges of its flattening revenues and declining operating profits, Avon’s CEO restructured the organization and significant changes were made. As the changes began, it was found that Avon had some issues with its talent, both with the existing talent and with the company’s ability to identify and produce talent. The change model in the Avon case was the 360-degree feedback assessment tool. According to Silzer & Dowell (2010), the rise of 360-degree feedback assessments encouraged greater use of competency models built specifically around leadership behaviors.Silzer & Dowell (2010) go on to say that â€Å"organizations soon had lists of the leadership behaviors they expected from their managers,† which was the case with Avon. Avon was found to be opaque. As such, the talent practices within the organization were not known to the managers or to the associates. The resulting change was that of new practices and a re-making of the existing practices to become more transparent except for when there were confidentiality concerns. Another weakness that existed with Avon’s talent practices was that the company was egalitari an and needed to turn around the quality of its talent.Once this was understood, Avon made a change to differentiate its investment in its talent. This allowed for the company to better match the effectiveness of its talent investment with the expected return since before the turnaround the high performers were not engaged and the low performers were not managed very effectively. Avon’s level of complexity in its talent management practices was another noted weakness. Quality talent was not grown as quickly as was needed by the company so Avon simplified its talent process to ensure a balanced process. Employee surveys and talent reviews were performed episodically.Decisions concerning promotions and other objectives were more or less influenced by as much by individual knowledge and emotion as by objective facts. The turnaround that was made here was that relationships became stronger and as the business grew, leaders know of other’s performance or development needs a nd used this factor in determining talent management. Finally, meaningless talent practices such as Human Resources professionals not being able to answer most of the basic questions posed by managers about talent practices and there was not existing accountability.With the new talent practices, questions were answered and talent reviews were done and notations of progress were made. This was indication that effective communication had begun to take place. In this case, feedback was helpful and resulted in changed behaviors and overall things were done differently. According to Silzer & Dowell (2010) as leadership concepts and education gained greater currency, it became clear that the followers (subordinates) of leaders should share their views on their leader’s effectiveness.Greater use of this model encouraged greater use of competency models built around leadership behaviors. 3. Illustrate the types of evaluation information that were collected and how they are used to be nefit the company. As stated earlier, Avon faced challenges of flattening revenues and declining operating profits. Regarding this situation there were many contributing causes. One underlying issue was that Avon had grown faster than portions of its infrastructure and talent could support (Goldsmith, & Carter, 2010, p. 2).Avon’s structure, including people and processes, had grown from that of a company with $5 billion in revenues to that of a company with $10 billion in revenues. With this growth Avon’s structure was no longer a good fit and was in need of a turnaround. To begin the process of turning the business around, the talent management group (TM) started by requesting copies of the 360-degree assessment of each VP, not to take any action against anyone, but to gain more knowledge about the behavioral information of the top leaders. Every enterprise must build knowledge into its value proposition. Knowledge cannot be separated but needs to be an explicit part of everything about an enterprise† (Edersheim, 2007, p. 189). The 360-degree feedback is a performance measurement which involves rating individuals on work-related behaviors. According to Noe, Hollenbeck, Gerhart, & Wright (2011), there are benefits of the 360-degree feedback. Organizations collect multiple perspectives of managers’ performance, allowing employees to compare their own personal evaluations with the views of others.The request for the 360-degree assessment was denied citing confidentiality. This matter was addressed and a new and simpler 360-degree assessment process was designed and implemented which now allowed for the disclosure of behavioral information to be used when making decisions relative to promotions and assignments. The new process aided in making the talent process less complex and more transparent The performance management form within Avon was a ten page long form and many of the associates had not had a review in a number of years.With t he turnaround process, Associates were now aligned with a different set of goals and could expect fair rewards. More value was added to the process because now managers had a simpler tool to use and it allowed them to manage their teams more effectively. Prior to the turnaround at Avon, accountability for talent practices was non-existent. With the implementation of the new process, it was believed that the focus could be on people issues and that mangers could be held accountable for the improvement thereof.Further, Associates were empowered to hold the managers accountable and to inform Human Resource leaders if things were not happening. The issues concerning accountability were applied such that talent management was the responsibility of the leaders within the organization and in keeping with Avon’s culture. 4. Speculate about success of the changes within the next five (5) years and how adjustments could be made if the results become less than ideal. The talent practice s at Avon had some weaknesses which were addressed during a year to a year and a half turnaround period.This process after turnaround saw great effective improvements in the talent practices. The most noticeable changes were in the areas of clear goal setting, feedback, development planning, and people effectiveness. As the talent management process has become simpler and more transparent at Avon, the development of leaders is now on a faster track. The work experience is also improved having made leaders more accountable for their behaviors. The effectiveness of the new process has contributed to Avon’s goals of reducing expenses and increasing revenue.It is expected that Avon will continue to grow and with the updated talent practices, better leaders will be developed faster and those leaders must continue with the development and growth of potential new leaders. This process is expected to continue and Avon will stay on track in achieving its goals for continued success. I t is important that the vision of the organization is continuously communicated. According to Bates (n. d. ) many leaders fail to get their messages across even though they are intelligent, analytical, and decisive leaders.Human resource professionals know that the consequences are serious if leaders cannot successfully communicate a vision. Executives have to motivate and inspire, or they will fail. One role of HR professionals is to recognize when there is an issue and help leaders develop this skill. References Bates, S. (n. d. ). Communicating vision: How HR professionals can help leaders articulate big ideas and get people moving in one direction. Retrieved 11/03/12 from http://www. hrcrossing. com/article/270140 Edersheim, E.H. (2007). The definitive Drucker. New York, NY: McGraw-Hill Goldsmith, M. , & Carter, Louis. (2010). Best practices in talent management: how the world’s leading corporations manage, develop, and retain top talent. San Francisco, CA: Pfeiffer Noe, R. A. , Hollenbeck, J. R. , Gerhart, B. , & Wright, P. M. (2011). Fundamentals of human resource management. New York, NY: McGraw-Hill Silzer, R. , & Dowell, B. E. (2010). Strategy-driven talent management: A leadership imperative. San Francisco, CA: Jossey-Bass.

Ethics Essay ­ “Miss Evers’ Boys” Essay

One unethical issue that I saw in the film was that the patients that were being used for the experiment were misinformed about their medical status. This is unethical because the patients were being lied to about something that ultimately affected their health and lives. A second unethical issue that I saw in the movie was that the medical service was not provided so as best to promote the participant’s interest. The service was provide to conduct an experiment to study the progression of Syphilis in African American men. The only reason patients consented to the experiment was because they thought that they had â€Å"bad blood† and were waiting for treatment. Thirdly, another unethical issue that I saw in the movie was that the information about the Syphilis treatment, Penicillin, was kept hidden from the patients. Even when Penicillin became the official treatment, patients never received it whether they were looking for it or not. The fourth unethical issue that I saw in the movie was that the nurse did not recognize that considerations relating to the well ­being of the individual participants in the research should have taken precedence of any interest of the doctors, science, or society.

Global Warming Effects on Population Literature review

Global Warming Effects on Population - Literature review Example DeWeerdt gives a interesting international, almost post-colonial study of the effects of global warming. She divides her analysis into three distinct categories of areas that will impact human life the most signifincantly that are also the most vulnerable to climate change: Food, Disease and changes in sea level. It is somewhat superficially obvious that all of these things will be impacted by climate; sea levels will obviously rise as polar ice melts (DeWeerdt 2012), food production, which is obviously very climate dependent will have to shift, and disease will obviously increase with any serious shifts in population, as these often lead to unsanitary conditions.Her analysis is novel, however, in indicating the degree to which these changes will impact different parts of the world. DeWeerdt argues that development, which is usually meant in economic ways, can actually be considered a society’s degree of ability to adapt to climate changes (DeWeerdt, 2012). Places that have hi gh degrees of infrastructure, for instance, will be more able to cope with any of these changes. Firstly, they can establish things that directly mitigate the effects of climate: things like greenhouses to cope with cold weather or irrigation to cope with dryness. Secondly, they can use transportation infrastructure to cope with secondary effects of climate change as well, easily developing newly productive areas while abandoning areas that have become unproductive due to climate change. Undeveloped countries, however, will suffer more greatly: people cannot flee a newly formed desert, cannot build seawalls to hold back a rising tide, and so forth. Conceptualizing development as a society’s degree of ability to cope with climate allows one to see very clearly that the effects of climate change will disproportionately affect developing societies.Using DeWeerdt’s analysis indicates that one can actually conceptualize as climate change (including both its causes and effec ts) as a particular damaging form of pseudocolonial resource extraction. If one imagines a favorable climate as a resource, in that it provides a whole sleuth of production but can be depleted if not properly managed, then developed countries, which achieve their development largely through the highest use of fossil fuels (DeWeerdt 2012) are actually taking a valuable resource from developing countries every time they use fossil fuels. This is an especially damaging form of resource consumption because the people who get the resource depleted do not gain from the depletion, unlike mining, for instance, in which minerals can be sold. Climate can more be imagined as a river dammed outside of a country’s borders: they lose, and do not gain, while the other country gains but does not loose.

MANAGEMENT ACCOUNTING Essay Example | Topics and Well Written Essays - 2000 words - 2

MANAGEMENT ACCOUNTING - Essay Example â€Å"Costing is a powerful tool that helps managers to discover the true costs of products† (Ledgerwood & White 2008, p.317). Relying on the given case of the building firm â€Å"Home improvements and Extensions Ltd† we can consider different constructive costs. A firm may incur various types of operational costs during the course of their day to day business operations. In order to design situational business strategies and thereby fit the organizational interests into the persisting demand conditions, every corner of operational costs has to be essentially identified by a firm. According to Truett & Truett (2008), a firm may experience different types of costs such as ‘historical costs, opportunity costs, fixed costs, variable costs, incremental costs, private costs, and social costs’ (p.223). The explanations and reasons of variations of these costs are described below with suitable examples from construction industry. Historical cost: It is the cost incurred by the firm during the course of their operations for which the firm has either paid completely in the past or promised to pay in future. Historical cost is recorded in the books of accounts of the firm and is used for the preparation of income statement at the end of the financial period. It includes rent and interest payments, wages and salaries to the labors, and cost of raw materials. Historical cost may vary according to the market demand for the services of the construction firm. Opportunity costs: It is the implicit cost incurred by the firm if the business had employed to next best opportunity. Although opportunity cost is an imaginary cost, it is necessary for calculating the net income from operations. Implicit salary income to the directors and implicit interest income on capital are some of the examples for opportunity costs. It may fluctuate according to management strategies and

Discuss the role of Haskole (Jewish Enlightenment in Eastern Europe) Essay

Discuss the role of Haskole (Jewish Enlightenment in Eastern Europe) and of Hasidism in the formation and development of modern Yiddish culture - Essay Example From the initial days of the Yiddish, there were a few prayer books for women but were merely translations of existing Hebrew scripts. The first of the books published originally in Yiddish was â€Å"Come Out and See†, more commonly known by a slurring of the name as ‘Tsenerena’; written in the early 1600s. It was written for women, who generally did not read Hebrew and were not as well-versed in biblical commentary, so it is an easier read than some of the Hebrew commentaries written for men. When secular Jewish fiction began to emerge, the religious authorities of that time did not approve of these irreverent Yiddish writings dealing with modern secular and frivolous themes. Some strictly observant people refused to even set type for these writers because they were so offended by their works, but Jewish people throughout Europe embraced them wholeheartedly and treasured them. Foundations of the Yiddish theatre can be traced back to Abraham Haim Lipke Goldfaden w ho was (at that time) a pioneer of Yiddish plays. He wrote and produced numerous plays while travelling throughout the Eastern Europe. The culture, as we all know, was not preserved and was laid bare to the brutalism of the Nazis and the communists†¦ however; certain movements took place during that period in order to modernize the Jews and the Yiddish culture. Two off these movements are of great significance regarding the enlightenment of the Jews in Eastern Europe i.e. Haskole (Haskalah) and Hasidism (Hasidic Judaism). Haskole is a word of Yiddish, the alternative of which in Hebrew is Haskalah. The movement began in the late 1880s its aim being to enlighten the Jews of Eastern Europe through better mixing of the Jews with the eastern societies so that they could learn and live the ways of their neighbors in order to acquire a wider horizon. Haskole encouraged

The benefits and necessity of bilingual education in schools and Essay

The benefits and necessity of bilingual education in schools and universities - Essay Example The debate on bilingual education has been something that has been ongoing since the 1960s when the equality of different Minority groups in the United States was first asserted.This is because there were many American citizens who could not speak fluent English,but were quite fluent in other languages such as Spanish and FrenchToday there are many schools that champion the case of bilingual education. In some of such schools, the students are in English as well as their native languages. Comprehending the disadvantages as well as advantages of a bilingual education can help people to develop an impartial perspective on the necessity of giving the educational system a bilingual structure. While many people may think bilingual education is ineffective and unnecessary, I have discovered it to be otherwise. Why Bilingual Education is Unnecessary. There are many people who feel that the bilingual system of education will not be beneficial for English speakers in the long run. For example , in English speaking nations, there are people who feel that adopting a bilingual system will compromise the students’ grasp of the English language. To others, it appears that adopting a bilingual system is giving too much power to foreign elements within their own nation (Sizeron). To them, the effort to become a part of the society lies with the foreigners who relocate to English speaking nations. The citizens of an English speaking nation should not be inconvenienced in different ways in order to make foreigners feel that they are accepted by society. Another reason why some people may be against bilingual education in their schools is because they do not wish to be forced to start learning new languages. Bilingual education systems inevitably result in a larger fraction of citizens who speak two languages clearly. From there, it is quite easy for the reigning government to make the decision that it will have an additional national language. This means that it is not jus t the students who will be compelled to learn the new language. The English speaking citizens of the nation will also be compelled to learn the new language in order to be able to navigate around public systems that may start operating in two different languages. Not all people are talented and able to learn other languages easily. For many people, learning a new language is quite a difficult undertaking. Another reason why some people oppose the structuring of a bilingual educational system is that it might cost more than can be dedicated to expanding a nation’s educational system. Using different languages in different nations, states, or cities in the same region means that there will have to be different facilities to cater for the speakers of all languages. This will also further strain the educational systems of most countries which do not even fully cover the existing programs (National Latino Children’s Institute). Opponents of the bilingual educational system also often stress that children may be confused if they are suddenly required to start speaking different languages and master all of them. From their point of view, subjects such as English can only be mastered through the ‘immersion’ method. This means that the student has to have the opportunity to speak this language in all settings in order to become accustomed to it. Introducing a new language means that students will have to learn two languages concurrently. This could mean that none of the languages become well mastered. Why Bilingual Education is Necessary. According to the proponents of bilingual education, there are many reasons why bilingual educational systems are beneficial both for students as well as the general society. In addition to allowing children to develop a feeling for cultural pluralism, there are academic as well as cultural benefits associated with speaking two languages right from childhood. According to a research documented by

HR Training Class Research Paper Example | Topics and Well Written Essays - 1750 words

HR Training Class - Research Paper Example It employs more employees some of whom do not understand how maintain competition in the market. In relation of remaining relevant and competitive, our retail department in this company trains its employees in order to conform to this requirement. This approach benefits our employees by providing a clearly defined working culture in which employees have the necessary job satisfaction that enhances their productivity. The main concern of this training is to find out ways in which Al-Futtaim will expand into global retail through improved customers and employees relation. This will depend on the strategies and policies applied to the employees in relation to customer satisfaction. The policies and strategies will influence the employees. In relation to internationalization, a challenge on maintaining integrity, service and social responsibility is bound to arise Hipsher (2006). The nature of the training will be descriptive. The description will focus on explaining the key issues of how and when to strategize the policies of the training. The information presented in the training will be on the strategies and policies availed from the existing and potential customers of this retail company. The information will assist in developing a roadmap to global retail expansion with special consideration on the strategies that the employees will be equipped. The needs assessment for Al-Futtaim Retail Company gives the facilitators an opportunity to prepare adequate content to deliver to their expected audience. Inadequate number of employees is considered as a prerequisite to the training. The existence of web content with information containing this training may be reproduced for this training. The achievement of an organization depends on a number of factors such as the internal organization and coordination between the management and their sub-ordinate staff. Needs assessment helps an organization

How to Develop an Evangelistic Lifestyle Essay Example | Topics and Well Written Essays - 2250 words

How to Develop an Evangelistic Lifestyle - Essay Example In this book, the author reveals how one may keep the belief that Gospel can clearly and effectively communicate the message without any legalism that turns grace confusing. Moreover, with the help of this book, one can learn more about his ultimate faith in Christ as a Savior. As it is believed and gospel message that all the believers in Christ are the representatives and the ambassadors for Jesus having responsibility to declare His message to the entire world, in fact the fallen word. Apostle Paul is an enormous example of a loving and bold representative for Jesus. He came forward with the message of cooperation and reconciliation to all those who come in contact with him. He was the person who really did faithful evangelism throughout his life. Conversely, most of us cannot prove to be Apostle Paul. For many believers of Jesus Christ, the job of evangelism is something this may turn them into guilt and fear. For most of the believers, it is not a joyful experience than what it should be. Although many believers put in great efforts to develop evangelistic lifestyle and declaring the magnificent gospel’s messages but at the same time some fail to understand the urgency of this spiritual nourishment. So, Dr. Moyer points out what can be done to develop more faithful evangelistic efforts. In the first chapter, How to Develop an Evangelistic Lifestyle, Dr. ... Christ commands to his representatives â€Å"Go into the world and preach the gospel to every creature.† This verse simply tells that evangelism requires true dedication and the obedience of God. On the other hand, according to Dr. Moyer, some people excuse while spreading the message of gospel to their surroundings which is absolutely contradictory in achieving evangelistic lifestyle. It is also presented in the book that those individuals who present themselves entirely to evangelism also offer themselves to prayers. They also request God to offer them with courage to speak about Gospel. This gave them strength to understand that God can do this or He will surely do that. Personal contact is also essential in developing evangelism. We may have contact with non- Christians in a get together or in an informal discussion regarding spiritual things. Moreover, we should not let fright and panic set in our way. With the help of this initial information put forward by Dr. Moyer, it is obvious that evangelistic lifestyle always deals with true commitment and obedience. It also demands sincere efforts and involvements rather than just developing a mere intention. It is a great way to bring lost individuals to the right path of God. Next to these initial details presented by Dr. Moyer, the second chapter of the book under title How to Turn a Conversation to Spiritual Things is far more interesting. According to Dr. Moyer, talking about spiritual things is nothing difficult but only to those who are already interested. So, the question is that what to do with those non- Christians who are not interested in such topics. How we can bring up the subject of Spiritual things to discussion. How we can turn the topic of discussion from Golf to God, from secular to

Who Killed the Electric Car Essay Example for Free

Who Killed the Electric Car Essay In 1996, electric cars began to appear on roads all over California. They were quiet and fast, produced no exhaust and ran without gasoline.. Ten years later, these futuristic cars were almost completely gone. Who Killed the Electric Car is a documentary which unfolds a complex set of events around the development and demise of the modern electric car. The story stems from California from the early 1990s to 2006. Chris Paine, the film maker has woven together interviews and archival footage of over 65 people involved with the events. The narrative begins to unfold with a brief history of the first electric cars created in the early twentieth century. These electric vehicles were killed off nearly 100 years ago as gas/petroleum powered internal combustion engine (ICE) cars became cheaper. The worsening problems of gas/petrol cars are illustrated: smog, high child asthma rates, CO2 emissions and global warming. [Later we also see the use of the US Military in the Middle East. The loss of life and financial cost of war are not mentioned]. The film then commences the story of the modern EV in 1987 when General Motors and the SunRaycer, won the World Solar Challenge, a solar electric car race in Australia. General Motors CEO, Roger Smith challenged the same design team to build a prototype practical electric car which became known as the Impact when announced in 1990. The project expanded to small scale production vehicles with the aim that it would give GM several years lead over any competitor car companies. The Californian Air Resources Board (CARB) saw this as a way to solve their air quality problem and in 1990 passed the Zero Emissions Vehicle (ZEV) Mandate. The ZEV Mandate specified increasing numbers of vehicles sold would have to be Zero Emission Vehicles. For the car companies, there was only two options: Comply with the law or fight it. In then end, they would do both. The movie continues to reveal what the various suspects did to kill the reality of the electric car, and the efforts of EV supporters to save them. Oil companies stood to lose enormous profits if EV sales took off and they colluded with others to kill the electric car. To comply with the ZEV Mandate, in 1996, GM started leasing small numbers of the production car, called the EV1. Other car companies also produced electric vehicles by converting existing production models and leased them to drivers. But the GM board of directors never really wanted the car to succeed as they didnt think they would make profit from the car. They saw losses from development costs and the virtual absence of maintenance and replacement parts which, for gas cars, bring ongoing profits. They were worried that the popularity of the car was growing and that other US states were considering ZEV Mandate laws which meant that they may have to convert all their cars to electric drives which represented even bigger losses.

Decision Tree for Prognostic Classification

Decision Tree for Prognostic Classification Decision Tree for Prognostic Classification of Multivariate Survival Data and Competing Risks 1. Introduction Decision tree (DT) is one way to represent rules underlying data. It is the most popular tool for exploring complex data structures. Besides that it has become one of the most flexible, intuitive and powerful data analytic tools for determining distinct prognostic subgroups with similar outcome within each subgroup but different outcomes between the subgroups (i.e., prognostic grouping of patients). It is hierarchical, sequential classification structures that recursively partition the set of observations. Prognostic groups are important in assessing disease heterogeneity and for design and stratification of future clinical trials. Because patterns of medical treatment are changing so rapidly, it is important that the results of the present analysis be applicable to contemporary patients. Due to their mathematical simplicity, linear regression for continuous data, logistic regression for binary data, proportional hazard regression for censored survival data, marginal and frailty regression for multivariate survival data, and proportional subdistribution hazard regression for competing risks data are among the most commonly used statistical methods. These parametric and semiparametric regression methods, however, may not lead to faithful data descriptions when the underlying assumptions are not satisfied. Sometimes, model interpretation can be problematic in the presence of high-order interactions among predictors. DT has evolved to relax or remove the restrictive assumptions. In many cases, DT is used to explore data structures and to derive parsimonious models. DT is selected to analyze the data rather than the traditional regression analysis for several reasons. Discovery of interactions is difficult using traditional regression, because the interactions must be specified a priori. In contrast, DT automatically detects important interactions. Furthermore, unlike traditional regression analysis, DT is useful in uncovering variables that may be largely operative within a specific patient subgroup but may have minimal effect or none in other patient subgroups. Also, DT provides a superior means for prognostic classification. Rather than fitting a model to the data, DT sequentially divides the patient group into two subgroups based on prognostic factor values (e.g., tumor size The landmark work of DT in statistical community is the Classification and Regression Trees (CART) methodology of Breiman et al. (1984). A different approach was C4.5 proposed by Quinlan (1992). Original DT method was used in classification and regression for categorical and continuous response variable, respectively. In a clinical setting, however, the outcome of primary interest is often duration of survival, time to event, or some other incomplete (that is, censored) outcome. Therefore, several authors have developed extensions of original DT in the setting of censored survival data (Banerjee Noone, 2008). In science and technology, interest often lies in studying processes which generate events repeatedly over time. Such processes are referred to as recurrent event processes and the data they provide are called recurrent event data which includes in multivariate survival data. Such data arise frequently in medical studies, where information is often available on many individuals, each of whom may experience transient clinical events repeatedly over a period of observation. Examples include the occurrence of asthma attacks in respirology trials, epileptic seizures in neurology studies, and fractures in osteoporosis studies. In business, examples include the filing of warranty claims on automobiles, or insurance claims for policy holders. Since multivariate survival times frequently arise when individuals under observation are naturally clustered or when each individual might experience multiple events, then further extensions of DT are developed for such kind of data. In some studies, patients may be simultaneously exposed to several events, each competing for their mortality or morbidity. For example, suppose that a group of patients diagnosed with heart disease is followed in order to observe a myocardial infarction (MI). If by the end of the study each patient was either observed to have MI or was alive and well, then the usual survival techniques can be applied. In real life, however, some patients may die from other causes before experiencing an MI. This is a competing risks situation because death from other causes prohibits the occurrence of MI. MI is considered the event of interest, while death from other causes is considered a competing risk. The group of patients dead of other causes cannot be considered censored, since their observations are not incomplete. The extension of DT can also be employed for competing risks survival time data. These extensions can make one apply the technique to clinical trial data to aid in the development of prognostic classifications for chronic diseases. This chapter will cover DT for multivariate and competing risks survival time data as well as their application in the development of medical prognosis. Two kinds of multivariate survival time regression model, i.e. marginal and frailty regression model, have their own DT extensions. Whereas, the extension of DT for competing risks has two types of tree. First, the â€Å"single event† DT is developed based on splitting function using one event only. Second, the â€Å"composite events† tree which use all the events jointly. 2. Decision Tree A DT is a tree-like structure used for classification, decision theory, clustering, and prediction functions. It depicts rules for dividing data into groups based on the regularities in the data. A DT can be used for categorical and continuous response variables. When the response variables are continuous, the DT is often referred to as a regression tree. If the response variables are categorical, it is called a classification tree. However, the same concepts apply to both types of trees. DTs are widely used in computer science for data structures, in medical sciences for diagnosis, in botany for classification, in psychology for decision theory, and in economic analysis for evaluating investment alternatives. DTs learn from data and generate models containing explicit rule-like relationships among the variables. DT algorithms begin with the entire set of data, split the data into two or more subsets by testing the value of a predictor variable, and then repeatedly split each subset into finer subsets until the split size reaches an appropriate level. The entire modeling process can be illustrated in a tree-like structure. A DT model consists of two parts: creating the tree and applying the tree to the data. To achieve this, DTs use several different algorithms. The most popular algorithm in the statistical community is Classification and Regression Trees (CART) (Breiman et al., 1984). This algorithm helps DTs gain credibility and acceptance in the statistics community. It creates binary splits on nominal or interval predictor variables for a nominal, ordinal, or interval response. The most widely-used algorithms by computer scientists are ID3, C4.5, and C5.0 (Quinlan, 1993). The first version of C4.5 and C5.0 were limited to categorical predictors; however, the most recent versions are similar to CART. Other algorithms include Chi-Square Automatic Interaction Detection (CHAID) for categorical response (Kass, 1980), CLS, AID, TREEDISC, Angoss KnowledgeSEEKER, CRUISE, GUIDE and QUEST (Loh, 2008). These algorithms use different approaches for splitting variables. CART, CRUISE, GUIDE and QUEST use the sta tistical approach, while CLS, ID3, and C4.5 use an approach in which the number of branches off an internal node is equal to the number of possible categories. Another common approach, used by AID, CHAID, and TREEDISC, is the one in which the number of nodes on an internal node varies from two to the maximum number of possible categories. Angoss KnowledgeSEEKER uses a combination of these approaches. Each algorithm employs different mathematical processes to determine how to group and rank variables. Let us illustrate the DT method in a simplified example of credit evaluation. Suppose a credit card issuer wants to develop a model that can be used for evaluating potential candidates based on its historical customer data. The companys main concern is the default of payment by a cardholder. Therefore, the model should be able to help the company classify a candidate as a possible defaulter or not. The database may contain millions of records and hundreds of fields. A fragment of such a database is shown in Table 1. The input variables include income, age, education, occupation, and many others, determined by some quantitative or qualitative methods. The model building process is illustrated in the tree structure in 1. The DT algorithm first selects a variable, income, to split the dataset into two subsets. This variable, and also the splitting value of $31,000, is selected by a splitting criterion of the algorithm. There exist many splitting criteria (Mingers, 1989). The basic principle of these criteria is that they all attempt to divide the data into clusters such that variations within each cluster are minimized and variations between the clusters are maximized. The follow- Name Age Income Education Occupation Default Andrew 42 45600 College Manager No Allison 26 29000 High School Self Owned Yes Sabrina 58 36800 High School Clerk No Andy 35 37300 College Engineer No †¦ Table 1. Partial records and fields of a database table for credit evaluation up splits are similar to the first one. The process continues until an appropriate tree size is reached. 1 shows a segment of the DT. Based on this tree model, a candidate with income at least $31,000 and at least college degree is unlikely to default the payment; but a self-employed candidate whose income is less than $31,000 and age is less than 28 is more likely to default. We begin with a discussion of the general structure of a popular DT algorithm in statistical community, i.e. CART model. A CART model describes the conditional distribution of y given X, where y is the response variable and X is a set of predictor variables (X = (X1,X2,†¦,Xp)). This model has two main components: a tree T with b terminal nodes, and a parameter Q = (q1,q2,†¦, qb) ÃÅ' Rk which associates the parameter values qm, with the mth terminal node. Thus a tree model is fully specified by the pair (T, Q). If X lies in the region corresponding to the mth terminal node then y|X has the distribution f(y|qm), where we use f to represent a conditional distribution indexed by qm. The model is called a regression tree or a classification tree according to whether the response y is quantitative or qualitative, respectively. 2.1 Splitting a tree The DT T subdivides the predictor variable space as follows. Each internal node has an associated splitting rule which uses a predictor to assign observations to either its left or right child node. The internal nodes are thus partitioned into two subsequent nodes using the splitting rule. For quantitative predictors, the splitting rule is based on a split rule c, and assigns observations for which {xi For a regression tree, conventional algorithm models the response in each region Rm as a constant qm. Thus the overall tree model can be expressed as (Hastie et al., 2001): (1) where Rm, m = 1, 2,†¦,b consist of a partition of the predictors space, and therefore representing the space of b terminal nodes. If we adopt the method of minimizing the sum of squares as our criterion to characterize the best split, it is easy to see that the best , is just the average of yi in region Rm: (2) where Nm is the number of observations falling in node m. The residual sum of squares is (3) which will serve as an impurity measure for regression trees. If the response is a factor taking outcomes 1,2, K, the impurity measure Qm(T), defined in (3) is not suitable. Instead, we represent a region Rm with Nm observations with (4) which is the proportion of class k(k ÃŽ {1, 2,†¦,K}) observations in node m. We classify the observations in node m to a class , the majority class in node m. Different measures Qm(T) of node impurity include the following (Hastie et al., 2001): Misclassification error: Gini index: Cross-entropy or deviance: (5) For binary outcomes, if p is the proportion of the second class, these three measures are 1 max(p, 1 p), 2p(1 p) and -p log p (1 p) log(1 p), respectively. All three definitions of impurity are concave, having minimums at p = 0 and p = 1 and a maximum at p = 0.5. Entropy and the Gini index are the most common, and generally give very similar results except when there are two response categories. 2.2 Pruning a tree To be consistent with conventional notations, lets define the impurity of a node h as I(h) ((3) for a regression tree, and any one in (5) for a classification tree). We then choose the split with maximal impurity reduction (6) where hL and hR are the left and right children nodes of h and p(h) is proportion of sample fall in node h. How large should we grow the tree then? Clearly a very large tree might overfit the data, while a small tree may not be able to capture the important structure. Tree size is a tuning parameter governing the models complexity, and the optimal tree size should be adaptively chosen from the data. One approach would be to continue the splitting procedures until the decrease on impurity due to the split exceeds some threshold. This strategy is too short-sighted, however, since a seeming worthless split might lead to a very good split below it. The preferred strategy is to grow a large tree T0, stopping the splitting process when some minimum number of observations in a terminal node (say 10) is reached. Then this large tree is pruned using pruning algorithm, such as cost-complexity or split complexity pruning algorithm. To prune large tree T0 by using cost-complexity algorithm, we define a subtree T T0 to be any tree that can be obtained by pruning T0, and define to be the set of terminal nodes of T. That is, collapsing any number of its terminal nodes. As before, we index terminal nodes by m, with node m representing region Rm. Let denotes the number of terminal nodes in T (= b). We use instead of b following the conventional notation and define the risk of trees and define cost of tree as Regression tree: , Classification tree: , (7) where r(h) measures the impurity of node h in a classification tree (can be any one in (5)). We define the cost complexity criterion (Breiman et al., 1984) (8) where a(> 0) is the complexity parameter. The idea is, for each a, find the subtree Ta T0 to minimize Ra(T). The tuning parameter a > 0 governs the tradeoff between tree size and its goodness of fit to the data (Hastie et al., 2001). Large values of a result in smaller tree Ta and conversely for smaller values of a. As the notation suggests, with a = 0 the solution is the full tree T0. To find Ta we use weakest link pruning: we successively collapse the internal node that produces the smallest per-node increase in R(T), and continue until we produce the single-node (root) tree. This gives a (finite) sequence of subtrees, and one can show this sequence must contains Ta. See Brieman et al. (1984) and Ripley (1996) for details. Estimation of a () is achieved by five- or ten-fold cross-validation. Our final tree is then denoted as . It follows that, in CART and related algorithms, classification and regression trees are produced from data in two stages. In the first stage, a large initial tree is produced by splitting one node at a time in an iterative, greedy fashion. In the second stage, a small subtree of the initial tree is selected, using the same data set. Whereas the splitting procedure proceeds in a top-down fashion, the second stage, known as pruning, proceeds from the bottom-up by successively removing nodes from the initial tree. Theorem 1 (Brieman et al., 1984, Section 3.3) For any value of the complexity parameter a, there is a unique smallest subtree of T0 that minimizes the cost-complexity. Theorem 2 (Zhang Singer, 1999, Section 4.2) If a2 > al, the optimal sub-tree corresponding to a2 is a subtree of the optimal subtree corresponding to al. More general, suppose we end up with m thresholds, 0 (9) where means that is a subtree of . These are called nested optimal subtrees. 3. Decision Tree for Censored Survival Data Survival analysis is the phrase used to describe the analysis of data that correspond to the time from a well-defined time origin until the occurrence of some particular events or end-points. It is important to state what the event is and when the period of observation starts and finish. In medical research, the time origin will often correspond to the recruitment of an individual into an experimental study, and the end-point is the death of the patient or the occurrence of some adverse events. Survival data are rarely normally distributed, but are skewed and comprise typically of many early events and relatively few late ones. It is these features of the data that necessitate the special method survival analysis. The specific difficulties relating to survival analysis arise largely from the fact that only some individuals have experienced the event and, subsequently, survival times will be unknown for a subset of the study group. This phenomenon is called censoring and it may arise in the following ways: (a) a patient has not (yet) experienced the relevant outcome, such as relapse or death, by the time the study has to end; (b) a patient is lost to follow-up during the study period; (c) a patient experiences a different event that makes further follow-up impossible. Generally, censoring times may vary from individual to individual. Such censored survival time underestimated the true (but unknown) time to event. Visualising the survival process of an individual as a time-line, the event (assuming it is to occur) is beyond the end of the follow-up period. This situation is often called right censoring. Most survival data include right censored observation. In many biomedical and reliability studies, interest focuses on relating the time to event to a set of covariates. Cox proportional hazard model (Cox, 1972) has been established as the major framework for analysis of such survival data over the past three decades. But, often in practices, one primary goal of survival analysis is to extract meaningful subgroups of patients determined by the prognostic factors such as patient characteristics that are related to the level of disease. Although proportional hazard model and its extensions are powerful in studying the association between covariates and survival times, usually they are problematic in prognostic classification. One approach for classification is to compute a risk score based on the estimated coefficients from regression methods (Machin et al., 2006). This approach, however, may be problematic for several reasons. First, the definition of risk groups is arbitrary. Secondly, the risk score depends on the correct specification of the model. It is difficult to check whether the model is correct when many covariates are involved. Thirdly, when there are many interaction terms and the model becomes complicated, the result becomes difficult to interpret for the purpose of prognostic classification. Finally, a more serious problem is that an invalid prognostic group may be produced if no patient is included in a covariate profile. In contrast, DT methods do not suffer from these problems. Owing to the development of fast computers, computer-intensive methods such as DT methods have become popular. Since these investigate the significance of all potential risk factors automatically and provide interpretable models, they offer distinct advantages to analysts. Recently a large amount of DT methods have been developed for the analysis of survival data, where the basic concepts for growing and pruning trees remain unchanged, but the choice of the splitting criterion has been modified to incorporate the censored survival data. The application of DT methods for survival data are described by a number of authors (Gordon Olshen, 1985; Ciampi et al., 1986; Segal, 1988; Davis Anderson, 1989; Therneau et al., 1990; LeBlanc Crowley, 1992; LeBlanc Crowley, 1993; Ahn Loh, 1994; Bacchetti Segal, 1995; Huang et al., 1998; KeleÃ…Å ¸ Segal, 2002; Jin et al., 2004; Cappelli Zhang, 2007; Cho Hong, 2008), including the text by Zhang Singer (1999). 4. Decision Tree for Multivariate Censored Survival Data Multivariate survival data frequently arise when we faced the complexity of studies involving multiple treatment centres, family members and measurements repeatedly made on the same individual. For example, in multi-centre clinical trials, the outcomes for groups of patients at several centres are examined. In some instances, patients in a centre might exhibit similar responses due to uniformity of surroundings and procedures within a centre. This would result in correlated outcomes at the level of the treatment centre. For the situation of studies of family members or litters, correlation in outcome is likely for genetic reasons. In this case, the outcomes would be correlated at the family or litter level. Finally, when one person or animal is measured repeatedly over time, correlation will most definitely exist in those responses. Within the context of correlated data, the observations which are correlated for a group of individuals (within a treatment centre or a family) or for on e individual (because of repeated sampling) are referred to as a cluster, so that from this point on, the responses within a cluster will be assumed to be correlated. Analysis of multivariate survival data is complex due to the presence of dependence among survival times and unknown marginal distributions. Multivariate survival times frequently arise when individuals under observation are naturally clustered or when each individual might experience multiple events. A successful treatment of correlated failure times was made by Clayton and Cuzik (1985) who modelled the dependence structure with a frailty term. Another approach is based on a proportional hazard formulation of the marginal hazard function, which has been studied by Wei et al. (1989) and Liang et al. (1993). Noticeably, Prentice et al. (1981) and Andersen Gill (1982) also suggested two alternative approaches to analyze multiple event times. Extension of tree techniques to multivariate censored data is motivated by the classification issue associated with multivariate survival data. For example, clinical investigators design studies to form prognostic rules. Credit risk analysts collect account information to build up credit scoring criteria. Frequently, in such studies the outcomes of ultimate interest are correlated times to event, such as relapses, late payments, or bankruptcies. Since DT methods recursively partition the predictor space, they are an alternative to conventional regression tools. This section is concerned with the generalization of DT models to multivariate survival data. In attempt to facilitate an extension of DT methods to multivariate survival data, more difficulties need to be circumvented. 4.1 Decision tree for multivariate survival data based on marginal model DT methods for multivariate survival data are not many. Almost all the multivariate DT methods have been based on between-node heterogeneity, with the exception of Molinaro et al. (2004) who proposed a general within-node homogeneity approach for both univariate and multivariate data. The multivariate methods proposed by Su Fan (2001, 2004) and Gao et al. (2004, 2006) concentrated on between-node heterogeneity and used the results of regression models. Specifically, for recurrent event data and clustered event data, Su Fan (2004) used likelihood-ratio tests while Gao et al. (2004) used robust Wald tests from a gamma frailty model to maximize the between-node heterogeneity. Su Fan (2001) and Fan et al. (2006) used a robust log-rank statistic while Gao et al. (2006) used a robust Wald test from the marginal failure-time model of Wei et al. (1989). The generalization of DT for multivariate survival data is developed by using goodness of split approach. DT by goodness of split is grown by maximizing a measure of between-node difference. Therefore, only internal nodes have associated two-sample statistics. The tree structure is different from CART because, for trees grown by minimizing within-node error, each node, either terminal or internal, has an associated impurity measure. This is why the CART pruning procedure is not directly applicable to such types of trees. However, the split-complexity pruning algorithm of LeBlanc Crowley (1993) has resulted in trees by goodness of split that has become well-developed tools. This modified tree technique not only provides a convenient way of handling survival data, but also enlarges the applied scope of DT methods in a more general sense. Especially for those situations where defining prediction error terms is relatively difficult, growing trees by a two-sample statistic, together with the split-complexity pruning, offers a feasible way of performing tree analysis. The DT procedure consists of three parts: a method to partition the data recursively into a large tree, a method to prune the large tree into a subtree sequence, and a method to determine the optimal tree size. In the multivariate survival trees, the between-node difference is measured by a robust Wald statistic, which is derived from a marginal approach to multivariate survival data that was developed by Wei et al. (1989). We used split-complexity pruning borrowed from LeBlanc Crowley (1993) and use test sample for determining the right tree size. 4.1.1 The splitting statistic We consider n independent subjects but each subject to have K potential types or number of failures. If there are an unequal number of failures within the subjects, then K is the maximum. We let Tik = min(Yik,Cik ) where Yik = time of the failure in the ith subject for the kth type of failure and Cik = potential censoring time of the ith subject for the kth type of failure with i = 1,†¦,n and k = 1,†¦,K. Then dik = I (Yik ≠¤ Cik) is the indicator for failure and the vector of covariates is denoted Zik = (Z1ik,†¦, Zpik)T. To partition the data, we consider the hazard model for the ith unit for the kth type of failure, using the distinguishable baseline hazard as described by Wei et al. (1989), namely where the indicator function I(Zik Parameter b is estimated by maximizing the partial likelihood. If the observations within the same unit are independent, the partial likelihood functions for b for the distinguishable baseline model (10) would be, (11) Since the observations within the same unit are not independent for multivariate failure time, we refer to the above functions as the pseudo-partial likelihood. The estimator can be obtained by maximizing the likelihood by solving . Wei et al. (1989) showed that is normally distributed with mean 0. However the usual estimate, a-1(b), for the variance of , where (12) is not valid. We refer to a-1(b) as the naà ¯ve estimator. Wei et al. (1989) showed that the correct estimated (robust) variance estimator of is (13) where b(b) is weight and d(b) is often referred to as the robust or sandwich variance estimator. Hence, the robust Wald statistic corresponding to the null hypothesis H0 : b = 0 is (14) 4.1.2 Tree growing To grow a tree, the robust Wald statistic is evaluated for every possible binary split of the predictor space Z. The split, s, could be of several forms: splits on a single covariate, splits on linear combinations of predictors, and boolean combination of splits. The simplest form of split relates to only one covariate, where the split depends on the type of covariate whether it is ordered or nominal covariate. The â€Å"best split† is defined to be the one corresponding to the maximum robust Wald statistic. Subsequently the data are divided into two groups according to the best split. Apply this splitting scheme recursively to the learning sample until the predictor space is partitioned into many regions. There will be no further partition to a node when any of the following occurs: The node contains less than, say 10 or 20, subjects, if the overall sample size is large enough to permit this. We suggest using a larger minimum node size than used in CART where the default value is 5; All the observed times in the subset are censored, which results in unavailability of the robust Wald statistic for any split; All the subjects have identical covariate vectors. Or the node has only complete observations with identical survival times. In these situations, the node is considered as pure. The whole procedure results in a large tree, which could be used for the purpose of data structure exploration. 4.1.3 Tree pruning Let T denote either a particular tree or the set of all its nodes. Let S and denote the set of internal nodes and terminal nodes of T, respectively. Therefore, . Also let |Ãâ€"| denote the number of nodes. Let G(h) represent the maximum robust Wald statistic on a particular (internal) node h. In order to measure the performance of a tree, a split-complexity measure Ga(T) is introduced as in LeBlanc and Crowley (1993). That is, (15) where the number of internal nodes, |S|, measures complexity; G(T) measures goodness of split in T; and the complexity parameter a acts as a penalty for each additional split. Start with the large tree T0 obtained from the splitting procedure. For any internal node h of T0, i.e. h ÃŽ S0, a function g(h) is defined as (16) where Th denotes the branch with h as its root and Sh is the set of all internal nodes of Th. Then the weakest link in T0 is the node such that   < Decision Tree for Prognostic Classification Decision Tree for Prognostic Classification Decision Tree for Prognostic Classification of Multivariate Survival Data and Competing Risks 1. Introduction Decision tree (DT) is one way to represent rules underlying data. It is the most popular tool for exploring complex data structures. Besides that it has become one of the most flexible, intuitive and powerful data analytic tools for determining distinct prognostic subgroups with similar outcome within each subgroup but different outcomes between the subgroups (i.e., prognostic grouping of patients). It is hierarchical, sequential classification structures that recursively partition the set of observations. Prognostic groups are important in assessing disease heterogeneity and for design and stratification of future clinical trials. Because patterns of medical treatment are changing so rapidly, it is important that the results of the present analysis be applicable to contemporary patients. Due to their mathematical simplicity, linear regression for continuous data, logistic regression for binary data, proportional hazard regression for censored survival data, marginal and frailty regression for multivariate survival data, and proportional subdistribution hazard regression for competing risks data are among the most commonly used statistical methods. These parametric and semiparametric regression methods, however, may not lead to faithful data descriptions when the underlying assumptions are not satisfied. Sometimes, model interpretation can be problematic in the presence of high-order interactions among predictors. DT has evolved to relax or remove the restrictive assumptions. In many cases, DT is used to explore data structures and to derive parsimonious models. DT is selected to analyze the data rather than the traditional regression analysis for several reasons. Discovery of interactions is difficult using traditional regression, because the interactions must be specified a priori. In contrast, DT automatically detects important interactions. Furthermore, unlike traditional regression analysis, DT is useful in uncovering variables that may be largely operative within a specific patient subgroup but may have minimal effect or none in other patient subgroups. Also, DT provides a superior means for prognostic classification. Rather than fitting a model to the data, DT sequentially divides the patient group into two subgroups based on prognostic factor values (e.g., tumor size The landmark work of DT in statistical community is the Classification and Regression Trees (CART) methodology of Breiman et al. (1984). A different approach was C4.5 proposed by Quinlan (1992). Original DT method was used in classification and regression for categorical and continuous response variable, respectively. In a clinical setting, however, the outcome of primary interest is often duration of survival, time to event, or some other incomplete (that is, censored) outcome. Therefore, several authors have developed extensions of original DT in the setting of censored survival data (Banerjee Noone, 2008). In science and technology, interest often lies in studying processes which generate events repeatedly over time. Such processes are referred to as recurrent event processes and the data they provide are called recurrent event data which includes in multivariate survival data. Such data arise frequently in medical studies, where information is often available on many individuals, each of whom may experience transient clinical events repeatedly over a period of observation. Examples include the occurrence of asthma attacks in respirology trials, epileptic seizures in neurology studies, and fractures in osteoporosis studies. In business, examples include the filing of warranty claims on automobiles, or insurance claims for policy holders. Since multivariate survival times frequently arise when individuals under observation are naturally clustered or when each individual might experience multiple events, then further extensions of DT are developed for such kind of data. In some studies, patients may be simultaneously exposed to several events, each competing for their mortality or morbidity. For example, suppose that a group of patients diagnosed with heart disease is followed in order to observe a myocardial infarction (MI). If by the end of the study each patient was either observed to have MI or was alive and well, then the usual survival techniques can be applied. In real life, however, some patients may die from other causes before experiencing an MI. This is a competing risks situation because death from other causes prohibits the occurrence of MI. MI is considered the event of interest, while death from other causes is considered a competing risk. The group of patients dead of other causes cannot be considered censored, since their observations are not incomplete. The extension of DT can also be employed for competing risks survival time data. These extensions can make one apply the technique to clinical trial data to aid in the development of prognostic classifications for chronic diseases. This chapter will cover DT for multivariate and competing risks survival time data as well as their application in the development of medical prognosis. Two kinds of multivariate survival time regression model, i.e. marginal and frailty regression model, have their own DT extensions. Whereas, the extension of DT for competing risks has two types of tree. First, the â€Å"single event† DT is developed based on splitting function using one event only. Second, the â€Å"composite events† tree which use all the events jointly. 2. Decision Tree A DT is a tree-like structure used for classification, decision theory, clustering, and prediction functions. It depicts rules for dividing data into groups based on the regularities in the data. A DT can be used for categorical and continuous response variables. When the response variables are continuous, the DT is often referred to as a regression tree. If the response variables are categorical, it is called a classification tree. However, the same concepts apply to both types of trees. DTs are widely used in computer science for data structures, in medical sciences for diagnosis, in botany for classification, in psychology for decision theory, and in economic analysis for evaluating investment alternatives. DTs learn from data and generate models containing explicit rule-like relationships among the variables. DT algorithms begin with the entire set of data, split the data into two or more subsets by testing the value of a predictor variable, and then repeatedly split each subset into finer subsets until the split size reaches an appropriate level. The entire modeling process can be illustrated in a tree-like structure. A DT model consists of two parts: creating the tree and applying the tree to the data. To achieve this, DTs use several different algorithms. The most popular algorithm in the statistical community is Classification and Regression Trees (CART) (Breiman et al., 1984). This algorithm helps DTs gain credibility and acceptance in the statistics community. It creates binary splits on nominal or interval predictor variables for a nominal, ordinal, or interval response. The most widely-used algorithms by computer scientists are ID3, C4.5, and C5.0 (Quinlan, 1993). The first version of C4.5 and C5.0 were limited to categorical predictors; however, the most recent versions are similar to CART. Other algorithms include Chi-Square Automatic Interaction Detection (CHAID) for categorical response (Kass, 1980), CLS, AID, TREEDISC, Angoss KnowledgeSEEKER, CRUISE, GUIDE and QUEST (Loh, 2008). These algorithms use different approaches for splitting variables. CART, CRUISE, GUIDE and QUEST use the sta tistical approach, while CLS, ID3, and C4.5 use an approach in which the number of branches off an internal node is equal to the number of possible categories. Another common approach, used by AID, CHAID, and TREEDISC, is the one in which the number of nodes on an internal node varies from two to the maximum number of possible categories. Angoss KnowledgeSEEKER uses a combination of these approaches. Each algorithm employs different mathematical processes to determine how to group and rank variables. Let us illustrate the DT method in a simplified example of credit evaluation. Suppose a credit card issuer wants to develop a model that can be used for evaluating potential candidates based on its historical customer data. The companys main concern is the default of payment by a cardholder. Therefore, the model should be able to help the company classify a candidate as a possible defaulter or not. The database may contain millions of records and hundreds of fields. A fragment of such a database is shown in Table 1. The input variables include income, age, education, occupation, and many others, determined by some quantitative or qualitative methods. The model building process is illustrated in the tree structure in 1. The DT algorithm first selects a variable, income, to split the dataset into two subsets. This variable, and also the splitting value of $31,000, is selected by a splitting criterion of the algorithm. There exist many splitting criteria (Mingers, 1989). The basic principle of these criteria is that they all attempt to divide the data into clusters such that variations within each cluster are minimized and variations between the clusters are maximized. The follow- Name Age Income Education Occupation Default Andrew 42 45600 College Manager No Allison 26 29000 High School Self Owned Yes Sabrina 58 36800 High School Clerk No Andy 35 37300 College Engineer No †¦ Table 1. Partial records and fields of a database table for credit evaluation up splits are similar to the first one. The process continues until an appropriate tree size is reached. 1 shows a segment of the DT. Based on this tree model, a candidate with income at least $31,000 and at least college degree is unlikely to default the payment; but a self-employed candidate whose income is less than $31,000 and age is less than 28 is more likely to default. We begin with a discussion of the general structure of a popular DT algorithm in statistical community, i.e. CART model. A CART model describes the conditional distribution of y given X, where y is the response variable and X is a set of predictor variables (X = (X1,X2,†¦,Xp)). This model has two main components: a tree T with b terminal nodes, and a parameter Q = (q1,q2,†¦, qb) ÃÅ' Rk which associates the parameter values qm, with the mth terminal node. Thus a tree model is fully specified by the pair (T, Q). If X lies in the region corresponding to the mth terminal node then y|X has the distribution f(y|qm), where we use f to represent a conditional distribution indexed by qm. The model is called a regression tree or a classification tree according to whether the response y is quantitative or qualitative, respectively. 2.1 Splitting a tree The DT T subdivides the predictor variable space as follows. Each internal node has an associated splitting rule which uses a predictor to assign observations to either its left or right child node. The internal nodes are thus partitioned into two subsequent nodes using the splitting rule. For quantitative predictors, the splitting rule is based on a split rule c, and assigns observations for which {xi For a regression tree, conventional algorithm models the response in each region Rm as a constant qm. Thus the overall tree model can be expressed as (Hastie et al., 2001): (1) where Rm, m = 1, 2,†¦,b consist of a partition of the predictors space, and therefore representing the space of b terminal nodes. If we adopt the method of minimizing the sum of squares as our criterion to characterize the best split, it is easy to see that the best , is just the average of yi in region Rm: (2) where Nm is the number of observations falling in node m. The residual sum of squares is (3) which will serve as an impurity measure for regression trees. If the response is a factor taking outcomes 1,2, K, the impurity measure Qm(T), defined in (3) is not suitable. Instead, we represent a region Rm with Nm observations with (4) which is the proportion of class k(k ÃŽ {1, 2,†¦,K}) observations in node m. We classify the observations in node m to a class , the majority class in node m. Different measures Qm(T) of node impurity include the following (Hastie et al., 2001): Misclassification error: Gini index: Cross-entropy or deviance: (5) For binary outcomes, if p is the proportion of the second class, these three measures are 1 max(p, 1 p), 2p(1 p) and -p log p (1 p) log(1 p), respectively. All three definitions of impurity are concave, having minimums at p = 0 and p = 1 and a maximum at p = 0.5. Entropy and the Gini index are the most common, and generally give very similar results except when there are two response categories. 2.2 Pruning a tree To be consistent with conventional notations, lets define the impurity of a node h as I(h) ((3) for a regression tree, and any one in (5) for a classification tree). We then choose the split with maximal impurity reduction (6) where hL and hR are the left and right children nodes of h and p(h) is proportion of sample fall in node h. How large should we grow the tree then? Clearly a very large tree might overfit the data, while a small tree may not be able to capture the important structure. Tree size is a tuning parameter governing the models complexity, and the optimal tree size should be adaptively chosen from the data. One approach would be to continue the splitting procedures until the decrease on impurity due to the split exceeds some threshold. This strategy is too short-sighted, however, since a seeming worthless split might lead to a very good split below it. The preferred strategy is to grow a large tree T0, stopping the splitting process when some minimum number of observations in a terminal node (say 10) is reached. Then this large tree is pruned using pruning algorithm, such as cost-complexity or split complexity pruning algorithm. To prune large tree T0 by using cost-complexity algorithm, we define a subtree T T0 to be any tree that can be obtained by pruning T0, and define to be the set of terminal nodes of T. That is, collapsing any number of its terminal nodes. As before, we index terminal nodes by m, with node m representing region Rm. Let denotes the number of terminal nodes in T (= b). We use instead of b following the conventional notation and define the risk of trees and define cost of tree as Regression tree: , Classification tree: , (7) where r(h) measures the impurity of node h in a classification tree (can be any one in (5)). We define the cost complexity criterion (Breiman et al., 1984) (8) where a(> 0) is the complexity parameter. The idea is, for each a, find the subtree Ta T0 to minimize Ra(T). The tuning parameter a > 0 governs the tradeoff between tree size and its goodness of fit to the data (Hastie et al., 2001). Large values of a result in smaller tree Ta and conversely for smaller values of a. As the notation suggests, with a = 0 the solution is the full tree T0. To find Ta we use weakest link pruning: we successively collapse the internal node that produces the smallest per-node increase in R(T), and continue until we produce the single-node (root) tree. This gives a (finite) sequence of subtrees, and one can show this sequence must contains Ta. See Brieman et al. (1984) and Ripley (1996) for details. Estimation of a () is achieved by five- or ten-fold cross-validation. Our final tree is then denoted as . It follows that, in CART and related algorithms, classification and regression trees are produced from data in two stages. In the first stage, a large initial tree is produced by splitting one node at a time in an iterative, greedy fashion. In the second stage, a small subtree of the initial tree is selected, using the same data set. Whereas the splitting procedure proceeds in a top-down fashion, the second stage, known as pruning, proceeds from the bottom-up by successively removing nodes from the initial tree. Theorem 1 (Brieman et al., 1984, Section 3.3) For any value of the complexity parameter a, there is a unique smallest subtree of T0 that minimizes the cost-complexity. Theorem 2 (Zhang Singer, 1999, Section 4.2) If a2 > al, the optimal sub-tree corresponding to a2 is a subtree of the optimal subtree corresponding to al. More general, suppose we end up with m thresholds, 0 (9) where means that is a subtree of . These are called nested optimal subtrees. 3. Decision Tree for Censored Survival Data Survival analysis is the phrase used to describe the analysis of data that correspond to the time from a well-defined time origin until the occurrence of some particular events or end-points. It is important to state what the event is and when the period of observation starts and finish. In medical research, the time origin will often correspond to the recruitment of an individual into an experimental study, and the end-point is the death of the patient or the occurrence of some adverse events. Survival data are rarely normally distributed, but are skewed and comprise typically of many early events and relatively few late ones. It is these features of the data that necessitate the special method survival analysis. The specific difficulties relating to survival analysis arise largely from the fact that only some individuals have experienced the event and, subsequently, survival times will be unknown for a subset of the study group. This phenomenon is called censoring and it may arise in the following ways: (a) a patient has not (yet) experienced the relevant outcome, such as relapse or death, by the time the study has to end; (b) a patient is lost to follow-up during the study period; (c) a patient experiences a different event that makes further follow-up impossible. Generally, censoring times may vary from individual to individual. Such censored survival time underestimated the true (but unknown) time to event. Visualising the survival process of an individual as a time-line, the event (assuming it is to occur) is beyond the end of the follow-up period. This situation is often called right censoring. Most survival data include right censored observation. In many biomedical and reliability studies, interest focuses on relating the time to event to a set of covariates. Cox proportional hazard model (Cox, 1972) has been established as the major framework for analysis of such survival data over the past three decades. But, often in practices, one primary goal of survival analysis is to extract meaningful subgroups of patients determined by the prognostic factors such as patient characteristics that are related to the level of disease. Although proportional hazard model and its extensions are powerful in studying the association between covariates and survival times, usually they are problematic in prognostic classification. One approach for classification is to compute a risk score based on the estimated coefficients from regression methods (Machin et al., 2006). This approach, however, may be problematic for several reasons. First, the definition of risk groups is arbitrary. Secondly, the risk score depends on the correct specification of the model. It is difficult to check whether the model is correct when many covariates are involved. Thirdly, when there are many interaction terms and the model becomes complicated, the result becomes difficult to interpret for the purpose of prognostic classification. Finally, a more serious problem is that an invalid prognostic group may be produced if no patient is included in a covariate profile. In contrast, DT methods do not suffer from these problems. Owing to the development of fast computers, computer-intensive methods such as DT methods have become popular. Since these investigate the significance of all potential risk factors automatically and provide interpretable models, they offer distinct advantages to analysts. Recently a large amount of DT methods have been developed for the analysis of survival data, where the basic concepts for growing and pruning trees remain unchanged, but the choice of the splitting criterion has been modified to incorporate the censored survival data. The application of DT methods for survival data are described by a number of authors (Gordon Olshen, 1985; Ciampi et al., 1986; Segal, 1988; Davis Anderson, 1989; Therneau et al., 1990; LeBlanc Crowley, 1992; LeBlanc Crowley, 1993; Ahn Loh, 1994; Bacchetti Segal, 1995; Huang et al., 1998; KeleÃ…Å ¸ Segal, 2002; Jin et al., 2004; Cappelli Zhang, 2007; Cho Hong, 2008), including the text by Zhang Singer (1999). 4. Decision Tree for Multivariate Censored Survival Data Multivariate survival data frequently arise when we faced the complexity of studies involving multiple treatment centres, family members and measurements repeatedly made on the same individual. For example, in multi-centre clinical trials, the outcomes for groups of patients at several centres are examined. In some instances, patients in a centre might exhibit similar responses due to uniformity of surroundings and procedures within a centre. This would result in correlated outcomes at the level of the treatment centre. For the situation of studies of family members or litters, correlation in outcome is likely for genetic reasons. In this case, the outcomes would be correlated at the family or litter level. Finally, when one person or animal is measured repeatedly over time, correlation will most definitely exist in those responses. Within the context of correlated data, the observations which are correlated for a group of individuals (within a treatment centre or a family) or for on e individual (because of repeated sampling) are referred to as a cluster, so that from this point on, the responses within a cluster will be assumed to be correlated. Analysis of multivariate survival data is complex due to the presence of dependence among survival times and unknown marginal distributions. Multivariate survival times frequently arise when individuals under observation are naturally clustered or when each individual might experience multiple events. A successful treatment of correlated failure times was made by Clayton and Cuzik (1985) who modelled the dependence structure with a frailty term. Another approach is based on a proportional hazard formulation of the marginal hazard function, which has been studied by Wei et al. (1989) and Liang et al. (1993). Noticeably, Prentice et al. (1981) and Andersen Gill (1982) also suggested two alternative approaches to analyze multiple event times. Extension of tree techniques to multivariate censored data is motivated by the classification issue associated with multivariate survival data. For example, clinical investigators design studies to form prognostic rules. Credit risk analysts collect account information to build up credit scoring criteria. Frequently, in such studies the outcomes of ultimate interest are correlated times to event, such as relapses, late payments, or bankruptcies. Since DT methods recursively partition the predictor space, they are an alternative to conventional regression tools. This section is concerned with the generalization of DT models to multivariate survival data. In attempt to facilitate an extension of DT methods to multivariate survival data, more difficulties need to be circumvented. 4.1 Decision tree for multivariate survival data based on marginal model DT methods for multivariate survival data are not many. Almost all the multivariate DT methods have been based on between-node heterogeneity, with the exception of Molinaro et al. (2004) who proposed a general within-node homogeneity approach for both univariate and multivariate data. The multivariate methods proposed by Su Fan (2001, 2004) and Gao et al. (2004, 2006) concentrated on between-node heterogeneity and used the results of regression models. Specifically, for recurrent event data and clustered event data, Su Fan (2004) used likelihood-ratio tests while Gao et al. (2004) used robust Wald tests from a gamma frailty model to maximize the between-node heterogeneity. Su Fan (2001) and Fan et al. (2006) used a robust log-rank statistic while Gao et al. (2006) used a robust Wald test from the marginal failure-time model of Wei et al. (1989). The generalization of DT for multivariate survival data is developed by using goodness of split approach. DT by goodness of split is grown by maximizing a measure of between-node difference. Therefore, only internal nodes have associated two-sample statistics. The tree structure is different from CART because, for trees grown by minimizing within-node error, each node, either terminal or internal, has an associated impurity measure. This is why the CART pruning procedure is not directly applicable to such types of trees. However, the split-complexity pruning algorithm of LeBlanc Crowley (1993) has resulted in trees by goodness of split that has become well-developed tools. This modified tree technique not only provides a convenient way of handling survival data, but also enlarges the applied scope of DT methods in a more general sense. Especially for those situations where defining prediction error terms is relatively difficult, growing trees by a two-sample statistic, together with the split-complexity pruning, offers a feasible way of performing tree analysis. The DT procedure consists of three parts: a method to partition the data recursively into a large tree, a method to prune the large tree into a subtree sequence, and a method to determine the optimal tree size. In the multivariate survival trees, the between-node difference is measured by a robust Wald statistic, which is derived from a marginal approach to multivariate survival data that was developed by Wei et al. (1989). We used split-complexity pruning borrowed from LeBlanc Crowley (1993) and use test sample for determining the right tree size. 4.1.1 The splitting statistic We consider n independent subjects but each subject to have K potential types or number of failures. If there are an unequal number of failures within the subjects, then K is the maximum. We let Tik = min(Yik,Cik ) where Yik = time of the failure in the ith subject for the kth type of failure and Cik = potential censoring time of the ith subject for the kth type of failure with i = 1,†¦,n and k = 1,†¦,K. Then dik = I (Yik ≠¤ Cik) is the indicator for failure and the vector of covariates is denoted Zik = (Z1ik,†¦, Zpik)T. To partition the data, we consider the hazard model for the ith unit for the kth type of failure, using the distinguishable baseline hazard as described by Wei et al. (1989), namely where the indicator function I(Zik Parameter b is estimated by maximizing the partial likelihood. If the observations within the same unit are independent, the partial likelihood functions for b for the distinguishable baseline model (10) would be, (11) Since the observations within the same unit are not independent for multivariate failure time, we refer to the above functions as the pseudo-partial likelihood. The estimator can be obtained by maximizing the likelihood by solving . Wei et al. (1989) showed that is normally distributed with mean 0. However the usual estimate, a-1(b), for the variance of , where (12) is not valid. We refer to a-1(b) as the naà ¯ve estimator. Wei et al. (1989) showed that the correct estimated (robust) variance estimator of is (13) where b(b) is weight and d(b) is often referred to as the robust or sandwich variance estimator. Hence, the robust Wald statistic corresponding to the null hypothesis H0 : b = 0 is (14) 4.1.2 Tree growing To grow a tree, the robust Wald statistic is evaluated for every possible binary split of the predictor space Z. The split, s, could be of several forms: splits on a single covariate, splits on linear combinations of predictors, and boolean combination of splits. The simplest form of split relates to only one covariate, where the split depends on the type of covariate whether it is ordered or nominal covariate. The â€Å"best split† is defined to be the one corresponding to the maximum robust Wald statistic. Subsequently the data are divided into two groups according to the best split. Apply this splitting scheme recursively to the learning sample until the predictor space is partitioned into many regions. There will be no further partition to a node when any of the following occurs: The node contains less than, say 10 or 20, subjects, if the overall sample size is large enough to permit this. We suggest using a larger minimum node size than used in CART where the default value is 5; All the observed times in the subset are censored, which results in unavailability of the robust Wald statistic for any split; All the subjects have identical covariate vectors. Or the node has only complete observations with identical survival times. In these situations, the node is considered as pure. The whole procedure results in a large tree, which could be used for the purpose of data structure exploration. 4.1.3 Tree pruning Let T denote either a particular tree or the set of all its nodes. Let S and denote the set of internal nodes and terminal nodes of T, respectively. Therefore, . Also let |Ãâ€"| denote the number of nodes. Let G(h) represent the maximum robust Wald statistic on a particular (internal) node h. In order to measure the performance of a tree, a split-complexity measure Ga(T) is introduced as in LeBlanc and Crowley (1993). That is, (15) where the number of internal nodes, |S|, measures complexity; G(T) measures goodness of split in T; and the complexity parameter a acts as a penalty for each additional split. Start with the large tree T0 obtained from the splitting procedure. For any internal node h of T0, i.e. h ÃŽ S0, a function g(h) is defined as (16) where Th denotes the branch with h as its root and Sh is the set of all internal nodes of Th. Then the weakest link in T0 is the node such that   <

ESP Methodology And Syllabus

ESP Methodology And Syllabus It is debatable whether ESP has a distinctive methodology and syllabus. This paper argues that methodology and syllabus design in English Language Teaching (ELT) andESP differ little and that it is not possible to say whether general ELT has borrowed ideas for methodology from ESP or whether ESP has borrowed ideas from general ELT. two characteristic features of ESP methodology are identified: ESP can base activities on students specialism, and ESP activities can have a truly authentic purpose derived from students target needs. Dudley-Evans and St. John(1998) maintain that what characterizes ESP methodology is the use of tasks and activities reflecting the students specialist area Introduction In the 1970s, EFL teachers first ventured out of the Arts Faculty and the gentle landscape of language and literature into the land beyond the mountains inhabited by illiterate and savage tribes called scientists, businessmen and engineers, wrote Ramsden (2002). In the light of this quotation, Ramsden pours his scorn over the turning point in the history of language teaching from art to science; and from English for general purposes(EGP) to English for specific purposes(ESP) . Though ESP emanates from EGP, it has established itself as a distinct trend. The distinctions between ESP and EGP are quite fuzzy. To clarify the issue, Hutchinson and Waters (1987) pointed out that there is no difference in theory, but in practice, there is a great deal. This paper delves deeply into the literature of ESP and EGP to uncover their points of similarities and differences, chiefly at the level of syllabus design , methodology and instructional materials. For the sake of clarification, theoretical preliminaries will be provided from the outset. As expected, the current paper is comparative in nature and selective in illustration. Theoretical Preliminaries: Definitions of: EGP: According to Blackwell, EGP is polarized with ESP ( English for specific purposes) to refer to contexts such as the school where needs cannot readily be specified. This view is misleading, since purpose is always inherent. EGP is more usefully considered as providing a broad foundation rather than a detailed and selective specification of goals. EGP, then, refers to that basic linguistic code that could be used in larger context and in everyday conversation. It does not take into account neither the requirements of a workplace nor needs of learners. Being general in its nature, EGP holds a sway at the core level of language instruction. ESP According to Longman dictionary of applied linguistics, ESP refers to the role of English in a language course or program of instruction in which the content and aims of the course are fixed by the specific needs of a particular group of learners. For example courses in English for academic purposes, English for science and technology, and English for Nursing. In this regards, ESP is chiefly associated with special language or register. However, Hutchinson and Waters )1987, p.19) claimed that ESP is not a particular kind of language or methodology, nor does it consist of a particular type of teaching material. Understood properly, it is an approach to language teaching. From the above definitions, one can notice that there is no absolute clear cut between ESP and EGP. To ask which one embraces the other is likely to generate divergent views. In an attempt to answer this question, Hutchinson and waters ) 1987.p.18) have drawn a tree of ELT where the ESP is just one branch of EFL/ESL, which are themselves the main branches of English Language teaching in general.. However, A closer gaze at the tree and to the ramifications of ESP and EGP uncovers the distinctive features of each. These features will be tackled in subsequent section. Distinctive features of ESP and EGP: Despite the overlapping connections between EGP and ESP, there are several differences at the level of their concerns and practices. First, the focus in ESP is on training students to conform well to the requirements of the workplace; whereas, in EGP, the main focus is on education. Widdowson( 1983) sees the difference between Education and Training as that of creativity versus conformity (in White, 1988: p.18). Second, Designing a course content in EGP is much more difficult than in ESP for the difficulty of predicting the future needs of EGP students. Knowing about only learners survival needs is quite unbeneficial because it may lead to an oversimplified language, unauthentic communicative structure and unrealistic situational content. Third, ESP learners are usually adults with an average mastery of English language. Their main purpose is to communicate and learn a set of professional skills. In EGP, the age of learners , however, varies from childhood to adulthood. Their chief purpose behind learning English is to achieve communication in the basic everyday communication. At the level of macro-skills, the four language skills are integrated and reinforced in EGP instruction, while in ESP the selection of language skills is based on needs analysis. For instance, in studying English for science and technology, the emphasis is on context and subject of the course. At the level of micro skills, EGP has shed too much attention to teaching of grammar and language structure; yet the focus in ESP is on the context and subject of the course. Finally, a distinctive feature of ESP classroom is team- teaching, where the teacher of language collaborates with subject teacher in the delivery of the lesson. This feature is , however, absent in EGP classroom where the language teacher seems sufficient to instruct broad themes. To sum up, though ESP stems from EGP, it has preserved for itself distinctive characteristics as outlined before. To sum up, Stevens states that ESP has four absolute characteristics: 1. Is designed to meet specific needs. 2. Is related to themes and topics particular to occupation. 3. Is centered on language appropriate to those activities, in terms of lexis, syntax, discourse pragmatics, semantics and so on. 4. The above is in contrast to General English (Stevens 1988 in Dudley-Evans St. John 1998: p.4). In the subsequent section, the paper will take both EGP and ESP a stage further to list the similarities and differences at the level of syllabus design. To facilitate the process of comparing and contrasting, an example of each course content will be highlighted. Syllabus design in EGP A syllabus refers to a particular plan of a course. It is a document that details the structure and operation of ones class. It can also be called the basic reference document that guides students and the instructor through a course (Breen 1984). In the current section, this section aims to uncover the salient types of syllabus adopted in EGP and ESP based on contents of two textbooks: Natural English( EGP textbook) and English for Careers: Tourism, (ESP textbook) Based on their observations of general English language courses, Brown (1995) and Richards (1990) list the following types of syllabuses. They also point out that courses are often based on a combination of: Structural (organized primarily around grammar and sentence patterns). Functional (organized around communicative functions, such as identifying, reporting, correcting, describing). Notional (organized around conceptual categories, such as duration, quantity, location). Topical (organized around themes or topics, such as health, food, clothing). Situational (organized around speech settings and the transactions associated with them, such as shopping, at the bank, at the supermarket). Skills (organized around microskills, such as listening for gist, listening for specifi c information, listening for inferences). Task- or activity-based (organized around activities, such as drawing maps, following directions, following instructions). Extract.1: Contents of Natural English , As can be observed in the content of Natural English, one of the main aims of the textbook is to enable General English learners to improve the four language skills, especially speaking and listening to everyday English. Yet, the integration of the four language skills is not the sole distinctive feature of the textbook. The contents of course book also seem to respond to the general wants of GE learners in that it all covers functions, notions, vocabulary and grammar. Each unit introduces GE learners to notions, functions and grammatical structures in an equal weight of emphasis. Thus, a point that one can infer is that EGP syllabus is integrative. Language skills as well as functions, notions, forms and semantic entries are all fused together. For example, in unit 2, the book introduces notions such as shopping and work. Concerning functions, expressing request and responding with sympathy are the main functions found in unite 1. The grammatical forms are so varied from using the present continuous to passive voice. What is so remarkable is that the communication of a notion entails the use of adequate target functions. From the design of Natural English, it is evident that the units are organized on topics. Unit one is on Cartoon Mobile Invasion, unit two on Joke lost in desert, and three on Cartoon Perfect Day. However, a striking existence of situations looms chiefly in extended speaking. Students are in front several situations, such as on train , on holiday, and are encouraged to interact , following the necessities of imagined communicative setting. To conclude, the pertinent remark we can deduce from the course content of EGP is that its syllabus is integrative and synthetic in nature. Functions, notions , forms, situations and skills gain enough space in the EGP syllabus. Nevertheless, these elements are tackled more broadly. For instance, It seems that the subjects are too general, the functions and notions are recurrent in daily life issues, and language skills are not relevant to any professional field. Now , ESP makes extensive use of content-based approaches. According to Master and Brinton (1998), CBI has the following features. The syllabus is organized around subject content; for example, in English for Careers: Tourism, an ESP textbook, the subject matter is on a number of topics from tourism, such Registration Client perceptions and supply and demand. Teaching activities are specific to the subject matter being taught and are geared to stimulate students to think and learn through the use of the target language. Language is viewed holistically, and learners learn from working with whole chunks of language and multiple skills. Content-based approaches reject synthetic approaches to course design-the idea that language or skills can be atomized into discrete items to be presented and practiced by learners one at a time. The approach makes use of authentic texts to which learners are expected primarily to respond in relation to the content. It has been argued (Hutchinson Waters, 1987) that once we remove the text from its original context, it loses some of its authenticity. For example, the intended audience is changed once the authentic text is imported into the classroom. Authenticity also relates also to the readers purpose in reading the text. For example, recommendation reports for the purchase of technical equipment are, in their original context of use, devised for the purpose of helping the reader decide which of two or more items of equipment to buy. If, however, a recommendation report is transported into a language teaching classroom and students are given an activity whose purpose is to answer c omprehension questions on it, the match between text and task is artificial. Content-based instruction tries to avoid some of these potential problems by using content (authentic texts) in ways that were similar to those in real life. Content-based approaches involve also the integration of skills. Writing often follows on from listening and reading, and students are often required to synthesize facts and ideas from multiple sources as preparation for writing (Brinton et al., 1989). In fact, ESP syllabi (in this case an English Vocational Purposes syllabus) differ from English General Purposes (EGP) syllabi, both in goals and content. Below is an outline of some major differences adapted from Widdowson (1983 in White 1988: pp.18 26), Hutchinson Waters (1987) and Stevens (1988) (both in Dudley-Evans St. John 1998: pp. 2-4). The ESP syllabus must be based on a previous analysis of the students needs, which includes not only an analysis of the situations in which the language will be used and of the language appropriate in these situations, but also an analysis of the students wants and subjective needs. The whole business of the management of language learning is far too complex to be satisfactorily catered for by a pre-packaged set of decisions embodied in teaching materials. Quite simply, even with the best intentions no single textbook can possibly work in all situations.(Sheldon, 1987: 1)If we are to prescribe content, we need to ask, whose content? Methodology Having uncovered the nuances existing between ESP and EGP syllabi, This current chapter will move a stage further to draw a comparison and contrast at level of methodology, chiefly at the types of techniques employed by each and the roles they played in serving the students needs. As defined by Robinson (1991), methodology refers to what goes on in the classroom and to what students have to do. Using technical terms, it refers to classroom activities and techniques. There are too many techniques which largely emerged in EGP classroom such as tasks, role play, simulations, and so on and so forth. These techniques soon adopted by ESP practitioners . Concerning tasks, Little John and Hicks ( ) noticed that valuable tasks in EGP have certain characteristics: they should be motivating and absorbing; and exploit learners prior knowledge. In ESP, the above criteria are also predominant, but what is specific here is that ESP tasks comprise linguistic and professional skills. For instance, medical students studying English may be assigned to carry out a series of operations as outlined below: Moreover, the role play and simulations are used differently in ESP and EGP. While dealing with simulatons in ESP, Strutridge() noticed that they were originally used in business and military training with focus on outcome rather than the means -language- of training. In EGP, the outcome was ,however, less important than the means used to achieve fluency. One should not perceive hastily that means in esp have no disregarded. Stutridge concludes that in ESP end is as important as the means. Taking case studies into account, Nunan in an outsanding research tested the validity of the technique to ESP course. He found out that it helps ESP students to draw upon their professional skills, utilizing the cognitive and behavioral styles of their work rather than of traditional language classroom. Case studies may prove difficult to be conducted by EGP learners if we take into consideration their younger age and Worse of al their professional immaturity. For ESP students who are not fully qualified in their profession, the use of case studies help to induct them into some aspects of professional culture ( Charles 337,pp.28-31) Project work is out-of-class activity used in both ESP and EGP classroom. However, Fried() observed the more advanced examples of project work would be appropriate for ESP. A final technique which is common in ESP and EGP as well is the oral presentations. Usually, they are the culmination of project or case studies conducted outside the threshold of classroom. The utility of such activity is that it trains students to develop their self autonomy and master the four skills of the target language. Word processor and PowerPoint become familiar means for presentations, Succinctly, the methodology endorsed by ESP is quite similar to that of EGP chiefly if we consider the types of techniques and activities .Yet, the ways in which techniques are employed in ESP differ a lot from that in EGP. the next chapter will attempt to decipher how material design becomes a site of innovation after the emergence of ESP. Being in its heyday, ESP materials assume a divergent way from EGP. ESP designers come up with in-house materials quite plausible to the students needs more than the General ready-made textbooks which hold their strength in EGP classrooms. Materials Design One of the common characteristics of of material design in ESP is the existence of an established tradition of ESP teachers producing in-house materials. These materials are the outcome of needs analysis. the tailor made material accounts to the learners needs more than a general textbook can do., However, several questions may emerge to the surface: What are the major factors behind the over-existence of in-house materials in ESP in contrast to its acute shortage in EGP? -what are the key features that distinguish ESP materials from EGP? One of the key factors behind the profusion of in-house materials in ESP is because of its reliance on needs analysis. Need analysis is rarely carried out in GL classroom. This is partly because of the difficulty of specifying GL learners and partly because of a lack of literature on the particularities of analyzing needs data. Needs analysis tends to be associated with ESP and is neglected in GE classroom. Hutchinson and Waters(1987,p.53-54) say that what distinguishes ESP from GE is not the existence of a need as such but rather an awareness of the needà ¢Ã¢â€š ¬Ã‚ ¦ for the time being, the tradition persists in GE that learners needs cannot be specified and as a result no attempt is usually made to discover learners true needs. Secondly, The fact that ESP materials are tailored to the needs of specific group of learners makes its absolute adoption by other ESP teachers futile. Even when suitable materials are available, it may not be possible to buy them because of import restric tions pointed out Hutchinson and Waters (1987,p.). If textbooks are more available in EGP than in ESP,ESP textbooks have not been immune from criticism. Ever and Boys(p.57) mount a strong a attack on the EST textbooks suggesting that most of them are designed for, or are the outcome of, remedial or supplementary courses and assume that students already possess a knowledge of Englishà ¢Ã¢â€š ¬Ã‚ ¦.unhappily, this is not at all understood by potential users, especially in developing countries abroad where the greatest demand for EST exists. Another strongly worded attack was that the heavy concern of ESP practitioners with methodology and approach leads them to ignore issues such the accuracy of explanations ,validity of examples and suitability of linguistic content. Because ESP materials are relevanct to target needs, This may increase the motivation of ESP students, but there are other aspects which are also highly important, such as Waters (1987: 48) put it, ESP, as much as any good teaching, needs to be intrinsically motivating. () Students should get satisfaction from the actual experience of learning, not just from the prospect of eventually using what they have learnt. The following task, for instance, could be interesting for Engineering students:. Another characteristic of ESP materials is that it is more authentic than EGP materials. The latter might be produced for the purpose of teaching language, while in ESP authenticity refers to the materials used in the students specialist workplace or study institution. Additionally, for ESP authentic text selection usually follow the needs analysis. To conclude, the whole business of language learning management is far too complex to be satisfactorily catered for by a pre-packaged set of decisions embodied in teaching materials. Quite simply, even with the best intentions no single textbook can possibly work in all situations.(Sheldon, 1987: 1). However, designing tailor made materials would in principle be motivating, authentic and innovative. Conclusion This paper has highlighted some of the issues involved in ESP curriculum development. It can be argued that language varieties are based in and extend from a common core of language. Or it can be argued that language varieties are self-contained entities. Needs analysis can be seen as an entirely pragmatic and objective endeavour to help course developers identify course content that is truly relevant to the learners, or it can be argued to have a bias in favour of the institutions and may overemphasize objective needs at the cost of subjective needs. It can be argued that syllabuses should specify content (what is to be taught). Or it can be argued that they should specify method (how language is to be taught). Some argue that the ESP courses should be as narrow-angled as possible. Others argue that this is not practica EST is in a parlous state and is being abandoned by many tertiary institutions who, like Sultan Qaboos University, found that the English teachers seemed to learn a lot of science, but the students didnt seem to learn much English