註冊 登入


Draft: PhD and Beyond

Draft: PhD and Beyond E-mail 此主題給朋友

PhD andBeyond

1. Whatis PhD for?
2.Admissions & Financial Considerations
3. QualifyingExams & Second Year Paper
4Dissertation Research + Oral Defense
5.Research & Publications
6. JobSearch and Career Opportunities

回覆 引用 TOP

1. What is PhD for?

PhD is a research degree. The objective is NOT to provide the student with in-depth knowledge of a particular subject, although that will usually occur as a side-effect. The primary goal is to train the student, through an apprenticeship with an advisor, in doing PhD level research. The obvious question is then what is "hD level research". I add the words "hD level" in front of research to distinguish it from "research" done for other purposes ... such as a class project. There will be more discussion in later part of this document about what "professional research" actually means.

As a result, while PhD seems to many a "higher level" degree than the master, it is erroneous to assume that a PhD is a super master. Unless a person is really motivated and wants to devote his career (and probably his life) to research, he or she probably should NOT pursue a PhD. Trying to obtain a marketable degree is definitely NOT the right reason to pursue a PhD. If making money is your primary goal, stop reading now, go get a master in financial engineering and go to Wall Street. You will make 3x the money your professor is making.

回覆 引用 TOP

2. Admissions & Financial Considerations

Getting into the top PhD programs is hard. It is probably much much harder than getting into a top undergraduate program. The competition is much more intense. For example, last year, the strategy group at the MS&E dept at Stanford took about 3 new PhD students out of 30-40 applicants. Furthermore, every applicant had a perfect GRE score (btw, about 6% get a perfect score on the quant part of GRE). So if you don't have a perfect score (or a close to perfect one), forget about applying to top PhD programs in technical disciplines.

A great GRE score and good grades are just necessary conditions so people will consider you. You also need great recommendation letters, ideally from well-known professors whom you have worked with, and preferably some research experience (like interning at a lab during your undergrad years).

I have written recommendation letters for my interns for PhD program applications. Typically, the program would like to evaluate an applicant in 4 dimensions: the potential to do research, technical skills, communication skills and motivation. The ability to work independently and in teams can also help.

2.1Choosing a School, a Department and an Advisor

Considerationsgoing into making choices about which PhD program to go to, which department tojoin and which advisor to work with should be quite different from choosing amaster program or a college for undergraduate studies.

First,the overall reputation of the school and even the department should not factoras much. Not that I said “not as much”, not “not at all”. Obviously, thereputation of the department, and a lesser extent, the school, would still havean impact on your first job application. However, the effect will only limit tothe chances of whether people will filter you out in the first round. As longas you are in the consideration set, other aspects such as your own researchand recommendations will become dominant.

The maintwo issues to consider are a) your research interests and b) whether you canwork with the professor/group. Here is an approach I would recommend. Assumingyou already know the general research area you are interested in (by the way,it has to be more specific than just physics or operation research, somethinglike quantum computing is probably good enough), and also assume you have apretty good chance in getting into say the top 10 departments, you should startresearching faculty in these departments starting from the biggest names. Atthis point, it is not necessary to read every paper you come across because itwill take too much time.

What youshould do is to skim through papers from several professors (starting from thebig name) in a department to figure out the general gist of their research. Andthen you should figure out which department matches your interests the best. Interms of reputation, as long as you are going to the top 10 department in thecountry, research interests overrides reputation effect. In many cases, thereputations of the departments are different in specific areas anyway. Forexample, it is not difficult to figure out that if you want to do experimental economics,Caltech and George Mason are top choices (Arizona will be another good choice).

Thesecond issue, whether you can work with a particular professor, is a muchharder one to address before you actually entering the school. A lot of placeswill fly you out to visit the department in a open day kind of thing foraccepted PhD students. You should talk to the existing PhD students and get afeel for how it is like working for particular professors. During your firstyear, before you have to decide on a dissertation advisor, try to work with andtalk to as many professors as possible. Remember again that while your advisor’sreputation is important, it is more important that you can develop a good workingrelationship with him. You will be working very closely with him (meetingmultiple times a week and may be spending long hours in a room talking aboutresearch) and often the relationship will develop into life-longcollaborations/friendships. So you need to choose wisely. It is also notunheard of that sour relationships with advisors can sink one’s career.

2.2 Financial Support

It is typical for good research universities to FULLY support their PhD students. That means the tuition is usually waived and the student is paid a livable stipend. Many schools also have graduate student housing. These are usually apartments and are better than the typical undergraduate dorms.

Financial support for PhD students usually come in one of the three flavors. The really good students may receive fellowships (some may needed to be applied, an example is the SLOAN fellowship which you need to be recommended by professors before they will even consider you) which essentially support you to do your own research.

Otherwise, a student may have to be a research assistant (RA) to a professor or a teaching assistant (TA). Schools usually limit the amount of RAing and TAing to no more than 20 hrs a week. You will be “cheap labor” to help with teaching or doing grunt work for professors. While the work is only part-time, they will make sure you will be paid enough to survive.

[ 本帖最後由 narius 於 2007-8-13 02:35 AM 編輯 ]

回覆 引用 TOP

3. Qualifying Exams & Second Year Paper

I will focus the discussion on the US system. UK/HK is slightly different with many students doing a Mphil before a PhD. In the US, there is no distinction between taught and research master. Typically, students go to PhD programs directly after their undergrad degree. It is not uncommon to do a master either, but it is not required.

The first year of a US PhD program is very much like a master program with lots of courses. The material may be a bit more advance than master level (because teaching material will include more latest published papers).

Usually at the end of the first year, students take the qualifying exam. It is a multiple section exams (for example, the one at Caltech HSS dept is divided into micro economics, econometrics and political science/game theory) and it is pass-fail. The goal is to ensure students have the necessary knowledge and technical skills to complete their PhD. If a student passes, he/she will be known as a PhD candidate. If not, he/she will be given a master and asked to leave.

After a student passes, the real work begins.

Many departments require their PhD candidates to do a second year paper. Usually this is a smaller piece of research and the goal is to get the student’s feet wet into the whole process from doing the research, to write up the results to presenting it to an audience. So it involves not just writing a research paper but also a presentation of the research result. This usually happened in the later part (say Apr) of the 2nd year. Many students would expand on the work in the second year into either a chapter in their dissertation or even a future publishable paper.

回覆 引用 TOP

4 Dissertation Research + Oral Defense

In fields including OR, economics, management science, a dissertation consists of around three paper's worth of research material. It does NOT have to organize into only 3 chapters though. The student's primary focus is to finish a good dissertation. Secondly, it is typical to turn at least one chapter into a paper and submit it. That would take extra work. Typically, PhD candidates are NOT encouraged to take on too many projects because that will dilute their attention from their dissertation research. At this stage, quantity of research is not required or even encouraged. However, it is not uncommon for them to take on some additional small projects (as RAs or interns somewhere else) but it is also not required for them to turn these into papers.

Each candidate will have a dissertation committee. The make-up is different from school to school. However, it is usually made up of 4-6 professors including the candidate’s advisor. The rule is that if all of them agree and sign on a piece of paper, the candidate get his PhD.

Aside from the dissertation, the candidate is also required to defend his research orally. In essence, he/she gives a presentation of his research to the committee and answers questions to their satisfaction. After the defense, the candidate is usually asked to wait outside of the room. If he passes, the committee will invite him back in and congratulates him with the title of “Dr”.

It is not uncommon for some students to finish their dissertation AFTER they started their first job. Usually they are still considered a PhD even before they finish the dissertation (but obviously after their oral defense). If a person takes too long or is unable to finish, there may be bad consequence though.

A typical PhD in OR/IE department is 4-5years (in the US, typically without a Mphil). Economics and other business school disciplines have similar timing. 3 years is typical the time needed for dissertation research + writing. This is typically AFTER a year of course work, passing qualifying and doing a second year paper as training before the "real" research starts.

回覆 引用 TOP

5. Research & Publications

If one pursues a career in research, it is not sufficient to "just" get a PhD. You really need to do good research to distinguish yourself. In this section, we explore more about what PhD level research means. I will also contrast that with research students asked to do at a lower level.

In the US, students are asked to start writing "research" paper at a young age. Even my kids, who are going to elementary school, are asked to do "research" and write "papers". Typically, a research paper, in abroad sense, requires the author to find information relevant to the topic, add in his own thesis and arguments, and write it up in a coherent fashion. This definition can be used to describe research papers up to undergrad/master level work. Obviously, in some master thesis, there may be some additional professional research elements that may approach PhD level research. It is not common to see this at all in research paper as class assignment. Master thesis, if done well, may include some PhD level research.

So the obvious question is “what is PhD level research?”. There are several considerations.

First, the research has to be original. That means that it is not sufficient to just find information and put them together. An example in a scientific field is that the researcher developing a new mathematical model to explain a previously unexplained phenomenon. Another example is designing and conducting of a new experiment with a new discovery.

Second, the research has to be done "right" with rigor. This point is more relevant to scientific fields (including math & social sciences) than humanities. However, even in humanities, you need to do the research with "good" methodological considerations.

Third, the research has to have a minimal level of impact and add to the literature. That means that it is not sufficient if I take someone else's experiment and do a trivial variation.

These considerations are often hard to judge. Thus, the major method of evaluating research is through the peer-review process. Papers sent to journals are reviewed by multiple experts (typically 3) in the field in an anonymous fashion. Top journals, Management Science for example, has typical publication rate of 10-15%. Not all fields use journals as the primary publication vehicle. For example, in CS, top conferences (like ACM-e-commerce) are more prestigious and more difficult to get in.

PhD dissertation research is usually required to have the *same*quality as professional peer-reviewed research although it is done via a different process. Because of the time constraint, it is usually not possible for a PhD student in fields like business or econ to get their research published by journals (typical review cycles can go for months and even years). The usual way it is done is the PhD dissertation will be reviewed by a committee which can ok the research. The normal expectation is that the dissertation will then be turned (rewritten and may be broaden) into multiple papers.

Typically in business schools (and Econ, IE/OR depts), a graduating PhD student will have one or two submitted (and may be accepted) papers in journals. CS is different because conferences reviewing cycles are much shorter.

Research projects can take substantially amount of time. Usually each project is a significant investment of time on the part of the researcher. For example, a supply chain decision experiment can easily take 6 months to 1 year just to do the research. Furthermore, it is not uncommon to take another 6 months to write up the results.

Publishing in top journals is difficult. Even faculty would have problem publishing in a top journal like Management Science.

回覆 引用 TOP

6. Job Search and Career Opportunities

6.1 Getting the first job

PhDs are evaluated very differently from taught masters and other taught degrees. If you apply for professional research jobs (include tenure professionships, research positions in industry, government or think tanks ...), people will scrutinized your research, not your grades (in fact, grades won't matter at all). A typical interview will involve the candidate giving a talk on his research, then being grilled by everyone, followed by one-on-one interviews with the whole dept.

A person will be evaluated on multiple dimensions including his research, how smart and creative he is, his communication skills, motivation, independence and team work.

6.2 Industry vs Academia

Career opportunities exist both in industry (a typical job title is “scientist”) and academia. Once again, since the US is a major player in research, most of the discussion focuses on the US environment. There are trade offs between industry research positions and academia tenure track positions.

First, the obvious observation that you have more freedom in academia than in industry is correct. However, the picture is absolute. Let's start with industry.

The amount of freedom you have in industry depends on the organization as well as your track record. Research organizations like HP Labs (and the formal glorious Xarc) give researchers relatively plenty of freedom. This is easiest to illustrate with examples. Thus, I would use my own experience as anecdotal evidence.

First example, the guy in the next office to mine does nothing but publish.

However, do not confuse between freedom & usefulness of the research. I can pretty much do whatever research I want but it has to have some connections, no matter how thin or far into the future, to actual applications. Roughly speaking, my research is half basic research (the type of stuff you see in management science) and half applied research (Interfaces). Note that real problems are not necessarily less interesting. So industrial research is not a bad option if you research is applied (or have consequences that can be applied) oriented.

There are other (sometimes as important) trade offs aside from freedom of research. For example, resources are a bit more plentiful in the industrial side. I have a lab, several interns, a contractor to do experiment implementation, a team in India to do research software, enough money to run any experiments that I want and I spent less time on grants than my academic counterparts. I have no teaching duties (although I like to go give guest lectures from time to time) and we pay better than almost all academic departments except may be the top business schools.

The flip side is that you have to deal with business people (but sometimes they give you data which is a dream come true if you are in business school research), and you don't get summer off.

Now let's go to academic side of things with respect to research freedom. While, to some extent, it is true that you can pretty do whatever you want as long as you publish, there are more subtle considerations. For example, often it is necessary for new assistant professors to collaborate because otherwise they can't publish efficiently In fact, how many assistant professors can do management science papers on their own right from the beginning? The answer is not many and not often. It is much easier to publish with a partner in the beginning.

Thus, the ability to attract others to work on your problem is also important. So whether you research is viewed favorably in a dept has some impact on your ability to do research. Plus that it obviously has implications of whether you can get funding easily or not.

6.3 Financial Compensations

While one should not be in the research game purely for financial gains, professors and scientists typically make a very comfortable living. However, compensations can differ greatly from one field to another. For example, business schools, law schools and medical schools compensate professors a lot better than other departments.

In the US, schools have very high flexibilities (including public schools like Berkeley although they have a bit more restrictions and bureaucracy) to hire professors, particularly the stars. B-school are typically cash cow of the school with large endowments. Furthermore, they want to draw people from their own and other disciplines (economics for example). Thus, they typically offer much larger salaries (and research support) compared to other depts. B-schools also have to compete with industry for obvious reasons.

BTW, industry also pays well. For example, an industrial lab can pay better than almost all academic depts except business schools. It is not uncommon for Fresh PhD to start in the range of $100-120k (for 12 months). And we haven't talked about stocks and bonus yet.

6.4 Tenure Track Academic Positions

Obviously academic employs a large bulk of researchers. Almost all research universities have a tenure system. The idea is that once a professor has proven himself, he should be given freedom to pursue whatever research he wants to do. And to ensure this freedom is not tempered with, he would be offered a life-long contract at that point so no one can fire him because of his research.

The system usually works the following way. A fresh PhD is hired at the rank of the assistant professor. He/she will be reviewed for tenure in 6 years. (Usually there is also a mid-point review at 3 year but that is more like a progress report than a decision point.) If he/she receives tenure, he/she will be promoted to associate professor and no one can fire him/her without cause (such as committing a felony or unethical behavior).

After a grad got their first assistant professorships, normally he or she starts to turn his dissertation into papers. At this point, the amount of publication is important. A typical 2nd tier school (like Kansas) would require around 4 a-list publications (like management science) to get tenure. A place like Berkeley will require 8-10. This is not counting the quantity (about the same number or more of lesser papers). That is extremely tough to do in 6 years. We are talking about almost 1 top publication per year. Tenure review also requires recommendation letters.

6.5 Industry Positions

Different companies have different systems. However, industry labs usually have technical career track so that top research talent does not have to go into management. A typical promotion path is scientist -> senior scientist -> principal scientist -> fellow -> senior fellow.

A fellow or a senior fellow would be equivalent to a chair professor in academia. There is no true one-to-one correspondence because the criteria of judging a scientist can be different than that to judge a professor. For example, a scientist may be evaluated based on his publication, as well as his patents and/or the relevancy (to business) of his research.

6.6 Post-doc Position

In China,post-docs are extremely common and some actually expect it to be part of aresearcher’s career. In the US,the situation is much more varied. Post-doc, short for post-doctorate, is theunofficial name for visiting positions in a research organization. Bothindustry and academia offers these kinds of positions, usually to fresh PhDs. Theactual title may be visiting scholar, visiting professor and something of thatnature. Usually, the position is for a pre-determined limited amount of time (1year is the most common but it can be two or even 3).

Inindustry, there are also post-doc positions that are designed to act as a trialperiod and have the possibility (and sometimes with very high probability) ofturning the position into a regular scientist position.

Whetherit is a normal part of a researcher’s career depends highly on the field. Forexample, most business school PhDs and economists go directly to tenure trackacademic positions. It is more common to do post-doc in physics and otherphysical sciences.

While atfirst glance, a post-doc position may seem less desirable than a tenure trackposition because of the uncertainty involves (no one wants to be on the jobmarket AGAIN in a year), it has its own advantages. First, most post-doc position,even in academia, is research only and there are no teaching duties, so you canfocus on your search. Second, if you start as an assistant professor, thetenure clock starts ticking right away. A post-doc position will allow you moretime to build up a research track record, which will count towards any futuretenure review.

Asidefrom career considerations, a post-doc position may also open the opportunitiesto work with particular people in the field.

[ 本帖最後由 narius 於 2007-8-13 02:36 AM 編輯 ]

回覆 引用 TOP

reserve reserve reserve

回覆 引用 TOP

reserve reserve reserve

回覆 引用 TOP

reserve reserve reserve

回覆 引用 TOP

reserve reserve reserve

回覆 引用 TOP

reserve reserve reserve

回覆 引用 TOP

Hi Narius,

Would you please add a section about how to pick the right school or right group to work with?

I think that would be really useful.... because sometimes going to the famous school might not be the best thing to do.... If one knows exactly what he/she wants, picking the right group, or the right advisor might be more important....

Thanks for doing this!

[ 本帖最後由 RandomCoil 於 2007-8-12 06:19 AM 編輯 ]

回覆 引用 TOP

原帖由 RandomCoil 於 2007-8-12 05:55 AM 發表
Hi Narius,

Would you please add a section about how to pick the right school or right group to work with?

I think that would be really useful.... because sometimes it is going to the famous ...
Sure .. I will do that in a day or so.

回覆 引用 TOP

原帖由 narius 於 2007-8-12 06:03 AM 發表

Sure .. I will do that in a day or so.
Thanks Narius.

Another thing.... About the career path for an academic,  is postdoc experience common in your field? In basic sciences,  getting some postdoc experience seems unavoidable (guess the schools want you to prove you are really good before allowing us to try the tenure track position ... and that is another 3 years.)  On the other for engineering, the situation is less clear cut... There is always a chance we can go directly to get a assistant professorship if one is really good, though a lot of us will still end up becoming postdoc....  

Should you mention something about postdoc opportunity?  I guess it would depend on how sepecific you would like to get into.

Thanks for you effort.  We really appreciate it.

P.S. This is actually a question I am thinking about.  I mentioned in my other post that I am graduating probably in a year and have started looking for a job... Postdoc or assistant professorship??? I don't know which one would be a better choice for me (or other fresh PhDs).  Doing a postdoc will widen our research horizon and build up connections.... But it go backs to the question of how to pick a group, finance, and stuff like that.... It is difficult decision...
Anyway, I have applied for a lecturership in Europe (my first application)... will probably start applying for jobs in the States as well... I am just not sure if I should apply for postdoc at the same time....

回覆 引用 TOP

 提示:支持鍵盤翻頁 ←左 右→
[按此隱藏 Google 建議的相符內容]

重要聲明:本討論區是以即時上載留言的方式運作,香港討論區對所有留言的真實性、完整性及立場等,不負任何法律責任。而一切留言之言論只代表留言者個人意 見,並非本網站之立場,讀者及用戶不應信賴內容,並應自行判斷內容之真實性。於有關情形下,讀者及用戶應尋求專業意見(如涉及醫療、法律或投資等問題)。 由於本討論區受到「即時上載留言」運作方式所規限,故不能完全監察所有留言,若讀者及用戶發現有留言出現問題,請聯絡我們。香港討論區有權刪除任何留言及拒絕任何人士上載留言 (刪除前或不會作事先警告及通知 ), 同時亦有不刪除留言的權利,如有任何爭議,管理員擁有最終的詮釋權 。用戶切勿撰寫粗言穢語、誹謗、渲染色情暴力或人身攻擊的言論,敬請自律。本網站保留一切法律權利。

Copyright©2003- Discuss.com.hk Limited. All Right Reserved.