how to effectively do literature search and review?

[Xiaowen Lei]

hi dear all

I am a master student who is doing a project in economics now. I find that two problems are hard for me.

1. how to effectively organize literatures and provide a good literature review? Cos sometimes I need to search for a long time and then find that the literature exactly suits the issue that I am going to do. Also, is there any recommended book on how to do research work on economics?

2. How do you define and organize other people's contribution to the issue and figure out something that you can contribute? The question sounds easy but I am curious because everyone might have different thinking method and hope you could help me to figure that out.

A master student in economics

[Teresa Rutherford]

A literature review establishes where your dissertation ‘fits in' to the existing body of knowledge. However, many academic papers have very brief literature reviews, and sometimes they are confined to the introduction rather than being a separate section in their own right.

When researching your dissertation, it is not uncommon to read 20-30 journal articles. These will form the basis of discussion in any literature review. As you will probably have to read some articles anyway, the reading burden is not excessive.

Identifying which articles are important is a stumbling block. Asking members of your department or supervisors for key readings can get you started.

Check out this link. It might be helpful to you: http://www8.esc.edu/ESConline/Across_ESC/library.nsf/3cc42a422514347a8525671d0049f395/46c31e3773d8747b852570ad00700699?OpenDocument

Goodluck!

[robot_hamster]

Look up how often something is cited by others. If it is cited a lot, then it would probably be a good idea to cite it too. Otherwise someone else who is familiar with the area of study might read it and ask why you didn't include X article by Y and Z because it is a very important article.

[Eigen]

20-30 articles sounds like very few for a dissertation. I wrote a review article last year, for which I probably read over 200.

Keep in mind that you'll read a lot of articles that you won't include in the actual discussion, but will help keep it in perspective.

I'll also disagree that you should cite something just because it is cited a lot- if it's relevant to what you're writing, cite it. If you think it's a central concept, cite it. If it's by a central figure in the field, cite it.

I personally like to start a literature review by looking for recent review articles in the field- something that will give me a starting place. Then I track down each of the cited articles in that central review, and keep and reads the ones that I think are relevant. Then I track down the references from each of those articles, etc. After I've gotten a good body of related work through citation trees, I usually have a good enough feel for the field that I can start running keyword-type searches to find articles that fill in the gaps in what I have or to branch out into new areas.

For actually placing the literature review, I like the "by subject, by chronology" organizational scheme. I divide up the subfield I'm writing on into the major parts, and then review the major developments within each of those parts in a chronological fashion, expalining how it built from one itteration to another.

[robot_hamster]

Eigen - I didn't mean that if something is cited a lot that you absolutely should cite it. I just meant that if something is cited a lot and it is relevant, then it is probably something important. Believe me, I have been chewed out before for not including something that someone felt was an important article. This comes up a lot in discussions about how to write a good literature review, at least at my school it does. The advice that has been given to me was to look at how many others have cited the article if I am in doubt.

Edited February 29, 2012 by robot_hamster

[Eigen]

Ah, gotcha. Yeah, I took your last post as an "always" kind of thing.

I think it's also worth getting used to weathering being chewed out for leaving an "important" article out. My experience (and what I've been told) was that almost every review process, someone thinks that there's some seminal paper that hasn't been accounted for. You just add it, and roll on.

[Xiaowen Lei]

A literature review establishes where your dissertation ‘fits in' to the existing body of knowledge. However, many academic papers have very brief literature reviews, and sometimes they are confined to the introduction rather than being a separate section in their own right. When researching your dissertation, it is not uncommon to read 20-30 journal articles. These will form the basis of discussion in any literature review. As you will probably have to read some articles anyway, the reading burden is not excessive. Identifying which articles are important is a stumbling block. Asking members of your department or supervisors for key readings can get you started. Check out this link. It might be helpful to you: http://www8.esc.edu/ESConline/Across_ESC/library.nsf/3cc42a422514347a8525671d0049f395/46c31e3773d8747b852570ad00700699?OpenDocument Goodluck!

Thank you very much for that.

[RockStar]

Talk to a librarian. He or she will be your best friend in this matter. If you have a few references that you know you are interested in, start with a collection of key articles (probably 10 or more) that you already know you would like to have in your review. Your librarian can help you determine the controlled vocabulary that is being used to index these articles. (For example PubMed assigns subject headings to articles and Embase does something similar). Once you know how these articles are being index you can develop a search logic that will help you yield similar types of articles. If your research revolves around "the economic study of X" then your search logic may look something like : search for "Economics" AND ("synonym1 for X" OR "synonym2 for X" OR....). Then you may wish to look into a reference manager like endnote to organize your references. You can group them according to being included/excluded, discussion material or however you want to set it up.

[Elimba]

You already got great advice above. By the time you see my message, I bet you have already read a bunch of articles and ready to write your review.

When I was doing my PhD several years ago, it took me less than 3 months to read and write a pretty good literature review about a particular topic in C.S. It ended up as a "critical review" journal article praised world wide and used as reading material in several graduate courses . How many articles did I read ? Well, something around 150 although I did not make reference to all them in my journal article. I was reading like crazy, 24 / 7 (except when I was sleeping). I was literally breathing articles and became one with them.

My best advice is "do by example", read what other students have done and specially reviews that have been published.

Good luck to you and other PhD students.

[MoJingly]

I sometimes start with Wikipedia. LET ME EXPLAIN BEFORE YOU SHUN ME.

Don't use Wikipedia as your source, but you can use it as a launching block to find relevant papers. Sometimes those authors are really good at taking a complex topic and trimming it down into something that is easily digestible on a first run. Look in the sources section and follow the links to the papers they cite. It's a pretty good way to start.

[al289]

I heard that we should find key papers related to our topic to have a good overview about the topic. But could anyone please tell me how to find key papers ? (for instance, I need some key papers related to acoustic emission).

Thanks a lot.

[TheFez]

I put together a literature review article in economics last year. The survey included reading about 60 (empirical) papers of which about 40 were included. I had little help real help/direction on how to do this from mentors - so I developed a system of my own and a few tips:

1. I developed a standard review form I would fill out for each article. It contained brief citation information, as well as sections on the estimation method (OLS, FE, RE, Dynamic Programming, Nested Logit, etc.) specification (vars, controls) used, key variables, controls, data set size (N=) and origin (public, private). And a general comments section for interesting observations.

I had a section for main findings, key assumptions, and then a section for observed problems with the paper (e.g. poor specification, endogeneity, reverse causality, data problems, etc.)

I got this to a single page format with protected cells and "fill in the blanks". This was very helpful later on when I forgot many of the details. I numbered each paper and put this number on the form to make easy reference.

[i also read about 20 theoretical papers and used these to to help "round out" the discussion since the theoretical developments were often tied to theoretical progress.]

2. I was told when reviewing papers to read the abstract, introduction and conclusion first. (Actually some professors said they thought this was sufficient - I read every page of every paper - a very time consuming task - but immensely rewarding. [i actually found a mathematical error in one paper that effected a key finding of the paper. I wrote to the author (now at MIT) and he confirmed my calculation and said he would "look into what went wrong during publication" - that was rewarding.]

3. Read a couple of really good survey articles as models for your own. There is even a journal of economic surveys to help this.

4. Use a good citation manager (I now use Zotero, at the time I just used the citation manager in Word - there's tons of options). Make lots of (hard copy and electronic) folders and name the papers with file names in a way you can tell a lot about the paper. I used the main authors name, a description of the good being observed (in my case) and used a grading system A,B,C,D,F at the end of the file name to help rank importance. Don't be afraid to be creative with file names to make recalling them easy.

5. The best way to find good papers is by starting with a good paper and reading it's main citations. This stuff forms a giant tree to climb around in - so try to find one of the most recent articles in the field and work backwards in time.

The tough part of course is how to organize the paper, select important papers rather than including everything, highlight the most relevant aspects of the literature and show connections, and finally to "synthesize" the literature into some original thoughts of your own (e.g. what interesting questions remain unsettled, where is there current debate, what would help move the field forward, etc.)

Hope something here might be helpful.

Edited June 1, 2012 by TheFez

[Asis Shuran]

This assignment is a difficult one. To begin with, you should be focused on the task. Choose one day which you will dedicate to the search. Make notes and analyse all information carefully. Also do not hurry, take your time. In conclusion, check out websites for the professional writers, they might post useful tips. Good luck) 

[juilletmercredi]

It might seem counterintuitive, but I always write my literature reviews after I write the methods and results.

 

The reason why is because your lit review is supposed to situate your work in the larger literature, but it's difficult to do that until you know what kind of results you've gotten. So I work out what my results are and then I tell my story in the lit review, alluding to what's missing and how my work fills that gap.

 

I organize literature using a citation manager - I use Sente (a Mac only program), but a lot of people swear by Mendeley and Zotero (I also like Zotero). Mendeley and Zotero are free; Sente costs some money. All of these programs allow you to organize and tag your literature by subject so you know where to look for related articles. Also, I outline my literature reviews before I write them. That gives me a roadmap for what I'm going to write and how I'm going to organize it. It's a lot easier than starting from a blank page.

 

As for figuring out gaps - that's easier for me. When I read papers, they raise more questions than they answer. The discussion also usually suggests areas for future research. Then you do another search to see if anyone has answered some of those questions. It takes some time - it took me several years in graduate school to feel like I knew enough about the literature to know where the gaps were and where I wanted to establish my research agenda

[TakeruK]

It might seem counterintuitive, but I always write my literature reviews after I write the methods and results.

 

The reason why is because your lit review is supposed to situate your work in the larger literature, but it's difficult to do that until you know what kind of results you've gotten. So I work out what my results are and then I tell my story in the lit review, alluding to what's missing and how my work fills that gap.

 

I organize literature using a citation manager - I use Sente (a Mac only program), but a lot of people swear by Mendeley and Zotero (I also like Zotero). Mendeley and Zotero are free; Sente costs some money. All of these programs allow you to organize and tag your literature by subject so you know where to look for related articles. Also, I outline my literature reviews before I write them. That gives me a roadmap for what I'm going to write and how I'm going to organize it. It's a lot easier than starting from a blank page.

 

As for figuring out gaps - that's easier for me. When I read papers, they raise more questions than they answer. The discussion also usually suggests areas for future research. Then you do another search to see if anyone has answered some of those questions. It takes some time - it took me several years in graduate school to feel like I knew enough about the literature to know where the gaps were and where I wanted to establish my research agenda

 

I write mine the same way too. It's one of the last things written since even if you have a good grasp of the literature, it's hard to know what areas to focus on or how to frame the story until you know how your results fit into the literature.

 

Also, just before submission, sometimes our group sends our paper to key people that we know would be interested / also work on the same things. Often, they will point out interesting angles/references to the literature that we might have missed (especially if it's their own work, or their group's work). And, most people in my field post papers to a pre-print server upon submission so that people who feel that their own work has been missed can point it out and they can fix it in the first round of edits. 

[GeoDUDE!]

just to put numbers in a category, my 8 page double spaced prospectus (which is less than a proposal in my dept) cited ~50 articles. I probably read 300 to get to that point.