Notes from Class Discussions

Advice on writing Research Reports | Advice on Literature Reviews | How to Document Data

Advice on writing research reports

Audience
Pretend that you did the work of collecting the data as well as analyzing it. Your work might be a first draft of a journal article that would have many authors. Imagine that the other people who worked on the project are your co-authors, so you can say "We measured..." You can assume that your readers are knowledgeable about the subject area. You should not assume that they have any knowledge of your specific project.

Introduction and References
It is not very useful to your readers to cite sources that they cannot obtain. If possible, cite peer-reviewed articles. Don't cite more than three articles to support a single claim. A recent paper will likely contain references to the earlier work, if a reader wants to learn more. Just because some other author cited a paper doesn't mean that it says what they think it says!

Methods
What information should be reported, and what information is unnecessary? You should report enough information for another researcher to be able to replicate your study and for hte reader to be able to understand your study.

Results

Don't forget the obvious. Are your numbers large or small? Consistent or highly variable?

Don't describe the figures ("Figure 3 shows x vs. y.") I call this kind of sentence a throw-away sentence. You can omit it. Instead, describe patterns you want us to see ("X is high when y…") and cite the figures parenthetically "(Figure 3)."

Don't repeat information in figures and tables. Do tell readers, in your text, what you want them to observe in your figures and tables. So, don't say "x was 7.43," if this number is presented in your table, but do say "As expected, x was higher than y when..."

Both of you will need to address variation across sites.

Figures and Tables
You will have to choose whether your data are better represented in a table or figure (or neither). You cannot report the same information in more than one table or figure (or text). For example, if you have your four sites in a single figure, you don't also need a separate figure for each site.

In both figures and table, organize your variables to put the ones you want to compare near each other. In a table, it is easier to compare numbers in columns than rows.

Don't report meaningless digits. Standard errors can give you an indication of the accuracy of your mean. The precision of your measurement may be greater than your confidence in the mean. More commonly, computers give you lots of digits, more than you possibly need. Use judgment.

Assemble your report in the following order.
Title page
Abstract (not this time)
Text (Intro, Methods, Results, Discussion)
Acknowledgments (if any)
References
Tables
Figures captions with Figures (journals will want the captions separately)

Consult the instructions to authors if you're writing for a journal

The reference list gives exactly the papers that you cite in the paper.

Double-space your lines, because it makes it easier for reviewers to write on your paper. If reviewers will write a review, give them line numbers. Page numbers are important.

A cover page is optional.

Advice on Literature Reviews

About Finding References

Searching using keywords: start with lots of specific keywords; to find more references, reduce the number of terms or use more general terms.

Too few references can come up, for example, "low-cost hydropower"

Too many can come up, not the relevant ones, for example, "root species separation" turns up papers on linguistics. Adding "tree" didn't help.

You might find help by looking at the keywords provided by a paper of they type you want to find.

In addition to keyword searches, consider author searches, and searches of more recent papers that have cited a work you have. If you find papers from a conference or a report, you can look at other reports in the same series or from the same conference.

Contacting researchers directly is also a good idea (after you do your homework).

About using RefWorks

Import your citations into RefWorks right away; it's a pain to input it manually from the pdf.

You can save the pdfs, since you have them, and space is not an issue. Naming them so that you can find them again is important. You might want to note whether you have the pdf (Notes). RefWorks currently allows pdfs to be linked to the citations only if they are less than 5 MB.

Beware of having multiple entries for the same paper.

Folders are a good way to organize your references. It's easy to make mistakes, but you can go back to "view folders" and re-organize things.

One user can have multiple accounts. We can have a class account with multiple users.

Using RefWorks from home requires a "user group name," which is "RWSyracuse."

Question: Can we create folders within folders?

Using RefWorks to create the list of references in your paper: Creating the reference list seems to be a one-time thing, so you should probably save this version of your document under a different name, in case you want to add more later.

Assessing the paper

Read the abstract first. Sometimes you can tell right away that it's not what you're looking for.

If what you're looking for is in the abstract, you can start by assuming that the same language will be used in the text of the paper, and search on those keywords. It can be a lot faster to search a pdf than a printout!

If there are only a few paragraphs that are relevant, you can copy them into your notes section.

You can also copy references into the Notes section.

I always used to note (on my 3x5 card) where I found the citation, initially. This can be helpful for identifying "empty" citations. "Covington (1976) cited this paper to support the claim that decomposition increases after forest disturbance."

What makes a good annotated bibliography?

Your notes should describe what the paper contributes to your overall goal. Good notes will save you from going back to a reference that doesn't have what you need (in spite of the promising-sounding abstract). Or they tell you which ones you do want to go back to, and what for. Don't write an abstract of the paper—the authors already did that!

Note that what is "useful" will depend on your needs. This means that the notes for a given paper will depend on the purpose of your literature review. Someone else's notes on the same paper might not help you!

It's hard to have someone else help you search the literature, unless what you are looking for is very specific. For example, information on biodiversity in bioenergy plantations is easier to evaluate than "willow in the Northeast." Specify your goals before you start.

Reading papers in a new discipline involves learning what variables are important and what they mean (e.g. "return on investment"). You might want to capture the equations used in a particular paper, including how these variables are used (which can differ). Defining new terms is important (e.g. "emergy"). Building a glossary might be helpful.

How to Document Data

The first page of a spreadsheet is a good place to document your data.
Put the date and the name of the person providing the documentation.
You can also provide the names of other people involved in the data collection.
I have spreadsheets with a dozen entries by several different people. It could be important to know who did what and when.

A brief description of the overall project would be helpful.
For example, one of your data sets relates to the heartwood project that currently involves Andrew Mishler. The other reports soil chemistry from NAMP sites currently under study by Dusty Wood and Nicole Werner.
Column headings can be entered in a column by using "paste special" and "transpose."
Make sure to provide the units for each variable.
Provide the original measurements, even if they go into later calculations.
Provide equations for calculated variables.
Intermediate variables that are not necessary to providing the final results need not be reported. If you used a lot of columns to calculate a derived variable, you might consider replacing it with one column and a description in the documentation of the equation.

Additional documentation
For example, ring counts don't correspond to tree age when the center of the tree is too rotten to count.
QA/QC: How often were check standards run? Were duplicates run? Tissue standards?
We commonly don't report the concentrations of the standards (would these change your results?) but sometimes the number of standards. For example, on this ICP run we used a 3-point calibration curve.

"Known problems" should probably be a required section. You can say "none" for now if you don't know of any.
For example, we had a problem with drift with the ICP. Describe it.
Refrigeration: We failed to refrigerate the samples (say how long between extraction and analysis) and saw signs of microbial growth ("fuzz") in the solutions.
We don't know if we used gloves when handling filter papers; we can't remember.

The raw data (e.g. ICP output) should be in the same workbook, so it will never be hard to find if someone wants to reanalyze it (e.g. using the check standards for drift correction)

How to review your data:

Neil would sort by site and horizon.
Zach would just order his sites (how?)
Make each element a separate graph, otherwise you can't see the small numbers.
Visualization graphics.
Excel will automatically scale the y axis to your range. This will make small variations (compared to the total) look just as big as large variations. You may want to force the y axis to go to zero.

Research | Prospective students | Write/Publish Class | Research Class | Official Website

Methods What information should be reported, and what information is unnecessary? You should report enough information for another researcher to be able to replicate your study and for hte reader to be able to understand your study.