Richard N. Landers, Ph.D.
Assistant Professor in Industrial/Organizational Psychology
Primary Investigator of TNTLab
Old Dominion University
The Research Explicator for oNline Databases (TREND) Tool
The TREND software is designed to process data from online databases and output several lists of summary information for a quick snapshot of important trends and authors in any particular research literature.
It outputs the following:
- List of authors published in the research domain and their number of contributions
- Name of publications appearing in the research domain and the number of articles in the target literature contained within each
- All keywords assigned to publications in the target literature and the number of times they appear
- All areas of study assigned to publications/articles in the target literature and the number of times they appear
- All publication years and the number of publications/articles released in each year
- Considering the given literature, which publications/articles are most often cited by that literature and how many times those citations appear.
Download TREND v1.21 Basic for Windows XP (Released April 17, 2007)
For more feature-rich versions of this software, installation instructions, and usage guidelines, see below.
PLEASE NOTE: This software is NOT Vista-compatible.
Citing This Software:
If you plan to publish research data based on the use of this software, following is the approximate citation in APA format:
Landers, R. N. (2007). The Research Explicator for oNline Databases (TREND) Tool v1.21 [Computer software]. Retrieved March 12, 2010, from http://rlanders.net/explicator.html
The TREND software is limited to the contents of the database that is being examined and relies on text-processing algorithms to determine these lists. If you are forming a definitive list based on this software (e.g. if you plan to publish the lists you produce) you must visually check the data for mis-citations undetectable by the software. For example, consider the following citations:
Baron, R. M. & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality & Social Psychology, 51, 1173-1182.
Baron, R. M. & Kenny, D. A. (1986). The mediator-moderator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality & Social Psychology, 51, 1173-1182.
If this article is mis-cited as the second of these, then there is no way for the TREND software to automatically detect the error in the title, and that they are referring to identital articles. It is the researcher's responsibility to follow up on this information if needed.
To install the Basic version of the TREND software, you will only need to unzip the download into a directory of your choice, and run the program within. If you are unable to launch the program, you may also need to install the .NET Framework from Microsoft from this location.
Each version of the software adds export functionality, but requires the installation of additional 3rd party software. See the table below to identify the version best for you.
MySQL export requires access to a MySQL database with table creation and write privileges, although that database need not be installed locally. Be aware, however, that 1 INSERT query will be run for each distinct entry in the produced database, which could means an open connection to MySQL may need to be kept open for a substantial length of time. This is by far the most versatile export option. Excel export is limited in that Excel only supports a maximum of 65,535 rows. If a larger number of distinct entries exist than this number, Excel (and thus TREND) will crash in this export mode. Other export modes do not suffer from this problem.
Currently, the software only supports the processing of citation data outputted in BRS/Tagged format, an option in OVID's implementation of PsycINFO, although any data in this format can be processed. Records in this format appears like the following:
AU Last, First M.; Last, First M.
SO Journal of Something Important
PT Journal; Peer Reviewed Journal
RF Reference 1
To get a suitable data file from OVID, first run a search for the literature that you wish to examine. Next, in the Results Manager, choose All in this set, Complete Reference, BRS/Tagged and click Save. On the next screen, choose Windows Style Linefeeds and save the file somewhere you can find it later. Open TREND and use that downloaded file as your data source.
TREND must make a number of assumptions about your data in order to make judgments regarding the equivalence of pieces of data. A list of those assumptions appears here (if an output type is not listed, no special handling is used, and entries are compared letter by letter):
- Authors: Any word 2 characters in length or less is stripped, to eliminate middle initials, as some author lists include initials and some do not. This can be set in the menu options if you expect authors with identical first and last names to appear in your search.
- Keywords: Indication of primary keyword (indicated by an asterisk) stripped from keyword list.
- Citations: Any word 3 characters in length or less is stripped and all numbers are removed, as well as any text after the letters "http://". This is done to eliminate initials, incorrect citation of short words (i.e. of, an, the, etc), and mis-citations of page numbers or volume numbers. Each reference string is also split by period (.), and any text after and including a colon (:) is eliminated, in order to remove subtitles. This assumtion can be eliminated by setting a menu option.
Tips and Tricks, Part 1:
You may find that your database software limits the number of records that you can download simultaneously, which makes the results from TREND inconclusive. For example, you may wish to process a 4523 record datafile from OVID's implementation of PsycINFO, while OVID only allows you to download 1000 records at a time. To overcome this, you can split up the datafile into multiple segments and then re-combine them manually. As an example, assuming the search for keyword.mp came up with 4523 results, you could conduct the following list of searches:
from 1 keep 1-1000
from 1 keep 1001-2000
from 1 keep 2001-3000
from 1 keep 3001-4000
from 1 keep 4001-4523
You can then use the standard procedures described in Usage Guidelines to download each segment individually. After this is done, concatenate all of the segemented files into one master file. In Windows, you can do this easily by dropping to the command prompt (Start Menu > Run... > cmd.exe), navigating to the directory where the files you downloaded are, and entering the following command, all on one line:
type cites.txt cites(2).txt > master_citations_list.txt
In this example, I have combined two downloaded citation files (named cites.txt and cites(2).txt) into one file called master_citations_list.txt. Substitute your own names as needed. A GUI-based concatenation tool is also planned for future versions of this software.
Tips and Tricks, Part 2:
You may find that the literature you are trying to summarize has included some articles that you don't want to include. For example, Personnel Psychology, one of the top-tier journals in I-O psychology, routinely includes reviews of books published relevant to our field. In a review of the literature, you might know that such a segment of articles exists and that you do not wish to include them. Unfortunately, this cannot be done within TREND. It can, however, be done in OVID. If you were to examine each review article in Personnel Psychology, you would notice that the abstract begins with the phrase, "reviews the book." Thus, we could form a series of queries like the following to ignore those articles in our examination:
1> selection.mp and "personnel psychology".jn
2> "personnel psychology".jn and "reviews the book".ab
3) 1 not 2
The results of search #3 would thus include the results of the first search but ignore those articles with the phrase "reviews the book" in the abstract. If you plan to publish any results using the TREND software, I strongly encourage you to thoughtfully consider what kinds of publications appear in your search to avoid inflated estimates due to quirks of the literature like these.
Please don't hesitate to e-mail me with any questions.