Lura E. Joseph
Print version (Adobe Reader required)
Interdisciplinary research offers increasing information challenges for researchers and scholars as well as for librarians. Quaternary research is an example of a highly interdisciplinary area incorporating research ranging from geochemistry and microbiology to planetary science. This study compares retrieval performance of eleven online indexes that can be used for Quaternary research, and discusses three others. Recall, precision, and overlap and uniqueness were analyzed using search results (12,896 records) from the eleven databases for the publication year 2000. A broad search strategy was used in order to recover most of the relevant information from the databases for the whole discipline for one publication year in order to avoid problems encountered when using sampling and example searches. Implications for interdisciplinary research in general are discussed, and federated searching is suggested.
Given the exponential increase of information, staying current in a particular discipline, verifying a particular citation, or conducting an exhaustive search for information on a particular topic can be daunting to researchers and scholars. Helping the researchers and scholars with their information quests can be taxing for librarians and information specialists. The switch from print to online indexes is both a blessing and a curse. On one hand, the online indexes uncover much more information, but on the other, sorting through the mountain of information can be frustrating and time consuming. Quaternary research is an example of a highly interdisciplinary area whose researchers, scholars, and supporting librarians are faced with these problems.
Quaternary Research: Its Nature and Importance
Quaternary research is the study of the Quaternary, which is the period of time that spans approximately the last 2.6 million years of the Earth’s geologic history.1 The Quaternary geologic time period includes the Pleistocene, sometimes known as the Ice Age, and the Holocene, which is the geologic epoch in which we live.2 The Pleistocene covers the time period spanning approximately two million years to ten thousand years ago, and the Holocene includes the period spanning approximately ten thousand years ago to the present. Scientists currently disagree regarding the exact boundaries and dates for the beginning of the Quaternary, and whether it should be retained as a formal chronostratigraphic unit (a body of rock officially recognized as a unit based on the age of its boundaries).3 In fact, the most recent International Commission on Stratigraphy removed the term “Quaternary” from the International Geologic Time Scale and included that interval of time in the Neogene Period.4 The terminology is still being hotly debated among the geologic community, but no matter what the time period is called, or when the exact initial boundary is set, that geologic period of time will undoubtedly remain an extremely important area of research.
The study of the Quaternary time period is particularly interdisciplinary.5 Figure 1 shows many of the areas of specialty within Quaternary research. Quaternary researchers study ice cores; ocean sediments; ocean circulation; lake sediments; cycles of the earth and sun; atmosphere; fossils and modern plants and animals, including invertebrates, vertebrates, insects, and pollen; and other subjects. Current topics of importance include atmospheric, ocean, and terrestrial interactions and the building of testable computer models; for example, models that can be used to predict climate change, sea level rise, or ocean circulation patterns.
During the Quaternary, the Earth has experienced frequent, extreme, and often abrupt climate and environmental changes, including the advance and retreat of continental and mountain glaciers.6 These climate and environmental changes have been accompanied by global changes in plant and animal communities and include numerous extinctions and the evolution and dispersal of humans. No matter what the causes, Earth’s climate currently appears to be warming, and significant and rapid environmental changes could challenge modern civilization. Current major goals of Quaternary research include documenting past climate patterns at various time scales in order to understand current trends and to predict future climate patterns and environmental effects.
In 1928, a group of scientists who were conducting interdisciplinary research into environmental changes that occurred during the glacial ages founded the International Union for Quaternary Research (INQUA).7 International congresses are held every four or five years, with attendance as high as one thousand participants. There are more than thirty-five member countries worldwide that also hold regional and local meetings. Out of a total membership of 18,445 within fifteen divisions, the Quaternary Geology and Geomorphology Division of the Geological Society of America (GSA) had 2,695 members (14.6 percent) in 2004, further demonstrating the importance of Quaternary research. There are a number of excellent peer-reviewed journals devoted to Quaternary research, and many Quaternary research articles appear in other highly respected general and specialized scientific journals. Other publications include newsletters and monographic series. See Associations and Information Resources for Quaternary Research for further information about the various Quaternary research societies and publications.
The Research Question
This study compared the retrieval performance of eleven online indexes that can be used to find information pertaining to Quaternary research. The main purpose was to determine which databases contain the largest number of relevant references (recall and precision), and how much overlap exists between the databases. This information can be used to help researchers and librarians determine which databases are most appropriate for Quaternary research, which might safely be ignored (and possibly cancelled by librarians faced with budget cuts), and which should be included in metasearch (federated search) tools. Results also can be used to indicate which databases could best substitute for others that might not be available, and how much information would be missed if a particular database was not searched.
Literature Review
Research focusing on interdisciplinary, cross-disciplinary, and multidisciplinary information has continued in recent years. Ackerson discussed the challenges of research and teaching in cross-disciplinary areas, focusing on engineering, and including obstacles to intellectual and physical access to information.8
Other studies related to interdisciplinary research include database comparisons. These comparisons generally fall into two categories: descriptive comparisons, such as the number of journals indexed and the size, range, and subjects covered; and performance measures, such as recall, precision, and novelty versus overlap. Starr reviewed the literature and recommended that future database comparisons should emphasize analytic research and study reasons for search results differences rather than concentrating on descriptive comparisons or performance measures.9 Her criticism of the study of performance measures was that most results are based on a small number of sample searches and the studies do not use random samples. Also, the studies typically use only a few professionals as searchers, and therefore results cannot be generalized to other user groups. This study avoids those problems because it looks at the results of a search strategy encompassing the entire interdisciplinary area rather than sample searches within the area, and the search terms used are not unique to any particular user group within the discipline. The reasons for observed differences in search results between the databases, such as database differences, are not the focus of this study.
In the past ten years, database comparison research has continued. For example, a comparison of two business literature databases by Rey and Fereres resulted in the suggestion of simultaneously searching both databases.10 McCain used a database filtering approach by combining citation, indexing, and productivity analysis to identify core and important noncore journals supporting multidisciplinary research and development in the area of biotechnology.11 Hooper-Lane and Atkins examined the journal coverage, uniqueness versus overlap, and currency of three multidisciplinary databases in relation to marine science, and concluded that the multidisciplinary databases complemented the subject-specific databases and significantly improved recall and currency of returned results.12
Cavanagh compared the retrieval performance of five multidisciplinary tables of contents (TOC) databases with three specialized biological databases in terms of precision and recall.13 The TOC databases had higher precision but lower recall than subject-specific databases. The chosen databases, as a group, identified only 75 percent of the articles known from independent sources.
Brown, Edwards, and LaSee-Willemssen conducted a “deep analysis of indexing” for two education indexes.14 They found that one of the indexes covered twice the number of journals but only half the number of articles. Based on indexing policies, even so-called cover-to-cover indexing may be incomplete. McDonald, Taylor, and Adams analyzed coverage and overlap of databases indexing psychiatry journals at the journal rather than article level.15 They identified 213 abstracting and indexing services and 977 psychiatry journals. Four databases accounted for 90 percent of all the psychiatry journals that were indexed; more than four hundred journals were not indexed at all. Thirty-five percent of the indexed journals were included in only one of the top four databases, emphasizing the need to search multiple databases. Sutton and Foulke compared eight databases (three subject-specific, two multidisciplinary, and three general academic) at the journal level for coverage of literature related to anthropology.16 They found that coverage was not comprehensive, even by subject-specific databases.
Read and Smith compared three databases for library and information science using a subject profile technique, and found maximum overlap was only 21 percent, indicating the need to search more than one database if comprehensive results are desired.17 Jatkevicius compared BIOSIS Previews, BasicBIOSIS, and Cambridge Scientific Abstracts’ Biological Sciences Collection using the same forty-two searches for each database and analyzing recall rather than precision. BIOSIS Previews retrieved more than 67 percent of overall hits from the forty-two searches and was judged best for graduate biology students and faculty.18
Brettle and Long compared retrieval from six databases related to rehabilitation of people with severe mental illness, and found that 42 percent of the papers were found only in one database, signifying the importance of searching multiple databases.19 Tellman compared retrieval of information of interest to historians from two online European journal indexes and found that significant numbers of articles would be missed if using only one of the indexes.20
Janke compared journal coverage and currency of indexing for Current Contents Connect (CCC) and PubMed for the subjects of agriculture, biology and environmental studies, clinical medicine, and life sciences, and found that, of the 801 journals jointly indexed by CCC and PubMed, for 15.5 percent CCC was less current, for 29.6 percent CCC was more current, and for 53.7 percent currency of indexing was the same.21 Kristick compared Meteorological and Geoastrophysical Abstracts with eight other databases at the source title list level, and found the highest level of overlap was with Environmental Sciences and Pollution Management (ESPM)(59 percent), followed by GeoRef (53 percent). The combination of ESPM, GeoRef, and Chemical Abstracts yielded an overlap of 80.3 percent.22 Walters and Wilder evaluated the effectiveness of twelve databases that index the literature pertaining to the multidisciplinary field of “later-life migration.”23 An article level analysis was used. They found that four multidisciplinary databases each provide better index coverage than any of the specialized subject indexes. There was a relatively low degree of overlap among the twelve databases, therefore multiple databases should be searched.
The research overwhelmingly indicates the inadequacy of using a single, or even several, databases to find comprehensive information, especially in interdisciplinary areas. Bar and Finkler advocated using a multidatabase, multidisciplinary search approach using databases from a single vendor.24 Today it is possible to use the same multidatabase, multidisciplinary approach using multiple vendors. This procedure of using metasearch (also known as federated or broadcast search) technologies has been described by Mischo and Schlembach, and Luther; National Information Standards Organization (NISO) has developed standards for these tools.25 When creating metasearch tools, it is useful to know which databases to include for particular areas. This type of study can indicate which databases are most important to include. If federated searching is not possible, then the information can be used by researchers to select the most useful databases to search and by librarians to determine which databases to purchase or retain.