Annual Editions: Research Methods 01/02

Rating:
Summary: Very Good Supplement to Course
Review: The "Annual Editions" series from McGraw-Hill/Dushkin now includes seventy volumes covering a wide range of liberal arts subjects . Each volume presents a varied selection of journal articles which are intended to be used as supplemental reading with the course textbook A Web guide is given for each article to facilitate further exploration of each topic.
The editors of this particular volume are faculty from Indiana University of Pennsylvania with impressive research credentials in educational research .Moreover, each is an experienced instructor of research methods courses.
They have selected "concise, practical articles that could provide insight into the wide range of strategies that researchers can use." The authors state that their book is "... so varied in content and level of difficulty that it could simultaneously meet the demands of different audiences ranging from sophisticated undergraduates to graduate students."
The 32 articles found in this volume have been divided into 7 sections: the nature of research, research standards/ethics, theoretical bases of research, research means/methods, categories of research, reporting research, improving research.
In the first section [ the nature of research ], Mary M Kennedy discusses "The Connection Between Research and Practice " and the 4 hypotheses of why there is a "perceived lack of connection between research and practice." The fourth hypothesis concerns the simultaneous "Stability (and Instability) of the Education System" in which she comments on American textbooks :

The lack of coherent direction is particularly apparent in our textbooks. Both the Second International Mathematics Study [McKnight et al., 1989 ] and the Third International Mathematics and Sciences Study [ Schmidt et al., 1996 ] found American textbooks to be more fragmented and superficial than texts in most other countries. They were longer, covered more topics, and devoted more space to review and repetition, so that individual topics were repeated often but were treated with little depth. Similar conclusions have been drawn from examinations of mathematics curricula at other grade levels [ e.g., Porter , 1989 ] and from examinations of texts in other subjects [ Brophy, 1990; Gagnon , 1987 ]. Tyson-Bernstein and Woodward (1986) concluded from their review that textbook publishers, in an effort to please everyone, try to include as many topics as possible rather than taking the time to develop a few central ideas. [ p. 25]

This eye-opening paper was originally presented at the annual meeting of the American Educational Research Association in New York, April 1996.
In the second section [standards/ethics ], Carol A. Heintzelman discusses "Human Subjects and Informed Consent: The Legacy of the Tuskegee Syphilis Study" This concise paper on the "longest non-therapeutic experiment conducted on human subjects in medical history" should be required reading for any course devoted to research methods.
An interesting paper by Duane A. Lundervold & Marilyn F. Belwood in the third section [theoretical bases of research] is entitled "The Best Kept Secret in Counseling: Single Case (N=1) Experimental Designs" .It introduces the single subject experiment as a valid and important tool for the investigator. The problem of "auto correlation" arises in the visual analysis of ( n = 1 ) data:
Drawing valid conclusions based on visual analysis of graphed data are hampered by variability in the data Furlong & Wampold, 1982; Matyas & Greenwood, 1990; Wampold & Furlong, 1981 ].Statistically significant auto correlation in the data increases the likelihood of Type II errors occurring when visual inspection is the sole criterion used to judge intervention outcomes [ Matyas & Greenwood, 1990 ] Auto correlation , in general, refers to serial correlation or a correlation between data points "n" steps apart, or a lack of independence among observations.In the context of single-case design, the issue of auto correlation is specific to a correlation among the residuals or error terms of the model used in the analysis. [ p. 11 ]

Section five [categories of research ] contains a study by Matt Wilkerson & Mary R. Olson on "Misconceptions about Sample Size, Statistical Significance, and Treatment Effect" that would be a valuable supplement for any statistics course.

In section six [ reporting research ] , Prof. James P. Shaver discusses "Chance & Nonsense", an entertaining and instructive Two-part "Conversation About Interpreting Tests of Statistical Significance".
Like the most selections in this volume, Shaver's fictional dialogue offers many insights and makes for a rewarding investment of reading time.
However, some of these papers should be eliminated for either irrelevance or poor writing. The excerpt from E.O.Wilson's "Back from Chaos" is irrelevant. Putney, Green, Dixon and Kelly's article on qualitative research methodology is poorly written.Marienau's article on self-assessment at work , Chang's article on Quantitative Attitude Questionnaire, and Page's article on Future Directions in Qualitative Research could all be profitably replaced. Tanner's article on the Social Consequences of Bad Research is too biased to be taken seriously.