Kroenke K
Bull Med Libr Assoc. 1986 Jan;74(1):1-5.
In a study of book reviews published in four general medical journals over a six-month period, 480 reviews were analyzed. Twenty-five features that reviewers address when evaluating a text were identified, and the frequency of commentary for each feature was determined. The mean number of features addressed per review was 9.0. Reviews averaged 389 words, but review length did not correlate with the length or scope of the book, with the number of features addressed, nor with the reviewer's assessment of the text. Extraneous commentary by the reviewer occurred in 16% of the reviews. This editorializing appeared in lengthier reviews that addressed fewer features. Favorable reviews were far more common than unfavorable ones (88.5% vs. 11.5%). Consequently, for the fifty-five books reviewed in more than one journal, agreement regarding rating of the text was high (86%). Results of this study may provide useful guidelines for reviewers of medical texts.
在一项对四家综合医学期刊在六个月内发表的书评的研究中，分析了480篇书评。确定了审稿人在评估一篇文章时涉及的25个特征，并确定了每个特征的评论频率。每篇书评涉及的特征平均数量为9.0个。书评平均有389个单词，但书评长度与书籍的长度或范围、涉及的特征数量以及审稿人对文章的评价均无关联。16%的书评中出现了审稿人的无关评论。这种编辑评论出现在篇幅较长但涉及特征较少的书评中。好评远比差评常见（88.5%对11.5%）。因此，对于在不止一本期刊上被评论的55本书，关于文章评级的一致性很高（86%）。这项研究的结果可能为医学文本的审稿人提供有用的指导方针。
