Research literature on discourse processing often emphasizes structural importance and text-based interest as two variables that affect learning. Structural importance has been defined by Meyer and Rice (1984) as the way "ideas in a text are interrelated to convey a message to the reader. Some of the ideas in the text are of central importance to the author's message, while others are of less importance. Thus, text structure specifies the logical connections among ideas as well as a subordination of some ideas to others" (p. 319). This concept of structural importance relates to Kintsch and van Dijk's (1978) presentation of a two-level model of discourse processing with macrostructure representing the gist of the text and microstructure representing the local propositions. There is ample research literature on structural importance indicating that readers use the superordinate or macrostructure as a framework for encoding and retrieving (Brown & Smiley, 1977, 1978; Meyer, 1979).

On the other hand, text-based interest has been termed the "neglected variable" in the research literature on discourse process (Hidi & Baird, 1986, p. 179). As defined by Hidi and Baird (1986), text-based interest refers to the interestingness of materials across subjects, without regard to individual preferences. This concept of text-based interest relates to Kintsch's (1980) distinction between cognitive interest and emotional interest. Cognitive interest includes the structural cohesion of a passage and is influenced by the reader's background knowledge such that interest increases as more is known and diminishes if nothing new can be learned from the passage. Emotional interest occurs when events have a direct emotional impact. Death, danger, power, money, sex, romance, and disease are some inherently interesting topics (Schank, 1979).

In many cases, personalized anecdotes and descriptive elaborations have been added to textbooks to increase interest (Hidi, Baird, & Hildyard, 1982; Pearson, Gallagher, Goudvis, & Johnson, 1981). Previous research reports that readers often find such material interesting but unimportant (Hidi et al., 1982). "Seductive details," the term coined in the literature for this type of material (Garner, Gillingham, & White, 1989) has been found to both enhance and interfere with learning (Hidi & Baird, 1988). Graves, et al. (l988) determined that a text was recalled better with seductive details and narrative elaborations. These findings conflict with Hidi and Baird (1988) who found that strategies for creating text-based interest did not facilitate learning of main ideas.

Since both factors affect learning, and given the controversy concerning text-based interest, it is worthwhile to investigate readers' assessment of what is important and what is interesting in text. In this experiment, subjects rated paragraphs of a mixed text (see Appendix A) for both interest and importance. A mixed text contains biographical content with exposition (Pearson et al., 1981). These ratings were compared to expert readers' ratings of interest and importance for paragraphs of the same passage from a previous study (Jetton & Alexander, 1997). The expert readers were doctoral students in reading with expertise in analyzing texts according to main idea versus details. In other words, the expert readers were trained to rate main ideas high in importance and details low in importance. The expert readers were also requested to rate each paragraph in terms of interest according to the information in the text that was particularly intriguing to them.

In keeping with the results of previous studies (Garner et al., 1989; Hidi & Baird, 1986, 1988; Wade & Adams, 1990), the ratings by these expert readers in the study by Jetton and Alexander (1997) were found to diverge for several paragraphs. In a passage about Stephen Hawking, the noted physicist, paragraphs about Hawking's goal to resolve the puzzle of Grand Unification Theory (paragraph 2), his study of black holes (paragraph 3), and his discovery that black holes emit a stream of electrons (paragraph 4) were rated as moderate in interest but high in importance. On the other hand, paragraphs about Hawking's serious disease (paragraph 5), the change in his life following his marriage (paragraph 6), and how his illness had affected his work (paragraph 7) were rated as low in importance but moderate to high in interest. Thus, the expert readers rated the paragraphs in a manner that identified paragraphs 2, 3, and 4 as main idea paragraphs while paragraphs 4, 5, and 7 were identified as "seductive detail" paragraphs (Garner, Alexander, Gillingham, Kulikowich, and Brown, 1991). Paragraph 1 about Hawking's attempt to discover the connection between two branches of physics was rated as low in interest and moderate for importance and is classified as somewhere between boring trivia and important factual information (Wade, Shraw, Buxton, & Hayes, 1993).

Because Hidi et al. (1982) found that ratings of importance sharply contrast ratings of interest for mixed text, we expected that ratings of importance would also contrast with ratings of interest. In other words, it was expected that ratings by the readers in this study would diverge on interest and importance and be similar to the ratings by the trained, expert readers of a previous study (Jetton & Alexander, 1997).

This study also investigated the effect of reading ability on assessment of interest and importance of text. Since previous research indicates that skilled readers have greater ability than poor readers to distinguish important from unimportant information (Brown & Smiley, 1977, 1978; Smiley, Oakley, Worthen, Campione, & Brown, 1977), it was expected that ratings of main idea paragraphs by good readers would more often agree with the ratings by the trained, expert readers (moderate to high in importance and low in interest) than would ratings of poor readers. Since there is evidence that adding narrative elaborations to texts may interfere with learning of main ideas (Hidi & Baird, 1988), college readers' assessment of interest and importance for "seductive details" (or information considered by expert readers to be moderate to high in interest but low in importance) is worth investigating. Therefore, this study sought to answer the question of whether ratings by good readers on paragraphs with seductive details would be in agreement with the expert readers more often than ratings by poor readers.