Text Organization

We examined two indices of text organization: one involved rating and the other was obtained using Criterion. PERMANOVA, with occasion, task, ELP group, and student (within ELP) as predictors and the two indices of text structure as dependent variables, detected significant main effects for occasion (F(l, 275) = 8.72,p = .0001, rf = .19), task (F(l,275) = 28.27,p = .0001, ip = .44), ELP group (F(3, 275) = 2.89, p = .0007, rf- = .22), student (within ELP) (F(81, 275)= 1.51,

TABLE 6.6 Descriptive Statistics for Total Text Organization Ratings for Full Sample

ELP	Occasion	Before			After			Total
ELP	Task	N	M	SD	N	M	SD	N	M	SD
VL	Independent	12	2.19	.59	9	2.96	.46	21	2.52	.66
	Integrated	9	2.15	.62	10	2.53	.61	19	2.35	.63
	Total	21	2.17	.59	19	2.74	.58	40	2.44	.64
L	Independent	25	2.70	.60	16	3.22	.61	41	2.90	.65
	Integrated	28	2.53	.51	21	3.15	.66	49	2.80	.65
	Total	53	2.61	.55	37	3.18	.63	90	2.85	.65
M	Independent	26	2.97	.46	17	3.17	.56	43	3.05	.50
	Integrated	26	2.68	.43	16	3.18	.62	42	2.87	.56
	Total	52	2.82	.46	33	3.17	.58	85	2.96	.54
H	Independent	17	3.14	.38	13	3.41	.44	30	3.26	.42
	Integrated	17	3.12	.46	14	3.25	.43	31	3.18	.44
	Total	34	3.13	.41	27	3.33	.44	61	3.22	.43
Total	Independent	80	2.80	.59	55	3.21	.54	135	2.97	.60
	Integrated	80	2.66	.56	61	3.08	.63	141	2.84	.62
	Total	160	2.73	.57	116	3.14	.59	276	2.90	.61

p = .0014, l']² = .74), and occasion by task interaction (F(l, 275) = 3.46,p = .0129, r)² = .07) for the full sample. Similar results were obtained for the longitudinal sample.

Text Organization Ratings : Table 6.6 reports descriptive statistics for text organization ratings across test occasions, tasks, and ELP groups. For the full sample, ANOVA detected significant effects for occasion (F(l, 275) = 21.03, p = .0001, 1p = .60),ELP group (F(3,275) = 13.98,p = .0001, rf = .72),student (withinELP) (F(81,275) = 3.13,p = .0001, rf = .85), occasion by student interaction (F(59,275) = 3.41, p = .0001, ff = .82), and task by student interaction (F(76, 275)= 1.98, p = .0086, rf = .77). For the longitudinal sample, there were significant effects for occasion (F(l, 191) = 51.96, p = .0001, i]² = .69), ELP group (F(3, 191) = 14.23, p = .0001, ff = .66), student (within ELP) (F(44, 191) = 2.01, p = .0001, //' = .67), occasion by ELP interaction (F(44, 191) = 3.96, p = .0136, ff = .34), and task by student interaction (F(44, 191)= 2.37, p = .0022, //² = .70). None of the other main and interaction effects was significant at p < .01. As Table 6.6 shows, participants had significantly higher text organization ratings (M = 3.14) after instruction than they did before instruction (M = 2.73). In addition, participants with higher ELP had higher text organization ratings than did those with lower ELP. Except for the difference between the medium and low ELP groups, all pair-wise comparisons across ELP groups were statistically significant at p < .01.

FIGURE 6.3 Text Organization Ratings by Occasion and ELP for Longitudinal Sample

As for the occasion by ELP group interaction for the longitudinal sample, followup analyses indicated that text organization ratings increased significantly after instruction for each ELP group, except for the high ELP group (see Figure 6.3). In addition, both before and after instruction, the very low group had significantly lower text organization ratings than did each of the other ELP groups.

Text Organization: Criterion was used to examine whether each essay included each of three discourse elements (introduction, thesis statement, and conclusion). Table 6.7 reports the percentages of essays that included each of the three discourse elements across test occasions, tasks, and ELP groups. ANOVA indicated the following:

• Introduction: There were significant effects for occasion (F(l, 275) = 11.4, p = .0016, rf = .25), task (F(l, 275) = 27.50, p = .0001, rf = .53), student (within ELP) (F(81, 275) = 2.31, p = .002, if = .81), and task by student interaction (F(76, 275) = 1.99, p = .009, rf = .77). None of the other main and interaction effects was significant at p < .01. Similar results were obtained for the longitudinal sample as well. As Table 6.7 shows, a larger number of participants (51% for full sample) included an introduction after instruction than they did before instruction (36%). In addition, more integrated essays included an introduction (59%) than independent essays (25%).
• Thesis statement: There were significant effects for task (F(l, 275) = 73.08, p = .0001, f]² = .67) and occasion by task interaction (F(l, 275) = 11.62, p = .0015, /]- = .21) for the full sample. For the longitudinal sample, there were no significant main or interaction effects. As Table 6.7 shows, more independent essays (91%) included a thesis statement than did integrated essays (43%). As for the occasion by task interaction, more independent essays included a thesis statement than did integrated essays both before (88% vs. 51% of essays, respectively) and after (96% vs. 33%) instruction. However, there was a significant increase (from 88% to 96%) for the independent task after instruction. The change for the integrated task (from 51% to 33%) was not statistically significant at p < .01.
• Conclusion: There were significant effects for task (F(l, 275) = 86.94, p = .0001, f]² = .65) and ELP group (F(3, 275) = 4.64, p = .0057, if = .31) for the full sample. For the longitudinal sample, there was a significant effect only for task (F(l, 191) = 55.32, p= .0001, if = .58). As Table 6.7 shows, more independent essays (78%) included a conclusion than integrated essays (32%). Additionally, for the full sample, as ELP increased, the percentage of essays with a conclusion increased. There were significant differences between the very low ELP group (28%), on one hand, and each of the low (54%), medium (61%), and high (62%) ELP groups, on the other.

TABLE 6.7 Percentage of Essays with Three Discourse Elements for Full Sample

ELP	Task	Introduction			Thesis			Conclusion
ELP	Task	Before	After	Total	Before	After	Total	Before	After	Total
VL	Independent	17	22	19	75	1.00	86	33	56	43
	Integrated	33	60	47	44	40	42	22	00	11
	Total	24	42	33	62	68	65	29	26	28
L	Independent	20	44	29	84	88	85	80	88	83
	Integrated	57	81	67	54	43	49	36	24	31
	Total	40	65	50	68	62	66	57	51	54
M	Independent	15	18	16	96	1.00	98	81	82	81
	Integrated	50	75	60	54	19	40	50	25	40
	Total	33	45	38	75	61	69	65	55	61
H	Independent	35	38	37	88	1.00	93	88	92	90
	Integrated	53	50	52	47	29	39	35	36	35
	Total	44	44	44	68	63	66	62	63	62
Total	Independent	21	31	25	88	96	91	75	82	78
	Integrated	51	69	59	51	33	43	39	23	32
	Total	36	51	42	69	63	67	57	51	54

TABLE 6.8 Descriptive Statistics for Argument Quality Ratings for Full Sample

ELP	Before			After			Total
ELP	N	M	SD	N	M	SD	N	M	SD
VL	12	1.99	.39	9	2.45	.50	21	2.18	.49
L	25	2.39	.50	16	2.64	.48	41	2.48	.50
M	26	2.62	.51	17	2.68	.41	43	2.64	.47
H	17	2.61	.40	13	3.10	.39	30	2.82	.46
Total	80	2.45	.51	55	2.73	.48	135	2.56	.52

To summarize, the following patterns emerged for text organization.Text organization ratings increased significantly with ELP and after instruction. Participants with higher ELP tended to obtain higher scores than did participants with lower ELP. Essays written after instruction obtained significantly higher scores than did those written before instruction. However, text organization ratings increased significantly after instruction for each ELP group, except for the high ELP group. In addition, the participants tended to include an introduction in their essays more often after instruction than they did before instruction.They also were more likely to include a thesis statement for the independent task after instruction than before instruction. As for task effects, the integrated essays were more likely to include an introduction than the independent essays. In contrast, the independent essays were more likely to include a thesis statement and a conclusion. Finally, participants with higher ELP were more likely to include a conclusion in their essays than were participants with lower ELP.

Argument Quality

Table 6.8 reports descriptive statistics for argument quality ratings across test occasions and ELP groups. For the full sample, ANOVA detected significant effects for occasion (F(l, 134) = 12.33, p = .0005, ff = .20) and ELP group (F(3, 134)= 6.40,p = .0006, Ip = .31). Similar results were obtained for the longitudinal sample. As Table 6.8 shows, participants obtained significantly higher argument quality ratings (M = 2.73) after instruction than they did before instruction (M = 2.45). In addition, participants with higher ELP obtained higher argument quality ratings. However, only the differences between the very low group (Af = 2.18), on the one hand, and each of the other three groups, on the other, were significant.