Chapter Five

The main aim of this study was to investigate the effect of question types (essay versus multiple-choice) and question levels: Remember - an - instance (RI), ýRemember a Generality (RG), and Use a Generality(UG) on three levels of learning (RI, RG, and UG) based on Merrill’s Taxonomy . Thus, the discussion of the results of this study will take the same direction of its aim. 

Before discussing the results of the study, it is worthwhile to mention that this study was limited to "Merrill’s Taxonomy": Remember - an - Instance (RI), Remember - a - Generality (RG), Use - a - Generality (UG). The Find - a - Generality (FG) level was not used because it has low relevance to most classroom objectives, and because "Merrill" has not yet adequately specified the nature of questions for the Find - a - Generality level (See, Darwazeh, 1982). 

The results of this study were discussed according to the following items:

1. Type of questions: TOP
The results of this study revealed somewhat that essay type questions have a greater effect in facilitating learning than multiple-choice ones. Experimental groups which received essay type questions during experiment performed better than those who received multiple-choice questions on most sub, post, and the total tests (See, Table 24).

This trend of these results could be explained in that the essay type questions which asked students to write his/her answer in an open format and asked, in most cases, about general information require students to implement a deep mental process in understanding the text, hence to comprehend it more on different levels of learning. Whereas, the multiple-choice questions which asked students to check the right answer and asked, in most cases, about specific information don’t require students to implement a deep mental process in understanding the text, hence don’t lead to comprehend it on different levels of learning. It required them to remember one level - a remembrance one instead. 

This explanation were supported by Arrasmith, Sheehan & Applebaum 1984; Hammbleton & Merphy (1992); Lukhele, Thissen, & Wainer (1994); Roderick & Anderson (1968) when they found that essay questions have a higher effect on students’ achievement than multiple-choice; and, on the other hand, oppose the results of Perkins (1984) and Duchastel & Nugester (1982) which found that multiple-choice questions have a greater effect on students’ achievement than essay ones.

2. Level of questions: TOP
The results of this study revealed, on most sub-tests and the total tests, that the remember a generality (RG) level questions first, then the use a generality (UG) level questions have a greater effect in facilitating learning than the remember - an - instance (RI) question level. In other words, the experimental groups which received (RG) questions or the (UG) questions during experiment performed better than those groups who received (RI) questions on most sub post -tests and on the total tests (See, Table 25).

Post-tests	RI	RG	UG	P - Value
RIM	2.629	2.950	2.475	Sig.
RGM	2.225	2.375	(2.400)	Non-sig.
UGM	2.100	2.275	2.025	Non-sig.
RIE	1.875	1.925	1.125	Sig.
RGE	2.763	3.325	(3.525)	Non-sig.
UGE	1.050	1.450	1.350	Non-sig.
Total multiple-choice	6.950	7.600	6.900	Non-sig.
Total essay	5.688	6.700	6.000	Non-sig.
Overall learning	12.637	14.300	(12.900)	Non-sig.

This trend of these results could be explained as that (RG) level questions which require students to remember the basic and general information, hence to comprehend the main ideas in the text would help students to retrieve information not just on the remember a generality (RG) level of learning but also on the use a generality (UG) level, because the (UG) level of learning could not be reached unless the student can master the (RG) level. The (UG) level of learning from "Merrill’s" point of view means to use (apply) the learned general information in a new situation. Thus, the (RG) level is considered as a prerequisite of (UG) level. 

This explanation agreed with the results of Al-Nayef (1989), Redfield (1981), and the two experiments which were conducted by Andre et al. (1980) when they found that high level questions such as RG and UG have greater effects on students’ learning than low level questions such as RI level. On the other hand, these results disagreed with the results of Perkins et al. (1990) who found that there were significant differences for the factual questions (RI) but no significant differences between the generalization (RG) and inference (UG) levels of questions. RI question level has a greater effect on students’ learning than RG and UG question levels as they found.

These results could also be explained that giving students higher level questions such as RG and UG would help them to activate their cognitive strategies in a better way than giving them lower level question such as RI. This is why that the performance of RG, then the UG experimental groups performed better than the RI experimental group on most sub-tests and on the total tests (See, Table 25). 

This result agreed also with what Rayan (1973); Redfield (1981); Richards and Vesta (1974); Royer and Konold (1984), found in their studies that UG questions have a greater effect on student’ learning than RI ones. But, on the other hand, this result opposes to the findings of Felker and Dapra (1975); Perkins et al. (1990); Samson et al. (1987) which revealed that low - level questions (RI) had a greater effect on students’ achievement than application questions (UG).

3. Student’s ability: TOP
The results of this study revealed clearly and significantly that high student’s ability then the medium student’s ability performed better than the low student’s ability on all sub-tests and the total test (See, Table 26). 

Post-tests H. ability M. ability L. ability P-Value
RIM 2.825 2.683 2.286 Significant
RGM 2.552 2.268 1.857 Significant
UGM 2.379 2.073 1.571 Significant
RIE 1.862 1.610 1.095 Significant
RGE 3.698 3.122 2.000 Significant
UGE 1.457 1.085 1.190 Non sig.
Total Multiple 7.759 7.024 5.714 Significant
Total Essay 7.017 5.817 4.286 Significant
Overall learning 14.776 12.841 10.00 Significant

These results showed that high ability then the medium ability students performed better than low ability ones on all sub-tests and the total test. This is of course due to their high mental abilities which enable them to implement their cognitive strategies in understanding the text effectively and to comprehend the learned information easily, thus to be more effective in manipulating the text than the low ability students. This result supported by Darwazeh (1982); Memory (1983); and Sanders (1973) when they found in their studies that high ability students performed better than low ability ones on post tests measuring different levels of learning (RI, RG, UG).

In terms of interactions, results showed that there were no significant interactions between the three independent variables specially the two main variables: type and level of questions (See the results in chapter four).

4. Experimental versus control groups: TOP
Although there were no significant differences between the performance of the experimental groups compared against the performance of the control group (p=.16), but the mean scores showed that the performance of the control group, who did not receive any type of questions during the experiment was the lowest (x= 11.5) among all the groups who received different types of questions on different levels of learning (See, Table 27) 

Group RIM RGM UGM RIE RGE UGE Control
x (SD) n 12.175.3320 14.953.3320 12.553.6020 13.13.4620 13.654.0420 13.253.7420 11.54.1620

In other words, the means of the experimental groups were higher than the mean of the control group though the differences were not significant. The direction of this result was supported by Darwazeh (1996) when she found that different levels of adjunct questions during instruction had activated and facilitated different levels of learning. This means that the use of adjunct questions during instruction have a greater effect than the use of the passage only without adjunct questions. According to this study, the researcher recommends teachers and book designers to present different adjunct questions during instruction, because this will help students understand the given text and be able to answer the post-test questions effectively.

Application of the study
= = = = = = = = = = = = = = : TOP

The findings of the study will be of great importance for the following elements of the instructional process:

A. Students:
Students who are interested in questions and tests as a learning strategy will get benefit from this study as follows: Students will know that they shouldn’t complain all the time of the type of questions that the teachers use in evaluating their achievement. The results of the study revealed that the two types of questions (essay vs. multiple-choice) have different effects on learning depending on the level of learning they want to induce. Their effects will vary from promoting remember levels to application levels of learning.

B. Teachers

Teachers can benefit from this study as follows:

1 . This study should help teachers to weigh out their questions and students’ abilities and instructional process aim before they decide on the type and level of questions they use for evaluating their students’ achievement . 

2 . The results of this study will be used for educating teachers on how to use questions effectively either for teaching or for testing, and informing them that the RG questions are the basic questions if they want their students to learn not just on remember levels but also on higher levels such as application. 

3. Informing teachers that the essay type questions are better than multiple-choice ones in promoting different levels of learning. Although the effects of the two types vary from one learning situation to another, from one learning level to another, and from student’s ability level to another. 

C. Exam designers and researchers
This study should help test designers and researchers in the following :

1 . It shows test designers and researchers the importance of tests and exams in learning; exams are not only used to test students’ achievement, but they can also use as a mean for providing students with new information.

2 . Designers and examiners can benefit from this study in designing their tests. Particularly the General Secondary Examination (Tawjihi ), by considering type, level, students’ ability, and the instructional aims before putting these tests .

3. The researcher hopes that it would invite other interested researchers to build on the findings of this study to reach the ultimate goal of creating a reliable procedure for using questions effectively.

Recommendations 
= = = = == = = = = = = TOP

1.Teachers are advised to use the RG level questions frequently in order to help students to understand the intended text and then to answer the higher level of questions later on. 

2. Teachers are also advised to use the essay type questions frequently in order to promote students’ cognitive strategies deeply, hence to increase their academic achievement on all levels of learning. 

3. Researchers should 
a. conduct studies on each type and level of questions, 
b . apply further research directed to other topics matter rather than English,
c . apply further research directed to other students rather than freshman college ones, and 
d. test different levels of questions by using "Bloom’s Taxonomy" with its six levels, beside the Merrill’s taxonomy. 

4. The testing system should take into consideration the specific instructional objectives of the course to be measured in the light of the general educational taxonomies. Moreover, teachers should expose students to different types and different levels of questions with an emphasis on high cognitive level questions such as remember a generality and use a generality levels. 

5. The Ministry of Higher Education should organize continuous training programs for teachers to increase their efficiency in evaluation and test construction. 

6. The Ministry of Higher Education is strongly recommended to adopt a national testing center that would take care of all testing services including a testing bank to review the old tests and standardized them for benefit of students and interested researchers.