Reliability and validity
- Inaccurate tests
Hughes (1989:2) conceives that one of the reasons why the tests are not favoured is that they measure not exactly what they have to measure. The author of the paper supports the idea that it is impossible to evaluate someones true abilities by tests. An individual might be a bright student possessing a good knowledge of English, but, unfortunately, due to his/her nervousness may fail the test, or vice versa, the student might have crammed the tested material without a full comprehension of it. As a result, during the test s/he is just capable of producing what has been learnt by tremendous efforts, but not elaboration of the exact actual knowledge of the student (that, unfortunately, does not exist at all). Moreover, there could be even more disastrous case when the student has cheated and used his/her neighbours work. Apart from the above-mentioned there could be other factors that could influence an inadequate completion of the test (sleepless night, various personal and health problems, etc.)
However, very often the test itself can provoke the failure of the students to complete it. With the respect to the linguists, such as Hughes (1989) and Alderson (1996), we are able to state that there are two main causes of the test being inaccurate:
- Test content and techniques;
- Lack of reliability.
The first one means that the tests design should response to what is being tested. First, the test must content the exact material that is to be tested. Second, the activities, or techniques, used in the test should be adequate and relevant to what is being tested. This denotes they should not frustrate the learners, but, on the contrary, facilitate and help the students write the test successfully.
The next one denotes that one and the same test given at a different time must score the same points. The results should not be different because of the shift in time. For example, the test cannot be called reliable if the score gathered during the first time the test was completed by the students differs from that administered for the second time, though knowledge of the learners has not changed at all. Furthermore, reliability can fail due to the improper design of a test (unclear instructions and questions, etc.) and due to the ways it is scored. The teacher may evaluate various students differently taking different aspects into consideration (level of the students, participation, effort, and even personal preferences.) If there are two markers, then definitely there will be two different evaluations, for each marker will possess his/her own criteria of marking and evaluating one and the same work. For example, let us mention testing speaking skills. Here one of the makers will probably treat grammar as the most significant point to be evaluated, whereas the other will emphasise the fluency more. Sometimes this could lead to the arguments between the makers; nevertheless, we should never forget that still the main figure we have to deal with is the student.
Now we can come to one of the important aspects of testing validity. Concerning Hughes, every test should be reliable as well as valid. Both notions are very crucial elements of testing. However, according to Moss (1994) there can be validity without reliability, or sometimes the border between these two notions can just blur. Although, apart from those elements, a good test should be efficient as well.
According to Bynom (Forum, 2001), validity deals with what is tested and degree to which a test measures what is supposed to measure (Longman Dictionary, LTAL). For example, if we test the students writing skills giving them a composition test on Ways of Cooking, we cannot denote such test as valid, for it can be argued that it tests not our abilities to write, but the knowledge of cooking as a skill. Definitely, it is very difficult to design a proper test with a good validity, therefore, the author of the paper believes that it is very essential for the teacher to know and understand what validity really is.
Regarding Weir (1990:22), there are five types of validity:
- Construct validity;
- Content validity
- Face validity
- Wash back validity;
- Criterion-related validity.
Weir (ibid.) states that construct validity is a theoretical concept that involves other types of validity. Further, quoting Cronbach (1971), Weird writes that to construct or plan a test you should research into testees behaviour and mental organisation. It is the ground on which the test is based; it is the starting point for a constructing of test tasks. In addition, Weird displays the Kellys idea (1978) that test design requires some theory, even if it is indirect exposure to it. Moreover, being able to define the theoretical construct at the beginning of the test design, we will be able to use it when dealing with the results of the test. The author of the paper assumes that appropriately constructed at the beginning, the test will not provoke any difficulties in its administration and scoring later.
Another type of validity is content validity. Weir (ibid.) implies the idea that content validity and construct one are closely bound and sometimes even overlap with each other. Speaking about content validity, we should emphasise that it is inevitable element of a good test. What is meant is that usually duration of the classes or test time is rather limited, and if we teach a rather broad topic such as “computers”, we cannot design a test that would cover all the aspects of the following topic. Therefore, to check the students knowledge we have to choose what was taught: whether it was a specific vocabulary or various texts connected with the topic, for it is impossible to test the whole material. The teacher should not pick up tricky pieces that either were only mentioned once or were not discussed in the classroom at all, though belonging to the topic. S/he should not forget that the test is not a punishment or an opportunity for the teacher to show the students that they are less clever. Hence, we can state that content validity is closely connected with a definite item that was taught and is supposed to be tested.
Face validity, according to Weir (ibid.), is not theory or samples design. It is how the examinees and administration staff see the test: whether it is construct and content valid or not. This will definitely include debates and discussions about a test; it will involve the teachers cooperation and exchange of their ideas and experience.
Another type of validity to be discussed is wash back validity or backwash. According to Hughes (1989:1) backwash is the effect of testing on teaching and learning process. It could be both negative and positive. Hughes believes that if the test is considered to be a significant element, then preparation to it will occupy the most of the time and other teaching and learning activities will be ignored. As the author of the paper is concerned this is already a habitual situation in the schools of our country, for our teachers are faced with the centralised exams and everything they have to do is to prepare their students to them. Thus, the teacher starts concentrating purely on the material that could be encountered in the exam papers alluding to the examples taken from the past exams. Therefore, numerous interesting activities are left behind; the teachers are concerned just with the result and forget about different techniques that could be introduced and later used by their students to make the process of dealing with the exam tasks easier, such as guessing form the context, applying schemata, etc.
The problem arises here when the objectives of the course done during the study year differ from the objectives of the test. As a result we will have a negative backwash, e.g. the students were taught to write a review of a film, but during the test they are asked to write a letter of complaint. However, unfortunately, the teacher has not planned and taught that.
Often a negative backwash may be caused by inappropriate test design. Hughes further in his book speaks about multiple-choice activities that are designed to check writing skills of the students. The author of the paper is very confused by that, for it is unimaginable how writing an essay could be tested with the help of multiple choices. Testing essay the teacher first of all is interested in the students ability to apply their ideas in writing, how it has been done, what language has been used, whether the ideas are supported and discussed, etc. At this point multiple-choice technique is highly inappropriate.
Notwithstanding, according to Hughes apart form negative side of the backwash there is the positive backwash as well. It could be the creation of an entirely new course designed especially for the students to make them pass their final exams. The test given in a form of final exams imposes the teacher to re-organise the course, choose appropriate books and activities to achieve the set goal: pass the exam. Further, he emphasises the importance of partnership between teaching and testing. Teaching should meet the needs of testing. It could be understand in the following way that teaching should correspond the demands of the test. However, it is a rather complicated work, for according to the knowledge of the author of the paper the teachers in our schools are not supplied with specially designed materials that could assist them in their preparation the st