|
Answer to #1
The textbook includes a discussion of three types of evidence for validity.
Content validity: how well test items cover a particular trait, body of knowledge, or skill.
Predictive validity: how well a test score predicts later performance.
Construct validity: how well a test measures a particular attribute.
It is very important to determine validity because, when testing, we are measuring qualities of people that are not directly observable. We cannot directly observe a person's math ability, so we design tests to indirectly observe that trait through the use of tasks, or items, for the examinee to complete. We make inferences regarding
the person's level of that trait based upon the quality of the performance we observe, which is indicated by a score. In order to be accurate in our score-based inference making, we must collect evidence to support the assumption that the test is measuring what we intend it to measure.
Answer to #2
Reliability is defined as the similarity between two rankings of test scores obtained from the same individual. The more test scores are free of measurement error, the greater the probability that two scores from the same individual will be the same. Validity is defined as how well a test measures what it claims to measure. Reliability and validity are both critical issues in testing. In order to make meaningful interpretations of students' test scores, tests must yield consistent scores and be measuring what the user intends to measure.
Three types of reliability were discussed in the text.
Test-retest reliability: This pertains to the stability of an individual's test score when given the same test on two separate occasions. If a test does not have test-retest reliability a student may score very well on one occasion and very poorly on a
second occasion. This would make it impossible to finally determine an accurate indication of the individual's level of the characteristic being measured.
Alternate-form reliability: This pertains to the equivalence of two forms of a single test. Both forms should measure the same content using items of similar difficulty. If a test does not have this characteristic student performance would be affected by the form of the test they took when they should be able to take either form and score the same.
Split-half reliability: This pertains to the homogeneity or internal consistency of a test's items. They should all be working together to measure the same characteristic. If a test does not have this quality, some of its items will be measuring the intended
characteristic while others will be measuring something else. For example, math word problems should not have difficult vocabulary. This would result in items measuring language skills rather than math skills.
Tests must be reliable in order for them to be valid. If a test's scores are not consistent over time, or forms of a test are not equivalent, or its items measure the wrong characteristics, appropriate conclusions regarding student performance cannot be made. However, when a test is valid, meaning that it is measuring the right thing and appropriate conclusions regarding student performance can be made, then we know it must be reliable.
Answer to #3
Standardized tests have three characteristics that set them apart from all other types of tests. They are administered and scored and the results are interpreted in a standard fashion. Every student who takes a particular standardized test should have exactly the same testing experience. All forms of a particular test are scored with reference to the same criteria. Students' raw scores are all reported with reference to the same standard. The standard for interpreting scores may be the performance of a norm group, as with norm-referenced tests, or it may be an indication of the amount of the tested domain that a student has achieved, as with criterion-referenced tests. The purpose of standardized tests is to get an accurate and representative sample of some characteristic of a person, such as their level of math aptitude, intelligence, or reading comprehension. A sample of the characteristic is measured because it would be too costly, time consuming, and cumbersome to measure every aspect of any single quality. Educators may want to know whether a student has the potential to develop advanced capabilities in some particular area of study. A standardized aptitude test may be given to a student to determine their potential to achieve, given proper instruction. Standardized achievement tests may be given to students to find out how well they have learned a particular body of knowledge or skill.
The results of standardized tests may be used in a variety of ways by educators to decrease the differences in students' academic performances. Standardized test results may be used to identify particular strengths and weaknesses that students' possess. Once these strengths and weaknesses are identified, teachers may plan instruction to fill gaps in the curriculum or to provide remediation. Furthermore, these types of test scores provide part of the evidence necessary to select students for special programs, such as special education classes or gifted programs.
|
Quick Reply
