Many elementary school educators, possibly because they are involved in the administration but not the construction of the test, want all their students to do well. They therefore make requests for an "easy" test. Yet, if the standardized test is designed to be an informative diagnostic tool, its strength is its ability to differentiate among students. A standardized test can provide a diagnosis of specific qualitative strengths and weaknesses in a student's educational development in relation to themselves and in relation to a group broader than their own classroom only if appropriate levels are chosen for administration. If it is an effective measurement tool, then different measures for different individuals would be expected. A scale that said everybody weighed the same would be providing useless, as well as incorrect, information. A test that was too easy would provide this same useless information. And if it is made clear that the purpose of testing is to provide information to be used in conjunction with, not in place of, teacher judgment, then educators might view standardized tests as a beneficial addition, rather than an intrusion, to their classroom instruction.

Although the primary purpose of standardized achievement tests is to help inform instruction, there are other audiences interested in the kind of information that only standardized tests provide. These audiences tend to be educational researchers that take advantage of the inherent characteristics of standardized test data in order to explore similarities and differences among groups. One of the most commonly used group distinctions is gender. One objective of this research has been to document the nature of differences between the test performance of representative samples of mates and females. More ambitious objectives of this research have been to try to understand why these differences exist.

One of the most exciting aspects of current research regarding gender is that results have been found to depend on a wide variety of factors that are not easy for researchers to control in the context of any experimental design. The implications of such context dependent factors are that research contexts need to be carefully articulated when drawing conclusions. Further, broad generalizations that do not take into account these contextual differences may not hold for all potential comparisons between the test performances of males and females. Examples of contextual factors include ethnicity, SES status, achievement level, and course taking characteristics of students included in a particular study sample. In what follows, several studies are presented that look at gender differences, including a possible interacting factor that may influence results and thus inferences regarding gender differences in achievement.

Many studies agree that there is a difference in scores by gender for certain content areas. Most educators would probably tend to believe that females perform better on language-related tests and males tend to perform better on mathematical-related tests, relying on prior research evidence and commonly reported findings that have come to be common knowledge. However, often only the mean difference between males and females is reported. This merely gives an overall indication of gender differences and may not be appropriate at all achievement levels. . .