Almost all educational research focus on identifying predictors of academic achievement. This is so because success in its simplest form has always been defined as how well an individual does at school. Although assessment of academic achievement has evolved over the years, the fundamental principle behind what it is, how it is expressed, and why it is used remain unchanged.
Limitation of data
Researchers use school grades in various forms (e.g., letter grade, pass-fail, GPA and CGPA, percentage, percentile ranks, etc.) as primary data to test a number of hypotheses and statistical prediction models. Examples of hypotheses would be, “There is a significant difference in academic achievement of students studying at private and public schools,” and “Academic self-perception, goal-valuation, attitude toward teachers and classes, attitude toward school and motivation predict academic achievement.”
A close examination of these hypotheses indicates that both heavily rely on obtaining valid data on academic achievement of students, which is the dependent (or outcome) variable. To test these hypotheses correctly and make sensible use of the findings thereafter, a researcher has to make sure that data collected on academic achievement truly reflects students’ actual achievement levels. Otherwise, the outcome of the study is just plain useless.
From a research perspective, data is valid when it is collected from individuals who undergo similar experiences and possess as many comparable characteristics as possible, i.e., if a researcher wants to study the effect of cooperative learning on third graders’ end-of-unit performance in science, he must make sure that all the participants in both experimental and control groups are exposed to the same unit of instruction, are of the same age, come from similar socio-economic background, have no additional tutorial after school, do not watch educational videos that would enhance their understanding of the unit, etc.
If these are not kept constant, then the eventual differences in the end-of-unit science test scores cannot be attributed to the use of cooperative learning strategies alone.
In other words, when a researcher is unable to correlate academic achievement with a variable of his interest with certainty, he is allowing an uncontrollable number of other factors to become potential predictors of achievement. These other factors are known as extraneous variables. In reality, no degree of caution could be taken to minimize or remove extraneous variables. This is an inherent weakness in all educational research.
The first hypothesis cannot be tested using data collected from existing school records (e.g., previous semester’s or academic year’s achievement records). This is so because there are huge differences in assessment practices in private and public schools. Hence if one discovers a significant difference in achievement between students studying in private and public settings, it is highly possible that the difference is not due to actual difference in how well students perform in these two different settings. Rather, the difference may be due to dissimilarities in assessment practices (one would assume that private school teachers would use more authentic assessment compared to their public school counterparts).
The second hypothesis attempts to find out which of the five psychological factors significantly predict academic achievement. However, differences in scores on all five variables could reflect the extent to which students of differing language skills understood and interpreted the items. Hence, instead of measuring psychological characteristics as truly experienced by students, the researcher has systematically measured their language abilities, reducing internal and external validity of the study.
Teachers must be careful how they use information derived from educational research. Instead of focusing on findings, conclusions and recommendations, they must pay extra attention to the overall soundness of research (how the research was done, how and what type of data was collected, etc.), before determining its usefulness. This explains why research design is a core subject in all teacher training programs.