Score Stats a Statistical Analysis

Score Stats

A Statistical Analysis of ACT vs. SAT Scores of Low Income Students

Study Description

Apparent income disparities in standardized test scores have been noted in many previous studies, with the determination that the income level of a student's family -- along with other sociocultural factors -- has a major effect on their ability to achieve on standardized tests (Kohn, 2002). For tests like the ACT and the SAT, which are commonly (almost universally) used by colleges and universities in the United States as part of their admissions criteria and decision-making process, a gap in performance caused by income levels puts low-income students at a significant disadvantage for entry into four-year degree programs, which in turn limits earning potential and thus could in fact perpetuate lower income levels (Kohn, 2002). This study set out to determine if there is a significant difference in the ACT test scores of low-income students when compared to the SAT scores of the same student population, as a means of determining if test composition can mediate or make more pronounced any impacts of income standing on student performance. 30 students who completed both the ACT and the SAT tests and who matched income criteria of living at or below 150% of the poverty level were included in the study, with their total scores on both tests compared in order to determine if a significant difference exists. This would indicate that something in the test structures(s) worked to influence the impact that a low-income background has been observed to have on standardized test scores.

Statement of Hypothesis

The null hypothesis is that there will be no difference between the means. The alternative hypothesis, which is the hypothesis this study is investigating, is that there is a significant difference between the mean scores on the ACT and the SAT, indicating that test structure can determine the degree to which income impacts test scores.

Variable Description

Income level was one independent variable used to determine eligibility/inclusion for the study, with a family income of 150% of the defined poverty level the upper income limit. Subjects were randomly selected and were also polled for age and gender, though these variables were not analyzed further. Income level was not recorded past the point of inclusion, therefore figures for this data are not given; gender and age are given and a descriptive analysis was performed. Dependent variables of interest were test scores on the ACT and test scores on the SAT. These two sets of variables constitute the data points that were assessed using a t-test in order to determine if a significant difference in means exists for these two tests when taken by low income students.

Descriptive Statistics

Test subjects were 53.3% male and 46.7% female, as shown in the bar graph on the accompanying Excel spreadsheet. The test population has a mean age of 15.9 years (standard deviation=0.80). These descriptive statistics validate the randomized population of the research study, as the population is almost evenly split between males and females and the mean age is also very close to the median age (16) of the study's population (Healy, 2009). As ACT and SAT scores vary in their nominal values (a top score on the ACT test is 36, a top score on the SAT is 2400), scores were recorded as percentages of the total score to enable meaningful analysis of varying means (i.e. A score of 29 on the ACT and of 1920 on the SAT would both be recorded as 0.80) (Kaplan, 2004). Mean scores on the ACT were 0.47 (standard deviation=0.27), while mean scores on the SAT were 0.41 (standard deviation=0.31).

T-Test