1 of 22

High School Achievement Test Scores as a Predictor� of College Enrollment

Angel Morgan

ED 711

Summer 2019

2 of 22

INTRODUCTION

Standardized test scores are an important factor in college admissions decisions¹
3 Tests Administered to all NC Public High School Students: English 1 End of Course (EOC), Math 1 EOC and ACT
2 Optional National College Entrance Exams taken by many NC High School students: Scholastic Aptitude Test (SAT) and Advanced Placement (AP)

3 of 22

STUDY BACKGROUND

Past achievement is a good predictor of future achievement²

Test	College Level Proficiency Score
NC Math 1 and English 2 EOC³	4
ACT Composite⁴	17
SAT (UNC Freshman 2017)⁵	1130
AP⁵	3

4 of 22

HYPOTHESIS

High school achievement test scores are a good predictor of college enrollment.

PRIMARY RESEARCH QUESTION

Are district level high school achievement test scores a statistically significant predictor of college enrollment in North Carolina?

5 of 22

RESEARCH SUB QUESTIONS

Q1: How well do each of the achievement tests relate to college enrollment rates?
Q2: Is the combination of the three tests taken by all NC high school students a good predictor of College Enrollment Rates for North Carolina School Districts?
Q3: Is the combination of the three standard national college entrance exams scores a good predictor of College Enrollment Rates for North Carolina School Districts?

6 of 22

DATA SOURCE

NC State Report Card⁶ – Database of student performance and academic growth, school and student characteristics and college enrollment
Report cards include data for all NC public schools, including charter and alternative schools.
Data are reported at the school, district, and state levels.
District and state data: all schools in the same grade-range category are averaged to produce comparison results

7 of 22

VARIABLE DESCRIPTIONS

College Enrollment Rates (% College) – Percentage of students who enrolled in college
NC Math 1 EOC (% Math 1) and English 2 EOC (% Eng2) – Percentage of students who scored at least a level 4
ACT Proficiency Percentage (% ACT) – Percentage of students who scored at least 17
SAT Percentage Score (SAT %) – Mean student SAT score / 1130 x 100
AP Proficiency Percentage (% AP) – Percentage of students who scored at least a 3

8 of 22

DATA COLLECTION

Existing secondary data for the class of 2017 for each school district
College Enrollment Data: 6/2017 – 9/2018

Test	Data Collection Date
NC Math 1	May – June 2014
NC English 2	May – June 2015
ACT	May 2016
SAT	March – May 2016
AP	May 2016

9 of 22

DESCRIPTIVE STATISTICS

Variable	Observations	Mean	CV*
% College	115	57.6%	0.132
% Math 1	115	42.3%	0.303
% Eng2	115	45.3%	0.254
% ACT	115	54.6%	0.240
SAT %	114	92.4%	0.066
% AP	107	41.2%	0.386

Statistical Elements Calculated: Observations, Mean, Standard Deviation, Coefficient of Variation (CV) and Range

*CV = Std Dev/Mean; CV >= 1 🡪 high variation; CV < 1 🡪 low variation

Since the NC Department of Public Instruction (DPI) requires school districts to administer the Math 1 EOC, English 2 EOC and ACT, observations from all school districts from these tests were present in the data set.
Not all school districts reported SAT and AP data because these are optional tests which no students in the district elected to take.
The mean of all tests, except the SAT, was lower than the college enrollment rate. The SAT test mean was expectedly higher because the SAT % depicts a mean test score converted to percentage rather than the percentage of students in the school district who performed proficiently on the test.
Both Range and Standard Deviation are indicators of data variability. Standard deviation takes into account all data values within a data set, so it is a better reflection of the amount of variability in a data set. Since the researcher wanted to compare variability between variables, she calculated the coefficient of variability to assess how well the data are centered around the mean.
Since all variables had a CV < 0.4, the distribution for each variable is centered around the mean and all variables show low variation.

10 of 22

Q1: DATA ANALYSIS

Statistical Techniques: Scatterplots, Pearson’s Correlation*

Variable	Correlation Coefficient**	Observations	Significance
% Math 1 (1)	0.593	115	0.000
% Eng2 (1)	0.651	115	0.000
% ACT (1)	0.585	115	0.000
% ACT (2)	0.630	107	0.000
SAT % (2)	0.606	107	0.000
% AP (2)	0.674	107	0.000

*p < 0.05; **ranges from -1 (strong negative relationship) to +1 (strong positive relationship)

11 of 22

Q1: INTERPRETATION

Scatterplots infer a positive linear relationship between all tests and college enrollment rates
All achievement tests have a correlation coefficient between 0.5 and 0.7
Confirms that all achievement test variables have a moderately strong significant positive correlation with college enrollment⁷

12 of 22

Q2: DATA ANALYSIS

Statistical Techniques: Sequential Multiple Regression

* p < 0.05;

Model	Adjusted R²	Variable	Coeff.	P-value*
2.1	0.422	Constant	0.380	0.000
		% Math 1	0.098	0.206
		% Eng 2	0.341	0.000
2.2	0.417	Constant	0.378	0.000
		% Math 1	0.088	0.312
		% Eng 2	0.329	0.001
		% ACT	0.021	0.812

Note: N=115, p < 0.05 🡪 statistical power of greater than 0.98

13 of 22

% College = 0.378 + 0.088 (% Math 1) + 0.329 (% Eng 2)

+ 0.021 (% ACT)

College Enrollment Rate (%)

Achievement Test Proficiency (%)

Regression Plot for College Enrollment Rate by Achievement Test Proficiency

14 of 22

Q2: INTERPRETATION

Math 1 CCR Percentage, English 2 CCR Percentage and ACT Proficiency can statistically significantly predict College Enrollment Rates
No real change in Adjusted R² for 2^nd model 🡪 ACT Proficiency does not significantly contribute to the overall model as a predictor of College Enrollment Rates

15 of 22

Q3: DATA ANALYSIS

Statistical Techniques: Sequential Multiple Regression

Model	Adjusted R²	Variable	Coeff.	P-value*
3.1	0.345	Constant	0.354	0.012
		% ACT	0.312	0.001
		SAT %	0.058	0.766
3.2	0.498	Constant	0.376	0.008
		% ACT	0.182	0.034
		SAT %	0.020	0.918
		AP %	0.201	0.000

* p < 0.05

Note: N=107, p < 0.05 🡪 statistical power of greater than 0.98

16 of 22

College Enrollment Rate (%)

Achievement Test Proficiency (%)

Regression Plot for College Enrollment Rate by National Achievement Test Proficiency

% College = 0.376 + 0.182 (% ACT) + 0.020 (SAT %) + 0.201 (% AP)

17 of 22

Q3: INTERPRETATION

ACT Proficiency, SAT % and AP Pass % can statistically significantly predict College Enrollment Rates.
The predictive ability of the model increases by 15% with the addition of AP Pass %
Insignificant p-value for the SAT % variable likely because SAT % represents a mean test score converted to percentage rather than the percentage of students in the school district

18 of 22

IMPLICATIONS

High school students should strive to achieve a proficient score on achievement tests if they are interested in going to college.
NC DPI should sustain a strong curriculum for the Math 1 and English 2 courses and implement supports to assist more students to achieve a Level 4 or 5 on these tests.
Important for schools and districts to offer AP courses and prepare students for the associated AP tests to increase change of college enrollment

19 of 22

CONCLUSION

Findings confirmed the researcher’s hypothesis that district level high school achievement test scores are a statistically significant predictor of college enrollment in North Carolina.
Limitations

Selection of a specific state limits generalizability of findings to other states
SAT and AP not required 🡪 participation rates vary by school district and results are thus not equivalent between districts

Findings confirmed the researcher’s hypothesis that district level high school achievement test scores are a statistically significant predictor of college enrollment in North Carolina.
The data reported for this study represented school districts from the state of North Carolina for one graduating class. The selection of a specific state limits generalizability of findings to other states. The nature of Math 1 and English 2 course characteristics may not be representative of high school level math and English courses in other states.
SAT and AP score data were limited by the non-requirement of these tests by all schools and school districts. This presents a limitation since participation rates on those tests varied by school district and percentage results are thus not equivalent between school districts.

20 of 22

FUTURE RESEARCH

Predictive value of sub tests on college major.
Predictability of college enrollment based on test scores from schools across the state with similar demographics
Comparison of district data across multiple states

21 of 22

REFERENCES

Springer, S. P., Reider, J., & Morgan, J. V. (2017). Admission matters : What students and parents need to know about getting into college. John Wiley & Sons. Retrieved from ProQuest Ebook Central, https://ebookcentral.proquest.com/lib/ncsu/detail.action?docID=4835246
Harwell, M., Moreno, M., & Post, T. (2016, June). A study of the relationship between the ACT college mathematics readiness standard and college mathematics achievement. Journal of Psychoeducational Assessment, 34(3), 269-281.
NCDPI/North Carolina Testing Program. (2014, October 16). EOG and EOC Assessments Technical Brief. Retrieved from North Carolina Public Schools: http://www.ncpublicschools.org/docs/accountability/testing/eoceogtechbrief14.pdf

22 of 22

REFERENCES

NC Department of Public Instruction. (2019). ACT and ACT Workkeys Results. Retrieved from North Carolina Public Schools: http://www.ncpublicschools.org/accountability/act/
College Foundation of NC. (2019). Minimum Admission Requirements. Retrieved from NCPLANFORCOLLEGE.ORG: https://www.ncplanforcollege.org/college-planning/minimum-admission-requirements/
NC Department of Public Instruction. (2019). NC School Report Cards. Retrieved from Department of Public Instruction: https://ncreportcards.ondemand.sas.com/SASVisualAnalyticsViewer/VisualAnalyticsViewer_guest.jsp?reportPath=/ReportCard/NC_SRC&reportName=NC+Report+Cards
Dancey, C. P., & Reidy, J. (2004). Statistics Without Maths for Psychology. New York: Prentice Hall