How to Employ ACCUPLACER Scores for Comparison Group Equating Gary Greer University of Houston Downtown NCTA San Diego September, 2011
Intro • External and internal reporting requires gathering, organizing, and arranging data. • Grant reporting requires gathering, organizing, and arranging data. • ACCUPLACER’s facility for gathering, organizing, and arranging data enables flexible access to test our hypotheses about correlates of student success.
Survival analysis • Educational researchers seek correlates of student re-enrollment. • Educational researchers seek correlates of student attrition.
3 Things we [should] hold true • In order to compare academic performance we must compare matched (equivalent) groups. • Matched groups must include only equally able students. • Academic ability is precisely measured by ACCUPLACER scores. (Note IRT)
Current Problems • Non-equivalent group comparisons are invalid. • Non-equivalent groups preclude valid statistical inferences. • University researchers regularly compare non-equivalent groups. • Grants permit and encourage comparison of non-equivalent groups.
Solutions to Current Problems • Equivalent group comparisons are valid. • Match groups by equal ability categories. • Universities should onlycompare matched groups. • Grants should onlyrequire comparisons of matched groups. • Group equating corrects non-equivalent group comparisons.
Target Group vs Comparison Group • Our Target group is our focus. • Our Comparison group is our reference. • Target and Comparison must be equated before differences can be asserted.
Target Groups of interest • Minority students • Minority Males • 1st Generation students (1st Gens) • Pell recipients • First Time in College students (FTICs) • First Year Success students • Mentor Program students • Students required to take 1 [or more] Dev Ed course(s) • Students who register early*
Measures for comparing groups – Outcome Variables • Dev Ed course GPA • College course GPA • % of Dev Ed courses completed • % of college courses completed • % re-enrolling • % declaring major • % graduating in 6 (or 7,8,9) years
Researchable Questions • To what extent do [equated] groups differ on outcome measures? • To what extent do[equated] group differences change after academic intervention?
Hypotheses about [equated] groups • Pell group % completion = Non-Pell group % completion • Minority GPA = Non-Minority GPA • 1st Gen Minority = non-1st Gen Minority • FTIC 1st Gen Minority = FTIC non-1stGen Minority • Mentored group graduation rate = non-Mentored group graduation rate
Group equating To verify academic interventions, we must onlycompare equivalent groups. Equivalent groups have equal ability (equal ACCUPLACER scores) but differ only by whether intervention was received.
Reiteration Matched [equivalent] groups have equal ability categories. Equal placement scores categorize equal ability. Groups whose members have equal groupings of ACCUPLACER scores have equal ability and are therefore comparable.
Findings Equivalent groups have equal ability. Equal placement scores indicate equal ability. Group comparisons of equally able students reveal actual differences (if and where differences exist).
Valid Inferences • Males % completion = Females % completion • Minority GPA = non-Minority GPA • 1st Gen Minority GPA = non-1st Gen Minority GPA • Mentored group graduation rate > non-Mentored group graduation rate
Conclusions Only equated group comparisons are valid. Group matching is achieved by way of ACCUPLACER score groupings. Demographic grouping is achieved by way of ACCUPLACER background (demographic grouping) questions. Non-equivalent group comparisons disable university research.
Thanks for attending Let’s talk more about group equating Gary Gary Greer Assistant Dean, University College University of Houston Downtown email@example.com