1
|
|
2
|
- Define what to measure
- Determine availability of data
- Obtain data
- Consolidate data into usable format
- Build statistical model
|
3
|
- Institutional diversity is greatest strength of American higher
education
- Size
- Location
- Student demographics
- Public/independent/church-related
- Mission
|
4
|
- Collaborative approach to developing an admissions score involving=
li>
- Dean of Enrollment Management
- Institutional Research Office
- Faculty
- Registrar
|
5
|
- Longitudinal studies from the US Dept of Education (Cliff Adelman)=
li>
- “Intensity” of high school curriculum
- 6 years of data on enrolled freshmen
- Student information system
- High school transcripts
|
6
|
- At time of admission decision, want to predict variables such as
- First-term GPA
- Cumulative GPA
- Probability of completing first semester of junior year
- Probability of graduating
|
7
|
- Independent variables from student information system
- SAT or ACT scores
- High school GPA, percentile rank in class
- Student demographics
- Expected family contribution (FAFSA)
- Proportion of need (FAFSA) met by free money
|
8
|
- Independent variables from high school transcript
- Grades, highest levels and number of semesters of academic subjects=
- Multiple foreign languages, Latin specifically
- IB, Honors, AP
- Quality of high school (College Board)
- % attending college
- Average SAT scores
- Number AP scores ³3
|
9
|
- High School Transcripts – Wish List
- Grades for selected courses
- # of foreign language courses
- Complete certain math courses
- Class rank
- Overall GPA
|
10
|
- High School Transcripts – Analysis
- Similarities
- Differences
- Grade scales and GPA
- Course names
- Available information
|
11
|
- High School Transcripts – Choices
- Enter selected and summarized data
- Slow down data entry
- Limited data for model
- Enter course data – summarize later
- Slow down data entry
- All possible data for model
|
12
|
- Database – Design Goals
- Easy to navigate
- Logical layout
- Flexible design
- Facilitate data entry
|
13
|
- Database – Initial Design
- Student lookup
- Forms by subject area
- Math, English, foreign language, etc.
- All course names listed
- Enter grade, qualifier, level
- Forms for grading scales
- Tied to high school or student
|
14
|
- Database
- Test
- Modify
- Deploy
- ForTAIR.mdb
- Complete data entry
|
15
|
- Build Final Data File
- Clean
- Unify grades
- Many to one
- Add new data
- Add summary variables
|
16
|
- Build Final Data File – Clean It Up
- Expect it
- Don’t skip it
- Allow time for it
- Types of errors
- Generic code
- Specific code
- Note future DB considerations
|
17
|
- Build Final Data File – Rescale Grades
- 100 point scale
- A-F + -
- 4-point,5-point, 8-point, 100+
- Missing scale information
- Find it on the web
- Standard scale
|
18
|
- Build Final Data File – Many To One
- Every course record -> 2 new variables
- Generic course names
- Loss of specifics
- Retain individual course grades/level
|
19
|
- Build Final Data File – Summaries
- Subject areas
- Course counts
- Level counts
- Grade average
- Other - # previous high schools attended et al
|
20
|
- Build Final Data File – More Data
- Financial aid
- Student /Applicant Info
- Demographics
- Dependent variable values
|
21
|
- Build Final Data File – Output Form
- 2000+ records and 350+ variables
- Variable definitions
- Compatible format
|
22
|
- Final Thoughts
- User-proof the database
- Rethink the design and layout
- Got all the data
- Worked fast – took cooperation and commitment
|
23
|
- Current Status: Handed off to 2 faculty members in quantitative area=
s to
develop mathematical model to predict
- First-term GPA
- Cumulative GPA
- Probability of completing 1st semester of junior year
- Probability of graduating
- Ultimately: Easy-to-use equation into which admissions representativ=
es
can insert applicant information
|
24
|
- Hurdles and Obstacles
- Lack of data integrity for high school GPA
- Missing data for rank in class
- Categorizing information from high school transcripts
- Time required
|
25
|
- Questions/more information
- Connie Tull
- Dr. Denise Young
|