Attack of the Killer courses

1
Attack of the Killer courses

• How course taking patterns affect retention

2
Jaclyn Cameron

• Research Analyst
• DePaul University
• Chicago, IL
• Presented at National Symposium on Student
  Retention
• CSRDE 4th Annual Conference
• September 29th-October 1st 2008

3
The Inspiration

• What we know
• Academic performance and progress highly related
  to retention
• Under-preparedness in Math, science, English
  related to student success
• What we dont necessarily know
• Details concerning first year academic
  performance

4
Defining the Role

• Killer course A course in which high
  proportions of students earn non-passing grades
  or withdraw after the drop date.
• Previous appearances
• Supplemental instruction
• Learning communities
• Parallel courses
• Educational centers (labs)
• Implications

5
Characters

• First Year Freshmen (n 7,226)
• Community College Transfers (n 3,059)
• Killer courses
• Minimum total enrollment
• 150 Freshmen (50 avg/yr)
• 100 Transfers (33 avg/yr)
• Percent of D, F, W Grades gt 10

6
Character Development
Student Roster Keyed on student ID, Unique
All 1st Yr course enrollments Keyed on student
ID, Duplicated
All 1st Yr course enrollments of selected
students Keyed on student ID, Duplicated (one
record for each course enrollment)
Once killer courses are identified, they are
flagged as such in this table
All unique 1st Yr courses w/ count of grades
earned Keyed on individual course. Yields list
of Killer courses
All unique students w/counts of courses and
grades Keyed on student ID. Includes all
demographic and academic information
7
The Set
Credit hours of killer courses
Race (URM)
Credit hours of non-killer courses
1st Yr Retention (Y/N)
Total credit hours taken in the 1st Yr
GPA of killer courses
GPA of non-killer courses
CC Data Total credit hours transferred in
CC Data Location of CC (City/Suburban)
8
Act 1

• The Killer Courses

9
Act 1, Part 1 Freshmen Killer Course List
10
Act 1, Part 2 Community College Transfer Killer
Course List
11
Act 1, Part 3 Comparing the Lists

• Math, Math, and more Math
• Transfers have more Math and higher DFW rates in
  general
• Gen Eds vs. Major Requirements
• Freshmen have more Liberal Arts, general
  education courses
• Sequenced vs. Disordered
• For Freshmen, the DFW proportion is higher in the
  later courses of the sequence
• For Transfers, full sequences are rare, and early
  courses of a sequence have higher DFW rates.

12
Act 2

• The Investigation

13
Act 2, Part 1 Two Questions

• Do the amount of killer courses taken predict
  retention?
• Percent of killer courses taken
• Total number of course hours taken
• Total credit hours of non-killer courses taken
• Race
• Does performance in killer courses predict
  retention?
• Only students who took at least one killer course
• Included the all the above plus
• Killer course GPA
• Non-killer course GPA
• Total killer courses taken x killer course GPA
• CC Controls
• of hours transferred in
• Location of CC (city/suburban)

14
Act 2, Part 2 Regression 1 Freshmen Course
Taking Activity

• Two Blocks
• Race
• Ttl credit hrs, ttl Killer credit hrs, ttl
  non-killer credit hrs
• Overall fit was significant (?2 1399.6, df 4,
  n 6753, plt .05)
• 32 variance explained (Nagelkerke pseudo R2, -2
  Log likelihood 4625.77)
• Correctly predicted 88.4 (Correct 98.8
  retained, 35.4 non-retained)
• Significant Predictors
• Total Credit Hrs ()
• Non-Killer Credit Hrs ()
• Killer Credit Hrs ()

15
Act 2, Part 3 Regression 2 Transfer Course
Taking Activity

• Two Blocks
• Race, Transfer Hours, CC Location
• Ttl credit hrs, ttl Killer credit hrs, ttl
  non-killer credit hrs
• Overall fit was significant (?2 399.11, df 6,
  n 1755, plt .05)
• 35 variance explained (Nagelkerke pseudo R2, -2
  Log likelihood 1131.47)
• Correctly predicted 86.2 (Correct 96.1
  retained, 33.2 non-retained)
• Significant Predictors
• Mediated relationships race (ltgt), transfer hrs
  (), CC locale (City -)
• Non-killer credit hrs ()
• Killer credit hrs ()
• Total hrs (-) ?

16
Act 2, Part 4 Regression 3 Freshmen - Courses
Performance

• Three Blocks
• Race
• Ttl credit hrs, ttl killer credit hrs, ttl
  non-killer credit hrs
• Killer cum GPA, non-killer cum GPA, killer cum
  GPA x ttl killer hrs
• Overall fit was significant (?2 851.41, df 7,
  n 4804, plt .05)
• 30 variance explained (Nagelkerke pseudo R2, -2
  Log likelihood 2967.80)
• Correctly predicted 89.6 (Correct 99 retained,
  29.7 non-retained)
• Significant Predictors
• Mediated relationships race ()
• Total credit hrs ()
• Killer credit hrs (-)
• Non-killer credit hrs (-)
• Non-killer GPA

17
Act 2, Part 5 Regression 4 Transfer - Courses
Performance

• Three Blocks
• Race, transfer hrs, CC locale
• Ttl credit hrs, ttl killer credit hrs, ttl
  non-killer credit hrs
• Killer cum GPA, non-killer cum GPA, killer cum
  GPA x ttl killer hrs
• Overall fit was significant (?2 276.98, df 9,
  n 1282, plt .05)
• 40 variance explained (Nagelkerke pseudo R2, -2
  Log likelihood 573.11)
• Correctly predicted 91.4 (Correct 98.2
  retained, 32.6 non-retained)
• Significant Predictors
• Non-killer credit hrs ()
• Non-killer GPA ()
• Transfer hrs ()
• Killer GPA x Killer credit hrs ()

18
Act 3

• The Verdict

19
Act 3, Part 1 General Conclusion Observations

• Different killer courses Different potential
  pitfalls
• 1st Analysis More courses More likely to
  persist
• Freshmen order Total, Non-killer, Killer
• Transfer order Non-killer, Killer, Total
• 2nd Analysis Non-Killer GPA trumps all
• Freshmen Killer and Non-killer hours now
  negative
• Transfers Only Non-killer, but with Interaction
  term
• Overall, may be a proxy for previous research
• Uncovered possibility of mediated relationships

20
Act 3, Part 2 Limitations

• Lack of other known retention predictors
• ACT, high school GPA, student satisfactions
• Did not adequately predict attrition
• Use of individual courses as a group
• Aggregate killer courses by departments

21
Act 3, Part 3 The Wrap-Up

• Implications for advising, especially for
  different populations
• Identify areas of improvement for student
  learning
• GPA is a direct reflection of coursework courses
  may not be relevant, but GPA is.
• Faculty and staff respond to the name killer
  course

22
the

• end