Orientation, Graduation, and Anticipatory Socialization

1
Orientation, Graduation, and Anticipatory
Socialization

• Dissertation Defense

Beckie Hermansen Utah State University
12/12/06
2
Exchange
EXCHANGEOf students time, efforts, knowledge
for education offered by the institution
Student
Institution

• Explicit Contracts and Implicit Contracts
• Little or no guarantee uncertainty

3
Uncertainty
Persistence and Graduation
Uncertainty
Anticipatory Socialization
4
Postsecondary Socialization
Socialization for students not participation in
an orientation
Socialization process marked by high levels of
uncertainty and increased risk of exit from the
institution
Socialization for students participating in an
orientation
5
Persistence Study Model
CollegePersistence
6
Research Questions

• Participants Higher 1st semester and 1st year
  cumulative GPA
• Participants higher graduation and/or
  certificate completion.
• Non-Participants greater withdrawal at the end
  of the first year than participants
• Participants higher transfer to 4 year
  programs/institutions.

7
Descriptive Statistics (N 1143)

• 587 Start Smart 556 non-Start Smart
• 731 (64) female 408 (36) male
• Average age 19
• White 93.7
• Average family contribution 6,447
• Average High school GPA 3.4
• Average ACT score 20.65
• 50.1 declared a major at matriculation
• 52.6 received a degree
• 34 transferred to a higher educational
  institution

8
RQ1

• Participants Higher 1st semester and 1st year
  cumGPA

• Multiple Regression on First Semester
  Cumulative GPA
• Multiple Regression on First Year Cumulative
  GPA
• Dependent Variable GPA (T1 or T2)
• Independent Variables
• Age Gender Ethnicity Income Level
  High School GPA ACT Score Start Smart
  Participation

These variables were included to account for the
socioeconomic factors known to influence GPA and
college success.
9
RQ1

• Participants Higher 1st semester and 1st year
  cumGPA

• Multiple Regression on First Semester
  Cumulative GPA
• Coefficient of Determination (r2) .391 or
  39 (F(7,483) 44.295, p .000)
• Significant relationships
• ACT Score (t(490) 4.581, p .000)
• Start Smart (t(490) 4.720, p .000)
• High school GPA (t(490) 10.998, p .000)

This indicated that Start Smart enrollment did
have an effect however, it was less powerful or
indirect when combined with high school GPA and
ACT score.
Cohort average high school GPA 3.4 (Coding 0
is lt average 1 is gt average)
10
RQ1

• Participants Higher 1st semester and 1st year
  cumGPA

• Tests of Between-Subjects Effects
• Start Smart X High School GPA F(7,1069)
  3.635, p .057

Averages T1 GPA 2.84High school GPA 3.4ACT
Score 20.65
11
RQ1

• Participants Higher 1st semester and 1st year
  cumGPA

• Multiple Regression on First Year Cumulative
  GPA
• Coefficient of Determination (r2) .337 or
  34 (F(7,401) 29.075, p .000)
• Significant relationships
• Start Smart (t(408) 3.627, p .000)
• ACT Score (t(408) 4.009, p .000)
• High school GPA (t(408) 10.254, p .000)

These results were consistent with first semester
GPA with high school GPA have the most powerful
effect followed by ACT score and Start Smart
participation.
Cohort average high school GPA 3.4 (Coding 0
is lt average 1 is gt average)
Cohort average ACT score 20.65 (Coding 0 is
lt average 1 is gt average)
12
RQ1

• Participants Higher 1st semester and 1st year
  cumGPA

• Tests of Between-Subjects Effects
• Start Smart x High School GPA Start Smart x
  ACT Start Smart x ACT x High School GPA not
  significant!

Averages T2 GPA 2.89High school GPA 3.4ACT
Score 20.65
13
RQ2

• Participants Higher graduation rates

• Graduation Rate Comparison
• Dependent Variable Graduation
• Independent Variables
• Start Smart Group Non Start Smart Group

14
RQ2

• Participants Higher graduation rates

• Correlation on Graduation Rate and Group
• Start Smart result was r .185, a .01

Start Smart students graduated almost 2 to 1
(1.7) compared to non-Start Smart students at the
4 the semester.
15
RQ3

• Non-Participants greater withdrawal at the end
  of the first year than participants

• Survival Analysis
• Dependent Variable Time and Status -- for this
  cohort there were 12 time intervals or semesters,
  excluding summer terms-- status was either
  censored (no event) or uncensored (terminating
  event)
• Independent Variables
• Age Gender Ethnicity Income Level
  High School GPA ACT Score Start Smart
  Participation

16
Survival-Time Analysis
RQ3

• Logistic regression does not deal well with
  sample attrition
• Unique characteristic of stop-out from
  college/university.
• Mission, marriage, maternity, money.
• Examine distributions given a time period between
  two events (matriculation and graduation)
• Life-Tables, Kaplan-Meier, and Cox Regression
  analysis

17
RQ3

• Non-Participants greater withdrawal at the end
  of the first year than participants

• Kaplan Meier survival probabilities (see
  handout)
• Mean Life statistic
• Start Smart 4.4 semesters/ non-Start Smart
  4.2 semesters
• Hazard Probabilities (see handout)
• Log-Rank Statistic
• Log-Rank value .628 (a .428) . . . not
  significant.

It is difficult for the log-rank test to find a
difference when survival curve lines cross, as
was the case in this study. In the absence of a
significant log-rank statistic, reliance on
graphical representation of survival curves and
associated survival probabilities is paramount.
18
RQ3

• Non-Participants greater withdrawal at the end
  of the first year than participants

Predicted Survival and Hazard Functions for the
Fall 200 Freshman Cohort (00 equals non-Start
Smart or Orientation participants 1.00 Start
Smart Orientation students).
19
RQ3

• Non-Participants greater withdrawal at the end
  of the first year than participants

• Cox Regression Analysis
• Accounts for the influence of different
  variables on survival over time
• Unique ability to analyze interactions between
  variables.

20
RQ3

• Non-Participants greater withdrawal at the end
  of the first year than participants

Cox Regression with Variables
Variables in the Equation
21
RQ3

• Non-Participants greater withdrawal at the end
  of the first year than participants

Cox Regression with Interaction Terms
Variables in the Equation
22
RQ3

• Non-Participants greater withdrawal at the end
  of the first year than participants

• With betas of -.346 for gender and -.393 for
  income (p .000), persistence significance was
  found for female students with lower than average
  family income contributions.
• No significance was found for high school GPA,
  gender, or Start Smart participation, even with
  interaction terms.
• In fact, high school GPA did not have a
  significant influence on persistence beyond the
  first year of college.
• This confirms the Kaplan-Meier findings
  (similar curves).
• Start Smart was not a factor in long-term
  student persistence participants and
  non-participants experienced equal or close to
  equal termination and persistence rates over time.

23
RQ4

• Participants higher transfer to 4 year
  programs/schools.

• Correlation between participation status and
  transfer
• Dependent Variable Transfer
• Independent Variables
• Start Smart Group Non-Start Smart Group

24
RQ4

• Participants higher transfer to 4 year
  programs/schools.

• Pearson Correlation r -.079 a .05

Non-Start Smart Transfer Rate 116/556 or
21 Start Smart Transfer Rate 96/587 or 16
It seemed that Start Smart students were less
likely to transfer than their non-Start Smart
peers.
25
Repetitive Findings (N 4,536)

• RQ1 Start Smart and GPA
• Same variation in scores (2001 33, 2002 29,
  2003 21)
• High school average t-score 11.42
• ACT average t-score 6.43
• Start Smart average t-score 4.155
• RQ2 Graduation Rates
• Same observed pattern
• Start Smart average 263 compared to 164 (1.7
  1)
• Combined correlation 2.4 (r2 .0243) toward
  Start Smart and graduation

26
Repetitive Findings

• RQ3 Survival Analysis
• Non-significant log rank values (survival curves
  are similar)
• 2003 females higher persistence (more in the
  study)
• 2001 was significant (log rank 16.007, a
  .001). Median survival for Start Smart 4
  semesters non-Start Smart 3 semesters.
• Gender was the greatest predictor of persistence
  (females).
• RQ4 Transfer Rates
• Cohorts 2001 and 2002 were mixed
• No results for 2003
• Pearson correlation .054 resulting in a 1
  transfer difference in favor of Start Smart
  students (r2 .002916)

27
Conclusions

• RQ1 Start Smart had an indirect but significant
  impact on first term and first year cumulative
  GPA. High school GPA was most significant.
• RQ2 Start Smart students experienced higher,
  timely graduation rates compared to their
  non-Start Smart peers.
• RQ3 No significant relationship existed
  between Start Smart participation and long term
  survival or persistence.
• RQ4 Start Smart students did not experience
  equal transfer rates non-Start Smart students
  had greater transfer.

28
Conclusions

• All of these results, illustrate the complex
  dynamics of college student persistence and
  departure. This study affirmed the importance of
  high school GPA on early college academics and
  the fact that attendance in a one credit
  orientation program positively effected timely
  graduation. In this sense, the anticipatory
  socialization expressed in Start Smart did help
  students negotiate the implicit contract(s)
  leading to degree completion. In addition, this
  study displayed the importance of studying the
  potential effects of retention-related variables
  on a semester-by-semester basis.

29
Limitations

• Historical Threat to Validity
• Programmatic changes over time
• Administrative policy changes over time
• Program delivery changes over time
• Missionary effect
• No consideration given to re-entry or
  re-enrollment.

30
Implications

• Survival analysis
• Incorporating time as a dependent variable
  (whether and when a terminating event occurs)
• Different elements affecting persistence
• Pre-college characteristics
• Collegiate characteristics
• Predictive ability
• Logistic regression goes beyond correlation to
  prediction
• High school GPA x Start Smart First semester or
  first year success
• In-depth assessment of effectiveness
• Fiscal support of Start Smart
• Comprehensive program assessment for
  accreditation
• Support to competing enrollments and retention

31
Recommendations

• Survival analysis applied to other
  intervention/retention programs
• Remedial education programs
• Upward Bound-type programs
• Early college programs
• Sports/Intramurals/Student Leadership
• Interactions between predictors and time
• Look at each predictor over time
• Determine transient or permanent effects
• Missionary effect
• Allow for re-entry either with original cohort or
  existing cohort
• Allow for part-time student analysis/study
• Survival analysis in terms of student
  decision-making
• Variables affecting decisions to withdraw or
  persist over time

32
Future Opportunities

• Application to other programs (athletics,
  remedial classes (Math or English).
• Study involving part-time students and
  persistence over time.
• Postsecondary IEP or SEPan advancement of
  educational anticipatory socialization theory
  (best practices).
• Start Smart combined with a capstone course for
  student success (measured college outcomes)

33
THANK YOU!