Reaching Out to Retain At-Risk Students

							        Reaching Out to Retain
          At-Risk Students
   An Exploratory Study Using the Student
     Adaptation to College Questionnaire



         Dr. Mike Meacham and Dr. Marsha Krotseng
                  Valdosta State University
                        June 4, 2007


Funded by a Faculty Research Grant from Valdosta State University
                    Introduction

   Student retention increasingly important to
    university administrators, boards, and legislators

   Literature review revealed students decisions
    involve:
     – Academics
     – Social reasons
     – Personal problems
     – Adjustment to school environment
     The Student Adaptation to College
          Questionnaire (SACQ)
   Measures student adaptation on four indices
    found in literature as important

   67 questions rated from “applies very closely
    to me” to “does not apply to me at all”
                  The SACQ        (cont.)



   Higher scores indicate better adaptation

   Research on SACQ
     – Consistent Internal Reliability
     – Numerous studies attest to high reliability
       and validity
               Steps in the Study

   Received permission from IRB and Freshman
    Experience Courses Director. Reviewed students’
    rights.

   Fall 2006, administered to students in program
    above.

   Advisors reviewed results with individual students

   Summary data from questionnaires analyzed
    statistically.
                   Study Focus

   Exploratory studies frequently change the area of
    focus as data are gathered.

   Original Intent
    – Which students are at greatest risk of leaving
      the university?
    – What intervention strategies might help VSU
      retain these students?

   Data suggested new hypotheses
             Research Hypotheses
   H1: Student characteristics (demographic
    variables) are not significantly associated with
    institutional attachment.

   H2: Adjustment cluster and subscale scores from
    the SACQ are not significantly associated with
    institutional attachment.

   Which variables, among demographic questions,
    individual items, clusters, and subscale scores on
    the SACQ can be used to predict institutional
    attachment?
    Participant Characteristics Analyzed
     Sample Characteristics similar to the University
                   Participants (n=74)
   Gender
     – Female              61%
     – Male                39%
   Age
     – 18                  77%
     – 19                  18%
     – Other                5% (17, 30, 40)
   Ethnic Background
     – Caucasian           72% Hispanic       4%
     – African American    16% Multiracial 4%
     – Other                4%
   Hometown
     – Major metro area    32%
     – Other areas         68%
       Other Characteristics Considered
   First-year Residence
     – On campus           32%
     – Off campus          68%
   High School GPA
     – 3.5 or above        26%
     – 2.5 to 3.49         54%
     – 2.49 or below       19%
   Class Type
     – Cohort (common)     57%
     – Non-cohort          43%
   Enrollment
     – Full-time           91%
     – Part-time            9%
   First Semester GPA
     – 3.5 or above        34%
     – 2.5 to 3.49         41%
     – 2.49 or below       26%
            What Differences Exist?
   Three of Four SACQ Subscales significant

    – Academic Environment
        Ethnic Background
    – Social Environment
        Ethnic Background
        Residence (On campus/Off)
        Class Type (Cohort/Non-cohort)
    – Attachment
        Ethnic Background
        Hometown
        High School GPA
        First Semester GPA
          What Differences Exist?            (cont.)


   African-American students reported the lowest
    adjustment:
     – Academic Environment     (M = 6.25)
     – Social Environment       (M = 5.83)
     – Attachment               (M = 7.68)

   Based on a small number (12)

   Important to test this finding with a larger sample
     Focus on Institutional Attachment

   Only three of the 74 did not persist

   Important factor in persistence (Tinto)

   Anecdotal evidence from students

   Attachment differed significantly across four
    demographic variables, including hometown

   Can we use the SACQ to predict institutional
    attachment?
                            Valdosta State University
            Six Year Headcount Enrollment Trend: First-Time Freshmen
                               Fall 2001 - Fall 2006


100%




75%




50%




                                                                            %
              %




                                                                                            %
                                                          %




                                                                          .5
            .0




                                                                                          .7
                             %




                                                        .6




                                                                       40
         42




                                            %




                                                                                       40
                           .5




                                                     38
                                          .3
                        38




                                       36 %




                                                                             %
                                                         %




                                                                                            %
                           %




                                        .9




                                                                           .6
                                                       .5
            %




                                                                                          .4
                         .4




                                     33




                                                                        33
                                                    33




                                                                                       31
          .8




25%
                      32
       30




 0%
          Fall 2001




                         Fall 2002




                                        Fall 2003




                                                       Fall 2004




                                                                           Fall 2005




                                                                                                Fall 2006
                           41-County Service Region                Metro Atlanta
      Institutional Attachment Variable

   Coded as ordinal variable for analysis
     – Frequency analysis suggested three ordered
       classifications:

       Low         Lowest thru 5.99, f = 22
       Average     6.0 thru 7.99, f = 22
       High        8.0 thru 9.0, f = 30
                       Analyses
   Examined data for statistically significant
    relationships between student characteristics
    and institutional attachment

   No significant correlations found
Relationship of Institutional Attachment
         to SACQ Subscales

   Three adjustment subscales (all statistically
    significant at p < .01)

       1.   Academic Adjustment       .477
       2.   Social                    .550
       3.   Personal-Emotional        .498
    Relationship of Institutional Attachment
              to SACQ Clusters
   Ten cluster scores (all statistically significant at p < .01)

        1. Motivation                     .520
        2. Application                    .285
        3. Performance                    .297
        4. Academic Environment           .476
        5. General Social Adjustment      .392
        6. Other People                   .446
        7. Nostalgia                      .498
        8. Social Environment             .552
        9. Psychological                  .424
        10. Physical                      .503
      Predicting Institutional Attachment
   Ordinal logistic regression

   Tests with subscales and clusters yielded no significant
    predictors

   At the item level, four predictors emerged:

    1. Item 8 (+)
       I am very involved with social activities in college.
    2. Item 30 (+)
       I am satisfied with the extracurricular activities
       available at college.
    3. Item 41 (–)
       I’m not doing well enough academically for the
       work I put in.
    4. Item 65 (+)
       I am quite satisfied with my social life at college.
    Predicting Institutional Attachment
↓ I am very involved with social activities in college.
  (B= -.281)

↑ I am satisfied with the extracurricular activities
  available at college. (B= .320)


↑ I’m doing well enough academically for the work I
  put in.* (B= .266)

↑ I am quite satisfied with my social life at college.
  (B= .682)



* This is a negative variable as stated on the SACQ and is reworded
  for interpretation.
                 Classifying Cases

   Percent correctly assigned (n = 71)* using the
    predicted probability

    Low Attachment                  13/21         (62%)
    Average Attachment              7/21          (33%)
    High Attachment                 24/29         (83%)




*After eliminating outliers, the model consisted of 71 cases
           Classifying Cases      (cont.)


Low                      High
 Low HS GPA (≤ 2.49)     Female
 Less than full-time     18 – 19
  (Enrollment)            Caucasian
 Low 1st Semester GPA    All Other (Hometown)
  (≤ 2.49)                Mid/High HS GPA (≥ 2.5)
                          Off campus (Residence)
                          Cohort (Class type)
                          Full-time (Enrollment)
                          Mid/High 1st Semester
                           GPA (≥ 2.5)
                    Implications

   Identify students whose scores on those four
    items suggest lower institutional attachment

   Offer targeted interventions
                    Discussion

   Established programs to increase student
    retention existed before university began
    intensive focus:

    – Students provided with individual advisor
    – Student Assistance Centers
    – Special Assistance Centers in various
      departments
    – Student Counseling Center
                   Discussion (cont.)
   Strategic Planning has begun new programs and
    opportunities for students

   Academic Support
     – Advising given priority
     – Expanded and updated library facilities
     – OASIS
     – Student Success Center

   Social Support
     – Expanded Student Food Services
     – Expanded Student Union
     – Outdoor recreation centers
     – Renovated and built new residence halls
     – Student Recreation/Exercise Center
               Discussion (Needs)

   New programs, etc. will address issues for two
    categories of students with indications of
    adaptation problems:
     – Low high school GPA
     – Low first semester GPA

   Closer advising will support non full-time
    students, but other programs likely not to affect
    as strongly. As distance learning courses
    increase, more students likely to be off campus
    and not full-time.
                    Limitations

   Small sample size (N = 74)

   Ordinal regression required that the nominal
    dependent variable be split into ordered groups

   Low number of students in each group
    (22/22/30) may have limited classification
                Future Research
   Plan follow up administration with larger group

   Subsequent SACQ administration for longitudinal
    comparison.
     – What are the effects of living on campus or of
       cohort classroom environments on institutional
       attachment one year after participation?
     – Can institutional attachment be used to
       approximate persistence and degree
       attainment?

   To what degree do students’ social networks
    influence institutional attachment?
     – A social network analysis may reveal important
       information not apparent in perception surveys.
Discussion and Questions


         Contact:

  mgmeacha@valdosta.edu
   krotseng@valdosta.edu
                   Analyses
 Descriptive statistics
 Correlation coefficients (rs)
 Chi-squares (χ²)
    – (Frequency distributions, magnitude and
      direction of association, and significant
      associations between variables)
   Ordinal logistic regression
    – (Which independent variables were predictors
      of ordinal institutional attachment?
    – Frequency analysis suggested three ordered
      classifications,
       Low          Lowest thru 5.99, f = 22
       Average      6.0 thru 7.99, f = 22
       High         8.0 thru 9.0, f = 30
          Student Characteristics
   Of the nine student characteristics, a slight, but
    statistically insignificant association was found on
    one variable, gender [χ² (2, n = 74) = 4.18,
    p = .124, Cramer's V = .238].

   No other slightly statistically significant
    associations were found between other student
    characteristics and ordinal institutional
    attachment.
          Regression Coefficients
                B     Wald      Sig.    OR
Item 8      -.281     5.122     .024    .755

Item 30      .320     3.841     .050    1.377

Item 41      .266     4.166     .041    1.305


Item 65      .682     11.330    .001    1.978




           Predictive (8, 30, 41, 65)