First-year student retention and success is challenging for many higher education institutions. Many universities around the world including Qatar University have made significant changes to their admissions policies, which have resulted in significantly larger first-year intakes. For universities to fulfill their role of developing their nation’s economies by improving student success and retention, it is necessary to determine the factors that contribute to overall student satisfaction. To this end, ensuring that first-year students graduate is a critical issue for this university, as it is for many higher education institutions. One of the major strategies adopted by this institution was to implement a “first-year experience” initiative to assist first-year students to adapt to college life and to give guidance to those who are underperforming. The “first-year experience” initiative has included programs and activities such as student orientations, first-year seminars, and success workshops. An important part of this initiative was to measure and explore the five factors contributing to overall first-year student satisfaction. A questionnaire was designed to provide a snapshot of overall student satisfaction and to assess the correlation with the three dimensions and five identified factors: academic (course effectiveness and citizenship knowledge and skills), social (a sense of belonging and interaction with key college members), and environmental (student awareness and utilization of campus resources). These factors, which were based on Astin’s interaction theory and Tinto’s model of departure, have been found to be the most important elements in first-year college life. This study sought to gain a greater insight into the contributions these factors had on overall first-year student satisfaction. To achieve this aim, the questionnaire was administered to 282 first-year students, and the resulting data were analyzed using regression analysis and artificial neural networks. In line with previous research, it was found that student course ratings were the best predictors of overall college satisfaction, with student citizenship knowledge and skills and a sense of belonging also found to be highly correlated. In contrast to previous research, however, this study found that student interactions with key members such as faculty, school administrators/staff, and other students and an awareness and utilization of campus resources were not highly correlated with overall student satisfaction.
Student satisfaction definitions have varied widely depending on the research approach and focus. In this study, student satisfaction was examined from an educational perspective. Elliott and Healy [
This study used a selective student satisfaction evaluation model in which certain student satisfaction determinants linked to the college learning outcomes for successful first-year experiences were explored. This model is described in more detail in Conceptual Framework.
Student satisfaction is a major concern for higher education institutions. Bryant and Bodfish [
Another important reason that student satisfaction is of importance is that dissatisfied students often withdraw during their first year at college. From a financial perspective, it has been found that retaining students is more efficient than recruiting new students. Tinto [
“Student satisfaction” is one of the most important dimensions for the assessment of first-year experiences. For this reason, it is essential to explore the determinants or factors that influence overall first-year student satisfaction. It has been found that a student’s perceptions and experiences during their first college year lay the foundation for future success and graduation. This notion is supported by the work of Barefoot [
Evaluating first-year student experiences and satisfaction is a complex process because “student life is seemingly a web of interrelated activities and experiences” (Elliott [
Previous studies have confirmed the multidimensionality of student satisfaction. Hanssen and Solvoll [
Most colleges measure student satisfaction by administering student satisfaction surveys such as CIRP (Freshmen Survey), NSSE (National Survey of Student Engagement), SSI (Student Strength Inventory), and Noel Levitz survey. Billups [
As highly satisfied students are more likely to persist in their studies and graduate, it is important to regularly evaluate student satisfaction. Schertzer and Schertzer [
Satisfaction can motivate students to work harder, achieve success, and persist until graduation. Oja [
Studies that have utilized comprehensive approaches to student satisfaction include Gruber et al. [
Several other studies have concentrated on specific aspects of college life. Yang et al. [
Other studies have attempted to use more creative approaches to determine student satisfaction. Douglas et al. [
Student satisfaction is usually associated with academic experience evaluations and teaching effectiveness perceptions. For example, Marzo Navarro et al. [
This determinant refers to the student knowledge and skills required to be a successful citizen, which we believe could influence student satisfaction. In higher education, students are expected to become involved in civil engagement activities. For example, at this university, students are expected to gain the knowledge, values, and skills necessary to be responsible citizens; therefore, it was expected that a student who had these positive behaviors would be motivated to graduate from college and would have a positive view toward higher education in general and toward the institution in particular. Frederick [
A sense of belonging in this study refers to feeling a part of the campus community and a commitment toward the institution. Strayhorn [
Several studies have confirmed the impact of student-faculty interactions on higher education academic and social outcomes and overall student satisfaction [
This section discusses whether student knowledge and the utilization of available support services influence satisfaction with the college experience. Nasser et al. [
Astin’s input-environment-output (IEO) model and Tinto’s theory of student departure formed the basis for the conceptual framework in this study.
Astin [
Conceptual model for student satisfaction adapted from Astin’s I-E-O model.
Tinto’s theory of student departure or Tinto’s “Model of Institutional Departure” states that to persist, students need to be integrated into formal (academic performance) and informal (faculty/staff interactions) academic systems and formal (extracurricular activities) and informal (peer-group interactions) social systems [
This study examined the influence of several social and academic engagement factors on overall student satisfaction with the college experience; therefore, this study was not intended to evaluate student satisfaction as a whole but rather to investigate the influence of some factors on student satisfaction. The factors examined were those that have been highlighted in previous research as being essential for new students to adapt to university life. This study roughly classified these factors into three dimensions: academic (student 100-level course evaluation and citizenship skills and knowledge obtained), social (a sense of belonging and interactions with key university members), and environmental (awareness and utilization of campus resources). Many studies have focused on how factors such as student attitudes, perceptions, and academic and social engagements impact first-year student success and retention; however, few studies have attempted to explore the influence these factors have on student satisfaction and their overall perceptions of the college experience.
This study sought to fill a research gap and offer greater understanding of the elements that have been strongly associated with overall student satisfaction. Specifically, the study sought to answer the following research questions: What is the correlation strength to overall student satisfaction for each of the five factors? How do these five factors contribute to the variances in overall student satisfaction? How can binary logistic regression and artificial neural networks be used to predict overall student satisfaction (satisfied and less satisfied)? Which method provides the best overall accuracy value?
This study explored the influence of the academic, social, and environmental factors on first-year college student satisfaction, for which regression and classification methods were used to explore the capacity of these factors to predict overall student satisfaction. The focus of this study was on first-year students at a public four-year institution.
A questionnaire was developed and administered to 282 first-year students during three 100-level course lectures in Spring 2014, all of which were from the five general college requirements or “core courses.” In this university, core courses are part of the five high-impact practices for the first-year experience, which means that these courses have high first-year student enrolments. With an initial sample of 300 students, the overall response rate was 94%.
The questionnaire had several sections, as shown in Table
Items used for each variable.
Student overall satisfaction (Likert scale used: strongly agree “5”; agree “4”; neutral “3”; disagree “2”; strongly disagree “1”) |
|
|
|
|
|
|
100-level course evaluation (Likert scale used: strongly agree “5”; agree “4”; neutral “3”; disagree “2”; strongly disagree “1”) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Citizenship skills and knowledge (Likert scale used: strongly agree “5”; agree “4”; neutral “3”; disagree “2”; strongly disagree “1”) |
|
|
|
|
|
|
Sense of belonging (Likert scale used: strongly agree “5”; agree “4”; neutral “3”; disagree “2”; strongly disagree “1”) |
|
|
|
|
Interaction (scale used: never “0”; 1-2 times per semester “1”; 1-2 times per month “2”; once a week “3”; 2-3 times per week “4”; daily “5”) |
|
|
|
|
|
|
Awareness and utilization of support services |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Descriptive statistics by course and gender.
Gender | |||||
---|---|---|---|---|---|
Male | Female | Frequency | Percent | ||
Course name | ENGL 110 | 36 | 66 | 102 | 36.2 |
MATH 103 | 43 | 51 | 94 | 33.3 | |
DAWA 111 | 26 | 60 | 86 | 30.5 | |
Total | 105 | 177 | 282 | 100.0 |
Three courses: DAWA 111 “Islamic Culture,” MATH 103 “Intermediate Algebra,” and ENGL 110 “English I,” from the 100-level courses were selected for the survey. These three courses were selected from the five courses identified by the college administration as “high-impact” practices, which meant that there was a high first-year enrolment as these were important elements of the first-year experience. An email was sent to instructors to seek permission for the researcher to administer the survey in the last 10 minutes of class time. Participation in this paper-and-pencil survey was voluntary, and it was made clear that students do not have to answer if they had already answered the survey in another class. The survey was administered in Arabic to make it more accessible, and completed questionnaires were returned to the researcher at the end of the class.
The independent variable in this study was overall student satisfaction. There were five dependent variables: course effectiveness evaluations, citizenship knowledge and skills, a sense of belonging, interaction with key college members, and awareness and utilization of campus resources.
Data were analyzed using SPSS v. 24. First, descriptive statistics were generated to gain an overall idea of the sample collected, after which Pearson’s correlation coefficient was used to explore the relationships between the variables. Multiple linear regression analyses were performed to calculate the total variance in student satisfaction that could be explained by the five factors, and then the overall student satisfaction data were used to classify students based on their satisfaction level (satisfied or less satisfied). Binary logistic regression and artificial neural network predictive analytic procedures were then used to check the capabilities of the five factors to predict student satisfaction levels and to group the students into one of the two satisfaction level categories. The predictive accuracy of the two models was determined using the value of the overall percentage of correct predictions calculated by each model.
The initial analysis of the data indicated that there were no statistically significant differences based on gender (male vs female) or residency status (local vs resident). As the inclusion of gender and residency did not increase the prediction accuracy of the models, these factors were not used in the subsequent analyses.
There were 105 males and 177 females in the sample. The sample included around 30% of the evaluations for each course (Table
Descriptive statistics were conducted on the dependent variable (overall college satisfaction) and the five predictive independent variables. Table
General descriptive statistics—minimum, maximum, mean, and standard deviation.
Minimum | Maximum | Mean | Std. deviation | ||
---|---|---|---|---|---|
Y | Overall college satisfaction | 1.00 | 5.00 | 3.6333 | 0.76176 |
X1 | 100-level course evaluations | 1.00 | 5.00 | 3.6313 | 0.68003 |
X2 | Citizenship knowledge and skills | 1.00 | 5.00 | 3.3064 | 0.78447 |
X3 | Sense of belonging | 1.00 | 5.00 | 3.9185 | 0.75753 |
X4 | Student interaction | 0.00 | 3.00 | 0.6290 | 0.50215 |
X5 | Awareness and utilization of campus resources | 0.00 | 5.00 | 2.5794 | 1.21644 |
Bivariate correlational analyses (Table
Pearson’s correlation between overall student satisfaction and the five factors.
Overall college satisfaction | 100-level course evaluations | Citizenship knowledge and skills | Sense of belonging | Interaction | Awareness and utilization of campus resources | |
---|---|---|---|---|---|---|
Overall college satisfaction | 1 | 0.579 |
0.354 |
0.537 |
0.171 |
0.107 |
100-level course evaluations | 0.579 |
1 | 0.299 |
0.440 |
0.276 |
0.014 |
Citizenship knowledge and skills | 0.354 |
0.299 |
1 | 0.555 |
0.279 |
0.112 |
Sense of belonging | 0.537 |
0.440 |
0.555 |
1 | 0.205 |
0.015 |
Interaction | 0.171 |
0.276 |
0.279 |
0.205 |
1 | 0.031 |
Awareness and utilization of campus resources | 0.107 | 0.014 | 0.112 | 0.015 | 0.031 | 1 |
Note:
A linear regression was generated for the five independent variables and the dependent variable (overall college satisfaction) to determine the independent variables most likely to influence overall student satisfaction.
The model summary table (Table
Model summary.
Model |
|
|
Adjusted |
Std. error of the estimate |
---|---|---|---|---|
1 | 0.667 | 0.445 | 0.434 | 0.57298 |
Note: predictors: (constant), awareness and utilization of campus resources, 100-level course evaluations, student interaction, citizenship knowledge and skills, and sense of belonging. Dependent variable: overall college satisfaction.
ANOVA.
Model | Sum of squares | df | Mean square |
|
Sig. | |
---|---|---|---|---|---|---|
1 | Regression | 66.146 | 5 | 13.229 | 40.295 | 0.000 |
Residual | 82.405 | 251 | 0.328 | |||
Total | 148.551 | 256 |
Note: dependent variable: overall college satisfaction. Predictors: (constant), awareness and utilization of campus resources, 100-level course evaluations, student interaction, citizenship knowledge and skills, and sense of belonging.
The coefficients table (Table
Coefficient of determination.
Model | Unstandardized coefficients | Standardized coefficients | Correlations | Collinearity statistics | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|
B | Std. error | Beta |
|
Sig. | Zero order | Partial | Part | Tolerance | VIF | ||
1 | Constant | 0.335 | 0.241 | 1.388 | 0.166 | ||||||
100-level course evaluations | 0.480 | 0.060 | 0.429 | 8.003 | 0.000 | 0.579 | 0.451 | 0.376 | 0.770 | 1.299 | |
Citizenship knowledge and skills | 0.039 | 0.056 | 0.040 | 0.687 | 0.493 | 0.354 | 0.043 | 0.032 | 0.653 | 1.531 | |
Sense of belonging | 0.333 | 0.061 | 0.331 | 5.489 | 0.000 | 0.537 | 0.327 | 0.258 | 0.608 | 1.645 | |
Student interaction | −0.044 | 0.076 | −0.029 | −0.578 | 0.564 | 0.171 | −0.036 | −0.027 | 0.881 | 1.135 | |
Awareness and utilization of campus resources | 0.058 | 0.030 | 0.093 | 1.954 | 0.052 | 0.107 | 0.122 | 0.092 | 0.984 | 1.016 |
The aforementioned multiple linear regressions were based on a continuous value for overall student satisfaction. In addition to modeling student satisfaction using linear regressions, other regression models based on a dichotomous dependent variable were also applied. The purpose of this step was to investigate whether the student satisfaction modeling could be further simplified and optimized. The dependent variable (Table
Overall student satisfaction classification.
Cat # | Scale | Label | Frequency | Percent | Valid percent | Cumulative percent |
---|---|---|---|---|---|---|
1 | <3.5 | Less satisfied | 127 | 45 | 45.8 | 45.8 |
2 | ≥3.5 | Satisfied | 150 | 53.2 | 54.2 | 100.0 |
Total | 277 | 98.2 | 100.0 | |||
Missing | 5 | 1.8 | ||||
Total | 282 | 100.0 |
In this section, two supervised learning regression methods: binary logistic regression and artificial neural networks, were applied.
Student satisfaction was modeled using binary logistic regression on IBM SPSS v. 24. Binary logistic regression is a predictive analysis used to describe data and explain the relationships between the dependent binary variable and the other independent variables. The data were examined to assess the multicollinearity between the predictors. As can be seen in the correlation matrix in Table
Table
Logistic regression model summary.
Step | −2 log likelihood | Cox and Snell |
Nagelkerke |
---|---|---|---|
1 | 252.087a | 0.324 | 0.433 |
aEstimation terminated at iteration number 6 because parameter estimates changed by less than 0.001.
The classification table (Table
Classification table.
Observed | Predicted | Percentage correct | |||
---|---|---|---|---|---|
Overall college satisfaction | |||||
Less satisfied | Satisfied | ||||
Step 1 | Overall college satisfaction | Less satisfied | 80 | 35 | 69.6 |
Satisfied | 24 | 117 | 83.0 | ||
Overall percentage | 77.0 |
The variation in the equation table (Table
Variables in the equation.
|
SE | Wald | df | Sig. | Exp( |
95% CI for Exp( | |||
---|---|---|---|---|---|---|---|---|---|
Lower | Upper | ||||||||
Step 1a | 100-level course evaluations | 2.013 | 0.357 | 31.712 | 1 | 0.000 | 7.483 | 3.714 | 15.077 |
Citizenship knowledge and skills | −0.030 | 0.257 | 0.014 | 1 | 0.906 | 0.970 | 0.586 | 1.605 | |
Sense of belonging | 1.421 | 0.320 | 19.684 | 1 | 0.000 | 4.143 | 2.211 | 7.762 | |
Student interaction | −0.561 | 0.342 | 2.690 | 1 | 0.101 | 0.571 | 0.292 | 1.116 | |
Awareness and utilization of campus resources | 0.148 | 0.133 | 1.243 | 1 | 0.265 | 1.159 | 0.894 | 1.504 | |
Constant | −12.678 | 1.788 | 50.247 | 1 | 0.000 | 0.000 |
aVariables entered in step 1: 100-level course evaluations, citizenship knowledge and skills, sense of belonging, student interaction, and awareness and utilization of campus resources.
In this section, a predictive model was developed using IBM SPSS artificial neural networks, which uses nonlinear data modeling to explore the complex relationships in the data to gain greater insights.
The Multilayer Perceptron (MLP) module of IBM SPSS v. 24 was used to build the neural network model and check the prediction capacity of the five independent variables. The case-processing summary (Table
Case-processing summary.
|
Percent | ||
---|---|---|---|
Sample | Training | 165 | 64.5 |
Testing | 91 | 35.5 | |
Valid | 256 | 100.0 | |
Excluded | 26 | ||
Total | 282 |
The network information table (Table
Network information.
Input layer | Covariates | 1 | 100-level course evaluation |
2 | Citizenship knowledge and skills | ||
3 | Sense of belonging | ||
4 | Student interaction | ||
5 | Awareness and utilization of campus resources | ||
Number of unitsa | 5 | ||
Rescaling method for covariates | Standardized | ||
|
|||
Hidden layer(s) | Number of hidden layers | 1 | |
Number of units in hidden layer 1a | 3 | ||
Activation function | Hyperbolic tangent | ||
|
|||
Output layer | Dependent variables | 1 | Overall college satisfaction |
Number of units | 2 | ||
Activation function | Softmax | ||
Error function | Cross-entropy |
aExcluding the bias unit.
A hyperbolic tangent was used for the activation function in the hidden layer, which included three units. The network diagram generated by SPSS (Figure
Network diagram.
The model neural network summary in Table
Model summary for the neural network results.
Training | Cross-entropy error | 71.055 |
Percent incorrect predictions | 21.8 | |
Stopping rule used | 1 consecutive step with no decrease in errora | |
Training time | 0:00:00.07 | |
|
||
Testing | Cross-entropy error | 37.224 |
Percent incorrect predictions | 16.5 |
Note: dependent variable: overall college satisfaction. aError computations are based on the testing sample.
A results summary is shown in the classification table (Table
Classification table.
Sample | Observed | Predicted | ||
---|---|---|---|---|
Less satisfied | Satisfied | Percent correct | ||
Training | Less satisfied | 63 | 7 | 90.0 |
Satisfied | 29 | 66 | 69.5 | |
Overall percent | 55.8 | 44.2 | 78.2 | |
|
||||
Testing | Less satisfied | 39 | 6 | 86.7 |
Satisfied | 9 | 37 | 80.4 | |
Overall percent | 52.7 | 47.3 | 83.5 |
Dependent variable: overall college satisfaction.
Independent variable importance is a measure of the degree to which the network model is able to predict value changes for the independent variables. The normalized importance in the independent variable importance table (Table
Independent variable importance.
Importance | Normalized importance (%) | |
---|---|---|
100-level course evaluations | 0.360 | 100.0 |
Citizenship knowledge and skills | 0.055 | 15.4 |
Sense of belonging | 0.308 | 85.6 |
Student interaction | 0.172 | 47.8 |
Awareness and utilization of campus resources | 0.105 | 29.3 |
Various student experiences in the first-year college lay the foundation for future success and final graduation. Student perceptions about the academic, social, and environmental aspects of the college community can have a significant impact on overall college experiences. Using IBM SPSS v. 24, data from a developed questionnaire were explored and analyzed, and bivariate correlations, linear regressions, and artificial neural networks were employed to determine the relationships and the capabilities of these factors to predict overall student college satisfaction.
It was found that student perceptions of the academic, social, and environmental aspects were positively correlated with overall satisfaction with the college experience. The regression analyses found that the students were satisfied with the quality of their first-year experience when the courses offered assisted them in adapting to college life, helped them participate in social and academic activities, increased their motivation, and improved their academic skills. This finding was consistent with studies conducted by Mai [
The other factor that was found to be highly correlated with overall student college satisfaction was students’ sense of belonging. Feeling part of the college community was found to contribute significantly to student satisfaction, which supported studies by Thomas and Galambos [
The third factor was student citizenship knowledge and skills. Knowledge about the local community and the training and research opportunities available were found to contribute to feelings of satisfaction with the first-year college experience. The more the knowledge about the society, the nation, and the world the student had, the more positive their views toward higher education. In contrast to previous studies, student interactions and student resource awareness and utilization were found to be less correlated with the overall student college experience, which was in contrast to findings by Astin [
The fifth factor, student awareness and utilization of campus resources, was found to have less predictive power than the other factors. While this finding was consistent with that by Mariani et al. [
The five factors were found to contribute 44% to the variance in overall student satisfaction with the first-year college experience. The binary logistics regression revealed an overall predictive accuracy of 77%, with student course satisfaction and sense of belonging being the best determinants of overall student satisfaction. The artificial neural network results further confirmed these significant correlations and the prediction capacities of the investigated factors. In the training dataset, the multilayer perceptron model showed the capabilities of the five factors to predict “satisfied” and “less satisfied” students with an accuracy rate of 78.2% for the training sample and 83.5% for the test sample, with the normalized importance chart indicating that student course satisfaction was the most important factor in this model, followed by the students’ sense of belonging. Based on the overall prediction accuracy value, it was concluded that the artificial neural network outperformed the binary logistic regression. Elliot and Shin [
This study indicated that overall student college satisfaction could be improved by enhancing those elements that could change students’ attitudes and perceptions and make them feel happier and more satisfied with their overall first-year experiences. In this case, more emphasis needs to be given to the learning experiences offered through the 100-level courses, and these courses should continue to enhance their content and delivery. Some important elements under the course satisfaction factor were encouraging students to participate in academic and social activities and assisting students to improve their writing, presentation, career, and time management skills.
These student survey results can be used to identify areas of strength and weakness in the first-year student experience. The results related to student satisfaction can also assist college leaders in setting appropriate goals and prioritizing initiatives for first-year students. Several steps are recommended based on the results: extending student orientation sessions, organizing more support service department workshops and training sessions, developing first-year seminars into more specialized courses, and providing professional development opportunities for the faculty members who are teaching first-year students.
An area for future research is to further explore the influence of other personal and attitudinal factors on students’ perceptions and satisfaction with the quality of their college experience and whether these associations differ by gender.
The data used to support the findings of this study are available from the corresponding author upon request.
The authors declare that they have no conflicts of interest.