SATISFACTION AND SELF-PERCEPTIONS: HOW ARE THE RELATED? STEVE GRAUNKE INDIANA UNIVERSITY-PURDUE UNIVERSITY INDIANAPOLIS

PRESENTATION Discuss the relationship between academic satisfaction and self- perceptions of learning Explore a model of academic satisfaction and self-perceptions of learning using the IUPUI Continuing Student Survey Demonstrate statistical model features in Mplus (that SPSS and SAS can’t do as well) Discuss implications

BACKGROUND Academic Satisfaction Indirect assessment

ACADEMIC SATISFACTION Edwards and Waters (1982) Satisfaction and GPA predict persistence Schreiner (2009) Comprehensive study of satisfaction and retention Pullins (2011) Sophomore students Satisfaction with… Campus Climate Advising

INDIRECT ASSESSMENT Direct Assessment demonstrations of skills Indirect Assessment Feelings about learning Perceptions of abilities

HOW THESE CONSTRUCTS RELATED? Student- Faculty interaction Grades Academic Persistence

RECIPROCAL CAUSATION BETWEEN SATISFACTION AND PERFORMANCE Suggests satisfied employees perform better (and vise versa) Work organization literature Based on organizational literature Satisfaction stronger Bean & Bradley (1986) Reciprocal causation model between grades and satisfaction fits well Satisfaction “wins” Pike (1991)

CURRENT STUDY Does reciprocal causation better describe the relationship between academic satisfaction and self-perceptions of learning than models in which academic satisfaction predicts self-perceptions (or self-perceptions predicts academic satisfaction)?

IUPUI CONTINUING STUDENT SURVEY Multipurpose survey α = “Overall, how satisfied are you with your academic experiences at IUPUI?” “How satisfied are you with the quality of the academic programs at IUPUI?” Overall Academic Satisfaction Communication Skills (4 items) Quantitative Skills (4 items) Exponent transformation Principles of Undergraduate Learning Indirect Assessment

LATENT VARIABLES Variables that are not observed directly Indicators Example: Socio-economic status Advantages Accounts for measurement error More consistent parameter estimates Disadvantages Large samples May need more complicated software to do this

THE LATENT VARIABLE PART Read and understand books, articles, and instruction manuals (n801) Formally communicate ideas and information (oral, visual, aural, etc.) (n802) Write a final report on a project or other work assignment (n803) Communicate with a team to solve problems (n804) Solve mathematical problems (n805) Use mathematics in everyday life (n806) Understand a statistical report (n807) Support an argument using quantitative data (n808)

STUDY 521 Senior respondents3 Exogenous observed variables Gender (65% female) IU Cumulative GPA STEM major Structural Equation model

MODEL 1/ SATISFACTION PREDICTS SELF- PERCEPTIONS Gender (Flag for female) IU Cumulative GPA STEM major Academic Satisfaction Quant. skills Comm. skills n801 n802 n803 n804 n805 n806 n808 n807

MODEL 2/ SELF- PERCEPTIONS PREDICT SATISFACTION Gender (Flag for female) IU Cumulative GPA STEM major Academic Satisfaction Quant. skills Comm. skills n801 n802 n803 n804 n805 n806 n808 n807

MODEL 3/ RECIPROCAL CAUSATION Gender (Flag for female) IU Cumulative GPA STEM major Academic Satisfaction Quant. skills Comm. skills n801 n802 n803 n804 n805 n806 n808 n807

MODEL ESTIMATION Maximum Likelihood estimation (ml) Enables clearer comparisons between models Chi-square difference test

DESCRIPTIVE STATISTICS FOR OBSERVED VARIABLES VariablesNMeanVariance Satisfaction with overall academic experiences a Read and understand books, articles, and instruction manuals b Formally communicate ideas and information (oral, visual, aural, etc.) b Write a final report on a project or other work assignment b Communicate with a team to solve problems b Solve mathematical problems b Use mathematics in everyday life b Understand a statistical report b Support an argument using quantitative data b Gender (Flag for Female) GPA STEM major a Scale for included items: 1 = Very Dissatisfied, 2= Dissatisfied, 3= Neutral, 4= Satisfied, 5= Very Satisfied b Scale: 1 = Not at all Effective, 2= Somewhat Effective, 3= Effective, 4= Very Effective. Exponent transformation used on these items. Means and variance account for transformation.

MODEL FIT STATISTICS Null model Model 1Model 2Model 3 Chi-Square Dof RMSEA CLI TLI

RESULTS Satisfaction does impact self-perceptions Self-perceptions do impact satisfaction All models fit data Not statistically significant improvement Significant effects

SO WHAT AM I GOING TO DO WITH THIS? Four page Research Brief What interventions effect both satisfaction and self-perceptions? Other factors?

FOR MORE INFORMATION ON SEM AND MPLUS… Byrne, B. M. (2012). Structural equation modeling with Mplus. New York, NY: Routledge. Any Questions?