Confirmatory factor analysis of the principal self-efficacy survey.

Smith, R. Wade ; Guarino, A.J.

ABSTRACT

This article describes the development and constructs validity of the Principal Self-Efficacy Survey (PSES). The item selection was based on the theoretical framework proposed by Bandura. Fourteen-items assessing two factors Instructional Leadership (nine items) and Management Skills (five items) and a demographic questionnaire comprised the PSES. Items were scored on a 1 to 4 Likert-type scale. Participants were two hundred eighty-four principals. Construct validity was supported by confirmatory factor analysis using AMOS 5.0. In conclusion, the PSES provides a promising measure of principal perceptions of their ability to effectively function in the areas of instructional leadership and management.

INTRODUCTION

Bandura (1997) defines self-efficacy as: "... beliefs in one's capabilities to organize and execute the courses of action required to produce given attainments" (p.3). According to Bandura, self efficacy influences, (1) the courses of action people choose to pursue, (2) how much effort people will put forth in a given endeavor, (3) how long they will persevere in the face of obstacles and failure, (4) people's resilience to adversity, (5) whether someone's thought patterns are self-hindering or self-aiding and (6) how much stress and depression is experienced in coping with taxing environmental demands.

The central role of self-efficacy in human agency makes it an important and useful construct for empirical research. Because self-efficacy is a task-specific construct (Bandura, 1997), any attempt to measure self-efficacy should be contextually sensitive to the setting in which the behaviors occur. A rich and robust body of literature documents the relationships between self-efficacy beliefs for teachers and students and their relationship to teaching and learning (e.g., Pajares, 1996; Tschannen-Moran, Hoy, and Hoy, 1998). However, a literature search for journal articles on principal self-efficacy and instructional effectiveness produced no articles specific to the topic. Currently there is tremendous interest in the role of the principal in affecting substantive, long-term improvement in schools. For example, the federal government, in The No Child Left Behind Act has weighed in with a mandate that principals in poorly performing schools shall be replaced if improvement is not forthcoming.

Given the central role that principals are expected to perform in maintaining quality teaching and learning environments in schools, it is important to begin to conceptualize and operationalize measures of principal self-efficacy. The following sections detail the development of the Principal Self-Efficacy Survey (PSES) along with its attendant psychometric properties.

ITEM GENERATION

The generation of items for the PSES used the rational-empirical approach to instrument development (Burisch, 1984). The rational component drew upon the knowledge and experience of professionals working as principals and the research literature to suggest potential items. The empirical component selected or rejected items based on their psychometric properties. The scale configuration was based on the theoretical framework proposed by Bandura. Fourteen-items assessing two factors Instructional Leadership (nine items) and Management Skills (five items) and a demographic questionnaire comprised the PSES. Items were scored on a 1 to 4 Likert-type scale.

ITEM SELECTION

The 14 items were then checked for violations of normalcy through the SPSS Statistical Package Version 11.0 (SPSS Inc., 2001), explore function. Items would be considered for elimination if they had a skew value equal or greater than two and kurtosis value equal or greater than seven.

PARTICIPANTS

Two hundred and eighty-four principals returned completed and valid surveys representing twelve states (5 in the southeast, 2 in the Midwest, 2 in the west, 2 in the northeast, and Alaska). There are 74 elementary schools, 30 middle schools, and 31 high schools represented in this study. Sixty-six percent of the respondents are males. Ethnic representation included 83% white, 14% black, and 1.4% other. Nearly 47% of the respondents indicated that they have a master's degree plus 30 hours and approximately 10% of respondents have an earned doctorate. The majority of the responses (54%) came from rural schools, while 17% were from suburban schools and 25% were from urban schools

RESULTS

Because missing data appeared to be randomly scattered among the variables, a full information maximum likelihood (FIML) imputation was performed to estimate missing data. The factor structures were examined using a confirmatory factor analysis. A series of models were tested in the following order: (a) a single-factor g model in which all items were free to load on only one common factor; (b) an orthogonal two-factor model in which each factor was set to be independent of each other; (c) a correlated two-factor model in which the factors were to each other. The first two models were included to aid in the assessment of the correlated two- factor model.

The models were examined by AMOS version (5.0) maximum likelihood factor analysis (Arbuckle, 2004). The models were evaluated by a variety of fit measures that are classified as absolute, relative, parsimonious, and population discrepancy. Absolute fit measures assess how well the proposed interrelationships among the variables match the interrelationships among the actual interrelationships. The measure of absolute fit used in this study was the chi-square test because AMOS does not provide other absolute measures when missing data is estimated with the FIML imputation procedure. Measures of relative fit compare the hypothesized model to the null model. The relative fit measures employed in this study were the Comparative Fit Index (CFI) (Bentler, 1990), the Tucker-Lewis Index (TLI) (Bentler and Bonett, 1980). Measures of parsimonious fit attempt to determine if the overall fit of the model has been accomplished by overfitting the data. The parsimonious fit measure in this study was the chi-square divided by the degrees of freedom. Lastly, population discrepancy measures are estimates from the sample coefficients to the population coefficients. The population discrepancy measure in this study was the Root Mean Square Error of Approximation (RMSEA) (Browne and Cudeck, 1993). Models were compared by examining differences in values of chi-square to identify statistically significant variations among the models. The fit indices for the three models are presented in Table 1.

The chi-square test for differences revealed that the correlated two-factor model is superior to the other models. The correlated two-factor model yielded acceptably high goodness of fit indices (i.e., > .99) for both the CFI and the TLI. The RMSEA achieved a value of .049 indicating a close fit between the sample coefficients and the estimated population coefficients. The correlation between the two factors is .69 demonstrating discriminate validity.

The factor loadings are provided in Table 2. All items loaded statistically significantly (p < .01) and demonstrated practical significance with loadings greater than .40 on their respective factors.

CONCLUSION

This study provides empirical evidence that the PSES operationalizes the latent constructs of instructional leadership and management skills for principals. Individual items demonstrated construct validity, (i.e., the items were shown to measure their respective hypothetical construct and factor loadings were all significant, p < .01). The instructional leadership and management constructs are both considered essential to principal effectiveness and as such, the PSES provides a promising measure for furthering understanding of self-beliefs of principals.

Because this research was exploratory in nature, further research is suggested to replicate the initial results. Also, future research should attempt to determine if the factor structure holds for various levels of the principalship (i.e., elementary, middle, and high school). Future research incorporating other important elements of principal self-efficacy beliefs (e.g., conflict resolution) would also be suggested. Finally, it would be important to understand principal self-efficacy for instructional effectiveness within the broader context of constructs known to be important for creating and facilitating an effective learning environment in schools. With this in mind, future studies should investigate the relationships between principal self-efficacy and other important constructs such as school culture, teacher self-efficacy, and student self-efficacy.

APPENDIX A: PRINCIPAL SELF-EFFICACY SURVEY

This administrator survey asks you to make a series of judgments about your experiences as a head administrator for a school. You are asked to read the following items and rate the strength of your beliefs in your abilities to attain the following outcomes. These items should be answered from your perspective as a school principal working to produce an effective teaching and learning environment. You are to indicate the degree to which you agree or disagree with each statement by darkening the appropriate oval.

Scale 1 = Very Weak Beliefs in My Abilities (VW)
 2 = Weak Beliefs in My Abilities (W)
 3 = Strong Beliefs in My Abilities (S)
 4 = Very Strong Beliefs in My Abilities (VS)

STATEMENTS:

My beliefs in my abilities to ...

1. influence teachers to utilize effective teaching and learning practices are

2. provide effective modeling for teachers regarding effective teaching and learning practices are

3. use research on teaching and learning to guide strategic planning for accomplishment of school goals are

4. plan effective activities and experiences which facilitate teachers' beliefs in their abilities to provide effective teaching and learning activities to their students are

5. use data collected from teacher observations to inform school-wide efforts for improving teaching and learning are

6. regularly perform effective observations of teachers are

7. stay abreast of current best practices for facilitating effective teaching and learning are

8. communicate needs and goals necessary to enhance effective instructional effectiveness to faculty are

9. provide experiences that foster and facilitate high levels of teacher motivation towards teaching and learning are

10. protect instructional time so that effective teaching and learning can take place

11. facilitate an atmosphere that provides fair and consistent discipline for all students are

12. maintain healthy school/community relations are

13. maintain a school-wide atmosphere that is conducive to teaching and learning are

14. buffer teacher from unnecessary paperwork

REFERENCES

Arbuckle, J.L. (1999). Amos 4.0 User's Guide. Chicago: Small Waters Corporation.

Bandura, A. (1997). Self-efficacy: the exercise of control. New York: Freeman.

Bandura, A. (2001). Guide for constructing self-efficacy scales (Revised). Available from Frank Pajares, Emory University.

Bentler, P.M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246.

Bentler, P.M. and Bonett, D.G. (1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88, 588-606.

Browne, M.W. and Cudeck, R. (1993). Alternative ways of assessing model fit. In Bollen, K.A. and Long, J.S. [Eds.] Testing structural equation models. Newbury Park, California: Sage, 136-62.

Burisch, M. (1984). Approaches to personality inventory construction: a comparison of merits. American Psychologist, 39(3) 214-227.

Pajares, F. (1996). Self-efficacy beliefs in academic settings. Review of Educational Research, 4 (66) 4. 543-578.

Tschannen-Moran, M.., Hoy, A.W., & Hoy, W. K. (1998). Teacher efficacy: Its meaning and measure. Review of Educational Research. 68(2) 202-248.

R. Wade Smith, Louisiana State University

A. J. Guarino, Auburn University

Table 1. Fit Indices for the Three Models

Factor Model c2 df c2 / df

Single (g) 180.37 * 77 2.34
Orthogonal 218.60 * 77 2.84
Correlated 127.1 * 76 167%

Factor Model CFI TLI RMSEA

Single (g) 993 .991 .069
Orthogonal 991 .987 .081
Correlated 997 .995 .049

* p < .05

Table 2. Item factor loadings
Factor Loadings of the Principal Efficacy Survey

Item Instructional Leadership Management Skills

Q1 .69
Q2 .62
Q3 .59
Q4 .65
Q5 .66
Q6 .64
Q7 .59
Q8 .65
Q9 .61
Q10 .66
Q11 .77
Q12 .47
Q13 .58
Q14 .44