Examining Factor Structure of the Community of Inquiry Framework: A Bifactor Tactic

Jingwen LOW¹, Teck Kiang TAN²*, Yuan Yi CHONG¹, and Tao Tao NG³
¹Department of Chemistry, Faculty of Science (FOS), NUS
²Institute for Applied Learning Sciences and Educational Technology (ALSET), NUS
³NUS Library
*alsttk@nus.edu.sg

Low, J. W., Tan, T. K., Chong, Y. Y., & Ng, T. T. (2023). Examining factor structure of the community of inquiry framework: A bifactor tactic [Paper presentation]. In Higher Education Campus Conference (HECC) 2023, 7 December, National University of Singapore. https://blog.nus.edu.sg/hecc2023proceedings/examining-factor-structure-of-the-community-of-inquiry-framework-a-bifactor-tactic/

SUB-THEME

Interdisciplinarity and Education

KEYWORDS

Community of inquiry, factorial structure, exploratory factor analysis, confirmatory factor analysis, bifactor model

ABSTRACT

The Community of Inquiry (CoI) framework was commonly used for blended as well as online teaching and learning in higher education. This paper presentation examines the factorial structure of CoI, providing evidence to demonstrate that the bifactor factorial structure gives the most preferred conceptualising of the CoI framework. The findings confirmed that the One & Ten Bifactor Model fitted best, indicating the construct-relevant multidimensionality of CoI. This factorial structure is valuable for educators and researchers to understand students’ experiences through the application of CoI in a chemistry course. It also helps researchers intending to use the results for future research to use this validated factorial structure CoI that is capable of understanding the conceptualisation of CoI and its constructs.

FRAMEWORK OF COMMUNITY OF INQUIRY

The Community of Inquiry (CoI) framework was commonly used for blended as well as online teaching and learning in higher education. This framework, from the socio-constructivist perspective, is based on a meaningful educational conception of teaching, social, and cognitive presence. These three presences are subdivided into ten subcategories, as shown in Figure 1 below.

*Figure 1.* Conceptual framework and factorial structure of the Community of Inquiry (CoI).

The factorial structure defines the CoI dimensionality and its hierarchical relationship. This conceptual structure links the structural configuration to its assessment items. However, the factorial structure and its dimensionality are far from confirmation. While the application of the CoI framework has gained attention, was widespread, extended, and changed from its original intent, its factorial structure was not satisfactorily settled and agreed upon empirically. Stenbom (2018) reported the review results of 103 journals published between 2008 and 2017, which showed high validity and reliability; however, CoI factorial structure remained unanswered. The numerous factor structures suggested in the literature include the original three-category, and the ten-category structure. However, there is no evidence to show whether the three categories or the ten subcategories fit better empirically. Validation of the factorial structure thus requires further validation. The results of the study allow researchers and practitioners to use it with confidence as a reliable structural source. More importantly, the bifactor model, which is the main model proposed in the current paper, specifies simultaneously two structural forms that make this model closer to reality. While Yang and Su (2021) supported the bifactor model as a valid and reliable representation of the CoI instrument but it was restricted to a general factor with the three main presences without further examining the ten subfactors. This study proceeds to examine the inclusion of a general factor with the ten subfactors.

RESULTS

This paper presentation examines eight factorial structures of CoI, comparing their fits (Table 1), and finds that the One & Ten Bifactor Model fits best, as depicted in Figure 2 below. This model allows for specifying both the general and specific factors. The general factor is a general common model that is loaded to all the CoI items together with a set of specific factors that represent a unique narrow set of dimensions that concurrently and essentially contributed under the CoI framework

Table 1 shows the results of the eight models, indicating M1, the hypothesised One & Ten Bifactor, fits best for all the five fit indices, namely χ2, CFI, TLI, RMSEA, and SRMR.

Table 1
Fit indices of hypothesised CFA and competing models

Model	χ2	p-value	df	CFI	TLI	RMSEA	SRMR
M1 Hypothesised One and Ten Bifactor	711	0.3495	458	0.9981	0.9976	0.0489	0.0455
M2 Competing One-Factor	5607	0	527	0.9610	0.9585	0.2043	0.1222
M3 Competing Three-Factor Oblique	1555	0	524	0.9921	0.9915	0.0923	0.0634
M4 Competing Ten-Factor Oblique	912	0	482	0.9967	0.9962	0.0622	0.0521
M5 Competing Second-Order, Three-Factor First-Order	1555	0	524	0.9921	0.9915	0.0923	0.0634
M6 Competing Second-Order, Ten-Factor First-Order	3555	0	517	0.9767	0.9747	0.1595	0.1022
M7 Competing Third-Order, Ten-Factor First-Order, Three-Factor Second-Order	1183	0	493	0.9947	0.9940	0.0778	0.0537
M8 Competing One and Three Bfactor	929	0	487	0.9966	0.9961	0.0627	0.0488

Note:

χ2=Chi-Square Statistics; df=Degrees of Freedom; CFI=Comparative Fit Index; TLI=Tucker-Lewis Index; RMSEA= Root Mean Square Error of Approximation; SRMR=Standardized Root Mean Square.

* p <0.01

CONCLUSION

While it is not out of expectation that the bifactor generally gives a better fit (Caci et al, 2015; Chen et al, 2006; Reise, 2012), it has been rediscovered and returned as an effective approach to modeling construct-relevant multidimensionality in a set of items (Reise, 2012). As the application of bifactor models remained poorly understood in the CoI research communities, the basic argument is that the utility of the bifactor model is capable of resolving important issues in conceptualising and measuring constructs. The current paper adds to the literature by providing evidence to demonstrate that the bifactor not only provides a better fit statistically but also provides a strong foundation for conceptualising the CoI framework and uses all ten subdimensions for measurement.

REFERENCES

Chen, F. F., West, S. G., & Sousa, K. H. (2006). A comparison of bifactor and second-order models of quality of life. Multivariate Behavioral Research, 41(2), 189–225. https://doi.org/10.1207/s15327906mbr4102_5

Caci, H., Morin, A. J. S., & Tran, A. (2015). Investigation of a bifactor model of the strengths and difficulties questionnaire. European Child & Adolescent Psychiatry, 24(1), 1291–1301. http://dx.doi.org/10.1007/s00787-015-0679-3

Reise, S. P. (2012). The rediscovery of bifactor measurement models. Multivariate Behavioral Research, 47(5), 667-96. http://dx.doi.org/10.1080/00273171.2012.715555

Stenbom, S. (2018). A systematic review of the Community of Inquiry survey. The Internet and Higher Education, 39, 22–32. https://doi.org/10.1016/j.iheduc.2018.06.001

Yang, H., & Su, J. (2021). A construct revalidation of the community of inquiry survey: Empirical evidence for a general factor under a bifactor structure. International Review of Research in Open and Distributed Learning, 22(4), 22-40. https://doi.org/10.19173/irrodl.v22i4.5587