Testing multi‐group measurement invariance of data from the knowledge creation practice inventory

• Author: Shinhee Jeong,Yu‐Yu Hsiao,Jihoon Song
• Published date: 01 September 2018
• Date: 01 September 2018
• DOI: http://doi.org/10.1002/hrdq.21323

QUANTITATIVE STUDY

Testing multi-group measurement invariance of

data from the knowledge creation practice

inventory

Shinhee Jeong

1

| Jihoon Song

2

| Yu-Yu Hsiao

3

1

Louisiana University, Baton Rouge, Louisiana

2

Educational Technology, Hanyang University,

Seoul, Korea

3

Center on Alcoholism, Substance Abuse, and

Addictions, University of New Mexico,

Albuquerque, New Mexico

Correspondence

Jihoon Song, Professor, Hanyang University,

Educational Technology, Seoul, Korea.

Email: psu.jihoonsong@gmail.com

This study examined the extent to which the factor structure of

the Knowledge-Creation Practice Inventory (KCPI) is invariant

across different occupational groups in South Korea, based on

the studies of Song et al in 2011 and 2012. Inspired by Nona-

ka's (1994) knowledge-conversion theory, the inventory consists

of four dimensions (i.e., knowledge sharing, creating concepts,

justifying concepts, and building prototypes), measured by

11 items. A total of 2,364 responses collected from school

teachers (n= 1,864), public-sector employees (n= 214), and

private-sector employees (n= 286) were utilized to examine the

measurement invariance, using a series of hierarchical multigroup

confirmatory factor analyses. The results indicated that the four-

factor, 11-item measurement model is equivalent across the

three groups, supporting configural invariance, metric invariance,

scalar invariance, and factor variance/covariance invariance,

except factor mean invariance. A series of post-hoc tests were

additionally conducted to identify the factor mean differences

among the three groups. Based on the results of this study,

future researchers should use more diverse samples including a

wide range of cultures or different subcultural groups

(e.g., occupations, genders, and educational backgrounds) to

expand the universality of the factor construct. Relevant organi-

zations can utilize the inventory to assess their current capacity

of knowledge creation and design for appropriate human

resource (HR) or organizational development interventions to

nurture and enhance employees' learning-related behaviors. Since

the generalizability and validity of the KCPI were supported in

this study, at least among these three groups, more reliable, rig-

orous group comparisons are available for future research utiliz-

ing this inventory.

DOI: 10.1002/hrdq.21323

Human Resource Development Quarterly. 2018;29:243–262. wileyonlinelibrary.com/journal/hrdq © 2018 Wiley Periodicals, Inc. 243

KEYWORDS

knowledge creation, measurement invariance/equivalence, South

Korea

Organizational knowledge creation through conversion between tacit and explicit knowledge is recognized as

one of the most essentia l driving forces for per formance-oriente d human resource devel opment (HRD) and thus

has received substantial attention in many types of organizations (Gourlay, 2001; Nonaka, Toyama, & Konno,

2000; Song & Chermack, 2008). Knowledge is an invaluable asset, and creating new knowledge is crucial to

ensure creative and innovative performance (Drucker, 1999; Nonaka & Takeuchi, 1995). It has been argued that

in this era, different types of knowledge are demanded for every worker or o ccupation; theref ore, it is difficult

to compare knowledge intensity (Choo & Bontis, 2002; Von Krogh, Ichijo, & Nonaka, 2000). That is, under-

standing the relevan t phenomena is not only im portant to for-prof it enterprises, but al so to all organizati ons,

regardless of the organizational or industrial type (e.g., nonprofit, educational settings) (Song, Bae, Park, &

Kim, 2013).

Because of its importance, there is a need to expand our understanding of the relevant phenomena, includ-

ing what promotes or hin ders knowledge-cr eating activities i n organizations, an d the interplay betwe en those

factors and different contexts or populations. To make such scientific, objective, and systematic investigations

possible, appropriate instrument development is necessary as it contributes to both theory building and practi-

cal application (DeVellis, 2012; Netemeyer, Bearden, & Sharma, 2003). Compared to the theoretical maturity of

concepts around know ledge-creation the ory, however, sever al authors have noted t he paucity of academic

efforts to validate the instrument from a quantitative research standpoint (Lloria & Moreno-Luzón, 2005;

Schulze & Hoegl, 2006). The Knowledge-Creation Practice Inventory (KCPI) (Song, 2008; Song, Uhm, & Yoon,

2011; Song, Yoon, & Uhm, 2012) has been one of the few efforts to capture and measure relevant activities

for knowledge creati on.

There has been an increasing emphasis on examining measurement invariance to ensure that “respondents from

different populations exhibit similar cognitive frameworks when interpreting and responding to a given measure”

(Hu, Pellegrini, & Scandura, 2011, p. 275). Researchers have argued that most group-comparison studies have

neglected or already assumed equivalence of the factor structure of their instruments across groups (Byrne, Shavel-

son, & Muthén, 1989; Ryan, Chan, Ployhart, & Slade, 1999; Steinmetz, Schmidt, Tina-Booh, Wieczorek, & Schwartz,

2008). However, it is crucial that a measure retains the same meaning across groups; otherwise, group comparison is

meaningless and erroneous, like comparing apples to oranges (Hoyle & Smith, 1994). Failure to achieve measurement

invariance implies that differences in groups potentially do not reflect true differences and are driven from the differ-

ent functioning of the instrument across groups (Millsap & Kwok, 2004). Thus, measurement invariance must be

accomplished before making any conclusions about a group comparison to develop a universal measurement for

research and practical applications.

For example, Wasti, Tan, Brower, and Önder (2007) assessed the measurement invariance of the trust

scales developed by Mayer and Davis (1999) across three cultures (i.e., United States, Turkey, and Singapore)

and found that the respondents from collectivist-high-power distant cultures interpreted several items differ-

ently from responde nts from individual ist-low-power dist ant cultures. As a re sult, they conclude d that culture

plays an important r ole in the operation o f the measure. Sinc e individuals from d ifferent cultures respond to

items in different way s, they recommended developing scales that are mo re applicable to mult iple cultures, not

a single culture.

244 JEONG ET AL.