Sung, S. Y., & Choi, J. N. (2014). Multiple dimensions of human resource development and organizational performance. Journal of Organizational Behavior, 35(6), 851‐870. https://doi.org/10.1002/job.1933

• Published date: 01 May 2020
• DOI: http://doi.org/10.1002/job.2428
• Date: 01 May 2020

CORRIGENDUM

Sung, S. Y. |& Choi, J. N. (2014). Multiple dimensions of human

resource development and organizational performance. Journal

of Organizational Behavior, 35(6), 851‐870. https://doi.org/

10.1002/job.1933

The authors have provided the following corrections and clarifications

to help readers better understand the reported findings and their

interpretations:

First, the first confirmatory factor analysis (CFA) reported on page

860 merits additional information and correction. The correct degrees

of freedom (df) for a CFA model with eight indicators for four corre-

lated human resource development (HRD) factors, of which two had

a single indicator (i.e., resource investment in HRD and employee

HRD exposure), should be 16. The df of this CFA became 15 as we

allowed two error terms to correlate because as reported in the mea-

sures section (pp. 857–858) two of the four items comprising the scale

of “management support for HRD”were exactly the same (i.e., “Our

company has a strong dedication to develop highly competent

employees”) but each rated by HRM directors and employees. This sit-

uation involving the same item rated by two sources generates a sig-

nificant method factor that cannot be accounted for by the

substantive latent factor (in this case, management support for

HRD). This measurement property requires the consideration of a

method factor in conducting CFA by allowing the correlated errors

between the two items with the same wording rated by two sources

(Brown, 2015), which reduced the df from 16 to 15. In addition, for

this CFA model, the correct CFI and RMSEA values were .94 and

.083 instead of .95 and .063, respectively.

Second, on page 860, we stated that “This model [HRD as a higher

order factor model] produced a significantly better model fit than the

four‐factor measurement model (Δχ2(Δdf = 1) = 16.46, p< .001).”This

statement was inappropriate because (a) the four correlated HRD

dimensions model and (b) the model with a higher order HRD factor

indicated by four HRD dimensions cannot be directly compared statis-

tically in terms of their model fit or through chi‐square comparisons

because the two CFA models are not nested to each other. Unlike

the model of four correlated HRD dimensions, in the model placing

HRD as a second‐order factor indicted by four HRD dimensions, the

four dimensions are not correlated to each other. In other words,

the two models represent completely distinct measurement hypothe-

ses that cannot be compared against each other instead of having

nested relationships to allow statistical comparisons of their degrees

of model fit. Considering that these two CFA models exhibited accept-

able and comparable levels of model fit, in the current data, the possi-

bility of having the four correlated HRD dimensions is equally likely to

the possibility of having a higher order HRD factor indicated by the

four HRD dimensions.

REFERENCE

Brown, T. A. (2015). Confirmatory factor analysis for applied research.

Guilford Press.

DOI: 10.1002/job.2428

J Organ Behav. 2020;41:403. © 2020 John Wiley & Sons, Ltd.wileyonlinelibrary.com/journal/job 403