The MTMM matrix approach: implications for HRM research
| DOI | https://doi.org/10.1108/PR-12-2014-0278 |
| Published date | 05 September 2016 |
| Date | 05 September 2016 |
| Pages | 1156-1175 |
| Author | Maria Riaz Hamdani,Sorin Valcea,Michael Ronald Buckley |
| Subject Matter | HR & organizational behaviour,Global HRM |
The MTMM matrix approach:
implications for HRM research
Maria Riaz Hamdani
Department of Management, The University of Akron, Akron, Ohio, USA
Sorin Valcea
Department of Management and Labor Relations, Cleveland State University,
Cleveland, Ohio, USA, and
Michael Ronald Buckley
Division of Management and Entrepreneurship, University of Oklahoma,
Norman, Oklahoma, USA
Abstract
Purpose –The purpose of this paper is to focus on the suitability of the multitrait-multimethod
(MTMM) matrix approach for examining construct validity in human resource management (HRM)
research. The authors also provide a number of suggestions on how to use MTMM more effectively in
HRM research.
Design/methodology/approach –The authors start by presenting a basic introduction to MTMM
approach. Next the authors briefly review the limitations of MTMM approach and suggested
improvements. The authors elaborate on these limitations by providing HRM examples. To further
illustrate these issues, the authors review employment interview research.
Findings –The construct validity analysis in HRM research suffers from three problematic
assumptions of the classical MTMM approach: uncorrelated trait-method units, uncorrelated methods,
and uncorrelated traits. The review of interview research shows that classical MTMM approach is
by far the most popular approach given its relative simplicity and modest sample size requirements.
This popularity stresses the significance of the review in highlighting these issues.
Originality/value –Several improvements to quantify the interpretations of MTMM analysis are
available to researchers. This review closely examines how these limitations and proposed
improvements influence HRM research, thereby making the methodological advances concerning the
MTMM approach more accessible to HRM researchers and practitioners.
Keywords Qualitative, Human resource management (HRM), Interview methods,
Recruitment and selection
Paper type Literature review
Accurate and valid measurement is a key concern in the human resource management
(HRM) discipline. Recru iting qualified employ ees, determining traini ng needs,
evaluating training outcomes, and assessing employee performance –all require
accurate and reliable measurement of constructs. However, many constructs in HRM
research cannot be observed directly (e.g. personality and job satisfaction), and instead
rely on subjective measures. In order to establish that such constructs are being
measured accurately and reliably, researchers need to provide evidence of construct
validity. Campbell and Fiske (1959) introduced the multitrait-multimethod (MTMM)
matrix as a tool to test the construct validity of psychological measures.
The MTMM matrix is based on a processof validating a focal construct by examining
the pattern of relationships with other constructs in its nomological network. Since the
introduction of MTMM approach, researchers have suggested many improvements to
quantify the interpretations of MTMM analysisand to address some of the limitations of
Personnel Review
Vol. 45 No. 6, 2016
pp. 1156-1175
©Emerald Group Publishing Limited
0048-3486
DOI 10.1108/PR-12-2014-0278
Received 10 December 2014
Revised 8 June 2015
8 August 2015
Accepted 24 October 2015
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/0048-3486.htm
1156
PR
45,6
the classical MTMM approach. In this paper, we discuss how these limitations and
proposed improvements in MTMM analysis influence HRM research. We believe
clarifying these issues for HRM research is necessary for the following reasons.
First, HRM research is context specific. While job performance models or selection
procedures may generalize across organizations to some extent, organization-specific
jobs, processes, or cultural norms need to be recognized in various HRM practices. For
example, each hiring company designs its own interview questions. As a result, often
times the operationalization of constructs across studies varies significantly (Bartram,
2005). Yet, in order to be valid, these organization-specific measurement instruments
have to tap into intended theoretical constructs. The MTMM approach can help
validate and integrate organization-specific measurement instruments with the broader
theoretical constructs. Additionally, in response to changes in technology,
demographics, and government regulations, HR-related functions need to be revised.
Consequently, the measurement of valid constructs becomes an ongoing endeavo r with
significant economic ramifications. This implies that HR practitioners should be
prepared to conduct MTMM analysis. Thus, understanding the issues surrounding the
classical MTMM approach is critical for both HRM researchers and practitioners.
Second, the field of HRM faces a particularly problematic research-practice divide
(e.g. Rynes et al., 2007). To help bridge this divide, it is important to propose
parsimonious theoretical models which can be translated into practice. Nevertheless,
construct ambiguity and construct proliferation issues are noticeable in HRM
research (Locke, 2012). MTMM analysis can aid in identifying redundant constructs
by highlighting significant overlap in their nomological networks. For instance,
Le et al. (2010) showed that job satisfaction and commitment are empirically
indistinguishable constructs. Such developments can produce more parsimonious
theories and facilitate practical applications.
Finally, the wealth of research available on the MTMM methodology is general in
scope and mostly presents solutions at the analysis stage. More attention needs to be
focussed on issues specific to HRM research at the design stage. The present study
takes such a focus and shows that many of the existing MTMM analysis approaches in
HRM research actually fail to meet basic assumptions of the MTMM method.
A significant number of issues emerge at the design stage (such as treating different
interviewers as distinct measurement methods), which complicate the analysis and
interpretation of MTMM. To resolve these issues we provide a number of suggestions.
Overall, our manuscript translates the general MTMM knowledge in a language that is
more meaningful and accessible for HRM researchers and practitioners.
MTMM can be a powerful tool in clarifying and improving core HRM constructs.
This requires understanding how the limitations and assumptions of MTMM approach
influence HRM research. With this goal in view, first, we briefly review the MTMM
method and its limitations, and then discuss how to prevent some of the more common
issues associated with the MTMM approach in HRM research. We conclude by
presenting theoretical and practical implications.
The MTMM matrix
Classical MTMM approach
In their classical article, Cronbach and Meehl (1955) state that construct validity is
required for any new assessment method which aims to measure a qualitative attribute
(e.g. personal initiative or persuasiveness); such assessments are commonplace in HRM.
For instance, valid performance evaluation measures are vital for diagnosing
1157
MTMM
matrix
approach
To continue reading
Request your trial