Measuring State, Effect and Response Uncertainty: Theoretical Construct Development and Empirical Va

1
Measuring State, Effect and Response Uncertainty
Theoretical Construct Development and Empirical
Validation

• Nick Ashill, AUS, UAE and Victoria University of
  Wellington, New Zealand
• David Jobber, School of Management, Bradford
  University, UK

2
Presentation Structure

• Background, Context and Research Gap
• Research Objectives
• Scale Development Process
• Data Analysis and Results
• Conclusions and Future Research

3
Background and Context

• The concept of Perceived Environmental
  Uncertainty (PEU)
• Application as a contingency variable
• Marketing (Achrol and Stern, 1988 Bstiler and
  Gross, 2003)
• Strategic Management (Boynton et al., 1993
  Sarker et al., 2001)
• Information Systems (Mangaliso, 1995 Choe, 1993)
• Accounting (Chenhall and Morris, 1986 Lal and
  Hassel, 1998)
• The Information Uncertainty Perspective (Bernard
  1938)
• Measurement problems
• Millikens (1987) critique of the PEU literature
• The Concepts of State, Effect and Response
  Uncertainty

4
Research Gap

• No full and rigorous psychometric development and
  testing of scales to measure the three constructs
• Gerloff et al. (1991)

5
Research Objective

• To develop valid and reliable scales for
  measuring Millikens (1987) three types of
  environmental uncertainty
• Antecedents and Outcomes
• A study of Senior Marketing Executives (SMEs) as
  boundary spanning organizational members

6
Scale Development Process

• Exploratory Research - an in-depth exploratory
  study (key informant survey with 20 senior
  marketing executives).
• Domain Specification and Item Generation
• The domain of each construct was defined and an
  initial set of 17 items was developed from (1) a
  review of the PEU literature and (2) the results
  of the exploratory research
• State PEU 5 items
• Effect PEU 5 items
• Response PEU 7 items
• Instrument Pretesting and Data Collection
• Phase 1
• Use of academic researchers as expert judges
• A total of 12 items survived this content
  validation process

7
Meta Matrix SME Uncertainty Perceptions
8
Scale Development Process

• Instrument Pretesting and Data Collection (Contd)
• Phase 2
• Pretest using a protocol approach with senior
  marketing executives and a separate mail survey
• Following rewording, the questionnaire was mailed
  to a sample of 568 senior marketing executives in
  New Zealand manufacturing and service businesses
• 204 usable responses
• Cross-industry pool ensured sample of SMEs faced
  a wide variety of environmental conditions which
  allows for maximum variance in uncertainty
  perceptions.
• Non-response analysis

9
Data Analysis and Results

• Evaluative Tests used to assess the Psychometric
  Properties of the PEU Scales (DeVellis, 2003)
• Sample 1 (N120) Item reduction and scale
  refinement using EFA
• Sample 2 (N84) Holdout sample for scale
  validation. Tests for convergent and
  discriminant validity performed using CFA.
• Nomological validity (linking the scales to other
  theoretically relevant constructs via a
  literature review using the SEM-based PLS
  methodology (PLS Graph version 3.00)

10
Exploratory Factor Analysis of PEU Items
11
Sample 2 Scale Validation
12
Data Analysis and Results

• Convergent and Discriminant Validity
• Convergent Validity was assessed with 3 measures
• Item Reliability (Chin 1998)
• Construct Reliability (Werts et al 1974)
• AVE (Fornell and Larcker 1981)
• Discriminant Validity
• Item loadings examined to ensure that no item
  loaded higher on another construct than it did on
  the construct it was intended to measure (Fornell
  and Larcker 1981)
• Average variance shared between constructs/their
  measures were compared to the variances shared
  between the constructs

13
Measurement Model Items
14
Structural (Inner) Model Results
15
Data Analysis and Results

• Structural Model Results
• All but one of the path coefficients were
  significant (plt0.05) providing support for the
  nomological validity of the measures
• Method Biases
• Harman one-factor test (Podsakoff et al., 1984)
• Marker-variable technique (Lindell and Brandt,
  2000)

16
Conclusions and Future Research

• Based on Millikens (1987) conceptualisation and
  an in-depth exploratory study we suggest that
  senior marketing executives experience different
  types of uncertainty as they make SENSE,
  INTERPRET and RESPOND to the external marketing
  environment.
• Our findings show that the measures of state,
  effect and response are psychometrically sound.
• The results of the a priori analysis indicated
  that 9 of the 12 uncertainty items load
  significantly onto three factors.
• The PLS CFA model also suggests that PEU should
  be represented by three separate components.
• The decomposition of PEU produces different
  information seeking behaviors

17
Contribution and Future Research

• The existence of reliable and valid scales to
  measure the three uncertainty constructs has high
  potential for academic research on managerial
  decision-making.
• In marketing research we believe our research
  represents a first attempt in validating
  Millikens (1987) conceptualization.
• Replication of the scales in different settings
• A platform to initiative empirical assessment of
  the antecedents and outcomes of different types
  of uncertainty
• Possible antecedents include
• example individual differences, organizational
  characteristics, resource dependence
  characteristics
• Possible behavioural outcomes include
• The contingency-information characteristic
  relationship and its consequent effect on
  managerial performance

18
Questions??