Summer 2003 Research Grant Results Presentation

1
Summer 2003 Research Grant Results Presentation

• Robert C. Hoell, Ph.D., SPHR
• Dept.of Management, Marketing, Logistics
• College of Business Administration

2
Proposal Title

• THE DEVELOPMENT OF SCALES APPROPRIATE FOR THE
  MEASUREMENT OF INTERPERSONAL TRUST AND
  PARTICIPATION ATTITUDES OF UNION MEMBERS

3
Abstract from Original Study

• Employees working in unionized environments have
  the unique position of having their loyalty
  courted by both their employer and their union.
  Some employees form a loyalty to both while some
  remain uncommitted to one or the other, or in
  some instances, to neither. There is no strong
  research evidence that explains these
  differences.
• It is hypothesized that interpersonal trust may
  explain these varying levels of employee
  commitment to their employer and their union.
  Initial findings indicate that individuals with
  low levels of trust do not form as high a degree
  of commitment as those who have high levels of
  trust.

4
Hypotheses from Original Study (these are still
valid for next iteration)

• Union Members with greater degrees of
  Interpersonal Trust will exhibit higher degrees
  of Union Commitment than members with lesser
  degrees of Interpersonal Trust.

• Union Members with positive attitudes towards
  participation will exhibit higher degrees of
  Union Commitment than members with less positive
  attitudes towards participation.

5
Abstract from Current Study

• This research will establish scale items best
  attuned to the measurement of union member
  commitment, trust, and participation attitudes.
  Existing scales that measure trust and
  participation have too strong of an
  organizational theme. These pro-management
  scales are confounded by the degree of
  commitment union members hold. Respondents with
  high union commitment view these scales as
  representing a managerial view and answer them
  from a negatively biased perspective. Through
  rewording, factor analysis, and re-sampling, new
  scale instruments and survey questions, devoid of
  any perceived managerial influence, can be
  developed.

6
One Step Further

• The data led to mixed results. Further
  inspection of the data, through factor analysis,
  yielded additional mixed results.
• It was determined that a second order factor
  analysis, conducted through SEM and the Lisrel
  Software Package, would be necessary before
  rewording and restructuring could continue.
• Todays presentation will focus on the
  re-analysis of the original data and the
  conclusions that can be drawn from the second
  order technique.

7
Trust Scales Used

• For this study, general measures of interpersonal
  trust were used. John Rotter (1967) developed a
  25-item scale that measures general trust in
  others. Entitled Interpersonal Trust Scale the
  items measure overall trust with 5-point Likert
  responses. The other scale used for this study
  is the Trust in People scale developed by The
  Survey Research Center in 1969. Again, the scale
  measures overall trust by an individual in other
  individuals. Three items are used, with a forced
  choice response to each. One of the two choices
  to each question indicates a trusting response.
  Both scales are additive, with higher scores
  indicating greater degrees of Interpersonal Trust.

8
Participation Scales Used

• Four distinct measures of Employee Participation
  were used in the survey. Participativeness is
  measured with a 12 item scale (? 0.85)
  developed by Verma and McKersie, 1987. This
  scale measures the desired participation in
  decision-making and the level of say employees
  want. The Interest in Work Innovation Index was
  developed in 1965 by Martin Patchen. This is a 6
  item scale, with reported test-retest
  reliabilities from .87 to .92. The scale
  measures the degree to which respondents are
  interested in finding new ways of doing things at
  their job.  Job Involvement is measured with a
  scale modified by Lorence and Mortimer from
  original scales developed by Quinn and Staines in
  the mid-70s. It consists of 7 items measuring
  effort and involvement attitudes. A locus of
  control scale, developed by Spector in 1988,
  attempts to overcome the limitations of the
  original developed by Rotter (1966). The Work
  Locus of Control Scale uses concepts and beliefs
  found in work settings, as opposed to the more
  general Rotter items.

9
Union Commitment Scales Used

• Union commitment was originally defined with 48
  items (Gordon, et al., 1980) 30 of these items
  have been identified as most salient in capturing
  aspects of commitment. Reliability for the scale
  was not reported in the original article, but
  additional research using subsets of the original
  scale has shown alpha values generally stable
  0.83 (Iverson Kuruvilla, 1995), 0.85 (Martin,
  Magenau Peterson, 1982), and 0.88 (Sherer
  Morishima, 1989). Four different factors within
  the scale have been identified union loyalty,
  responsibility to the union, willingness to work
  for the union, and belief in unionism. Fields
  Thacker (1992) reported the alphas on each of
  these four factors 0.89 for Loyalty to the
  Union, 0.72 for Responsibility to the Union, 0.80
  for Willingness to Work for the Union, and 0.82
  for Belief in Unionism.

10
Demographic Results
Age Range 23 to 65 Average 45.41
Gender 112 Male 19 Female
Ethnicity 110 White, 10 Black, 7 Hispanic
Education 55 beyond a High School degree
11
Structural Equation Model Conceptual Diagram
WLC
UNLOYAL
PARTIC
Trust
UNRESP
IT
UnCommit
JI
PartAtt
UNWORK
TIP
UNBELIEF
WII
12
Basic Model with T-Values(non-significant values
in red)
Chi-Square53.17, df32, P-value0.01078
WLC
7.06
4.77
UNLOYAL
7.33
PARTIC
8.26
3.66
Trust
5.08
0.77
UNRESP
IT
UnCommit
6.40
-1.80
7.31
-2.53
-4.65
JI
UNWORK
PartAtt
7.77
-3.60
TIP
0 15.
UNBELIEF
4.45
WII
3.34
-6.67
13
Structural Model with T-Values
0.00
Trust
0.77
Union Commit- ment
0.00
3.58
Partic- pation Attitudes
-2.53
0.00
14
Results

• SEM Fit Indices
• Root Mean Square Residual (RMR) 0.198
• Standardized RMR 0.0745
• Goodness of Fit Index (GFI) 0.930
• Adjusted Goodness of Fit Index (AGFI) 0.879

15
What to do now?

• Obviously, mixed results, yet a strong model.
• Scale items were examined for bias in the wording
  some bias noticed.
• Data were factor analyzed

16
Factor Analysis

• PCA was conducted on the 10 scales
• Eigenvalues of 1 were specified
• Rotation method Promax with Kaiser Normalization

17
SCALES FACTORS
Dependent Variables
Union Loyalty Responsibility to Union Willingness to Work for Union Belief in Unionism 5 2 1 1
Independent Variables
Trust Trust in People Interpersonal Trust Participation Attitudes Work Locus of Control Participativeness Job Involvement Work Innovation Index 1 8 5 2 2 2
18
8 Factor Example(Interpersonal Trust)
19
1 Factor Example(Willingness to Work for the
Union)
20
Again, now what???

• Possibility of an underlying factor
• One of the Big Five or something from the 16
  PF? Extroversion, for instance?
• Second Order Factor Analysis identified as a
  possible investigatory tool.

21
Second Order FA Results
22
Results

• Goodness of Fit Statistics
• Degrees of Freedom 32
• Minimum Fit Function Chi-Square 53.355 (P
  0.0103)
• Root Mean Square Residual (RMR) 0.198
• Standardized RMR 0.0745
• Goodness of Fit Index (GFI) 0.930
• Adjusted Goodness of Fit Index (AGFI) 0.879

23
Structural Model with T-Values
24
Further Steps

• How do these results steer further development of
  scales?
• Should the scales be re-worded, or dumped and
  re-written from scratch?
• Is SEM just witchcraft anyway???