The Generalizability of Sexual Harassment Models across Heterogeneous Organizations

1
The Generalizability of Sexual Harassment Models
across Heterogeneous Organizations

• Patrick Wadlington, Fritz Drasgow, Louise
  Fitzgerald
• University of Illinois at Urbana-Champaign
• Funded by the Department of Defense and the
  National Institute of Mental Health

2
Introduction

• The job is one of the most well-documented arenas
  for the incidence and severity of sexual
  harassment (Fitzgerald, Collinsworth, Harned,
  2002).
• Strong support for the Fitzgerald et al. (1995)
  model has been obtained within a variety of
  individual organizations (Fitzgerald, Drasgow,
  Hulin, Gelfand, Magley, 1997 Glomb, Munson,
  Hulin, Bergman, Drasgow, 1999).

3
Introduction (continued)

• Purpose of present study To determine whether
  the model of sexual harassment can be generalized
  across diverse organizations.
• This is a much more stringent test of the model.

4
Samples (women only)

• Agribusiness
• 264 individuals
• Utility
• 365 individuals
• University
• 273 individuals
• Non-US
• 355 individuals

5
Model with 3 samples
6
Model with 3 samples - Measures

• Antecedents of SH
• Job gender context (3 items alpha.42)
• Gender ratio of their workgroup
• Gender of immediate supervisor
• Whether they were one of the first of their sex
  to do their job
• Organizational tolerance
• Organizational Tolerance for Sexual Harassment
  Inventory (Hulin, Fitzgerald, Drasgow, 1996)
• 18 items
• alpha.96

7
Model with 3 samples Measures (continued)

• Sexual Harassment
• Sexual Experiences Questionnaire-Revised
• Three behavioral types
• Gender Harassment put down
• Unwanted Sexual Attention come on
• Sexual Coercion bribery for sexual
  cooperation
• 17 items
• Alpha.81

8
Model with 3 samples Measures (continued)

• Consequences of SH
• Psychological condition
• Mental Health Index (Veit Ware, 1983)
• Evaluates psychological distress
• 9 items
• Alpha.89
• An abbreviated Crime-Related Post-Traumatic
  Stress Disorder Scale (Saunders, Arata, and
  Kilpatrick, 1990)
• A second measure of psychological distress
• 10 items
• Alpha.80

9
Model with 3 samples Measures (continued)

• Satisfaction With Life Scale (Diener 1984)
• Evaluates subjective well-being
• 5 items
• Alpha.87
• Job satisfaction
• Job Descriptive Index (Smith et al., 1969)
• 30 items
• Alpha.90

10
Model with 3 samples Measures (continued)

• Work withdrawal
• Work Withdrawal scale
• 9 items
• Alpha.70
• Job withdrawal
• Job withdrawal scale
• 5 items
• Alpha.63

11
Model with 3 samples Measures (continued)

• Health condition
• Health conditions index (Brodman, Erdman, Lorge,
  Wolff, 1949)
• 5 items
• Alpha.71
• Health satisfaction
• Subscale of the retirement descriptive index
  (Smith et al., 1969)
• 9 items
• Alpha.69

12
Model with 3 samples Measures (continued)

• Control Variables
• Job stress
• Stress in General Scale (SIG Smith, Sademan,
  McCrary, 1992)
• 9 items
• Alpha.87

13
Model with 4 samples
14
Description of Analyses

• Measurement equivalence
• Do the measures assess the underlying constructs
  in the same way across organizations?
• Invariance across the samples would provide
  support for the generalizability of the
  measurement model.
• Relational invariance
• Evaluate the relationships among constructs
  across groups.
• If found invariant, support for the
  generalizability of the relations among the
  constructs.

15
Description of Analyses (continued)

• Method Simultaneous factor analysis in several
  populations (SIFASP) Jöreskog, 1971
• Examine model-data fit for all samples separately
• Examine model-data fit for all samples
  simultaneously
• Equality constraints on the factor loadings
• If model-data fit does not significantly
  decrease, then the model is generalizable.

16
Results Model with 3 samples
17
Results Model with 3 samples (continued)
18
Results Model with 3 samples (continued)

• The model-data fit was not significantly
  deceased, so the model is generalizable.
• Therefore, now the relationships between the
  constructs within the model can be interpreted.

19
Interpretation Model with 3 samples
20
Results Model with 4 samples
21
Results Model with 4 samples (continued)
22
Results Model with 4 samples (continued)

• The model-data fit was not significantly
  deceased, so the model is generalizable.
• Therefore, now the relationships between the
  constructs within the model can be interpreted.

23
Interpretation Model with 4 samples
24
Implications

• The rigorous analyses provide very strong
  evidence that these related models are useful for
  studying, understanding, and reducing sexual
  harassment within a variety of organizations.
• The direct significant path from organization
  tolerance to job satisfaction implies that
  regardless of the level of sexual harassment that
  occurs within an organization, the CEO or manager
  would be prudent to take this issue seriously.
• Employers and employees alike not only need to
  deter and stop the actual behaviors of sexual
  harassment but also invest their resources in
  creating an environment that is not tolerant of
  sexual harassment.