Presentation Title goes here by dffhrtcv3


									An Investigation of Race and Sex Similarity Effects in
 Assessment Centers in the South African context.

                      Christine de Villiers
              Masters Supervisor: Francois de Kock
                            15 March 2011

              Department of Industrial Psychology
                    University of Stellenbosch

                    Department of Industrial Psychology
                Faculty of Economic and Management Sciences

• The assessor plays a critical role in the evaluation process.
  However the evaluations made by the assessors are
  subjective and therefore susceptible to intentional and
  unintentional biases (Mount, Sytsma, Hazucha, & Holt,

• There are multiple environmental and cognitive variables
  that influences the accuracy of the inferences made by
  assessors and heighten the probability of similarity
  judgements in assessment centres (Lowry, 1993; Sagie &
  Magnezy, 1997; Lievens 2002; Sacco, Scheu, Ryan & Schmitt,

• Social identity theory: By striving to maintain a consistent
  identity individuals tend to evaluate others with similar
  characteristics more favourable than individuals with
  dissimilar characteristics (Sacco, Scheu, Rayn, & Schmitt,
  2003; Goldberg, 2005).

• Both sex and race similarities have been found to influence
  work and employee related judgments in employee
  evaluation (Schmidt, 1976; Oppler, White, & Borman,1989;
  Lin, Dobbins and Fahr, 1992; Graves and Powell, 1995;
  Goldberg, 2005; Purkiss, Perrewe, Gillespie, Mayes and
  Ferris, 2006 Dean, Bobko, & Roth, 2008)
                   Similarity in evaluations

•   Gender similarities effects have been found by various reserchers; Oppler,
    White and Borman (1989) found significant interactions as well as main
    effects for gender similarities in the performance appraisal context. Graves
    and Powell (1995) found that interviewers found members of the opposite
    sex more similar to themselves, although it affected only marginally higher
    rating. Goldberg (2005) also found that there is a sex-dissimilarity effect
    amongst male raters that indicated their preferences female applicants.
    Dean, Bobko and Roth (2008) in their Meta-analysis found that on average
    females get higher ratings than males in assessment centres.

•   There are also several studies that have found race similarity effects in the
    evaluation context. Schmidt (1976) found that racial and attitudinal
    similarity was related to higher ratings. Pulakos, Oppler, White and
    Borman (1989) found significant interactions as well as main effects for race
    similarities in the performance appraisal. In 1997 Mount et al, in their study
    on performance ratings, found that Blacks gave more favourable ratings to
    all employees of their own race. Goldberg (2005) found a significant race-
    similarity effect for white raters.
               Similarity in evaluations

• Therefore, it could be expected that Rater schema and
  similarity bias can influence the ratings in assessment
  centres. Similarity bias have been confirmed in other
  contexts, e.g. interviews (Graves et. Al., 1995) and
  performance appraisal ratings (Oppler et. al.,1989 ).

• However Little is known about possible demographic
  similarity effects in AC ratings.

                   • Research Question

   Social identity theory suggests that demographic similarity in
   rater-ratee dyads could bias assessor ratings. Does assessor-
   assessee similarity influence AC dimension scores, so that
   assessors assign higher dimension ratings to individuals that are
   demographically similar to themselves?

                          • Objectives
• To determine if rater and ratee demography act as main effects
  on AC PEDR;
• To determine if rater-ratee demographic similarity acts as
  interaction effects in AC PEDR.

•    Main effects for Rater (Assessor) and Ratees (Candidates) will be investigated prior
     to determining whether rater-ratee interaction effects exist. However, no hypotheses
     will be formulated a priori about main effects due to the fact that the South African
     context differs markedly from the US context, bringing into question the
     generalisability of these findings.

•    Interaction effects of Rater (Assessor) and Ratee (Assessee) Race

H3: Raters will rate same-race ratees higher than other ratees(Same-race positive
Or: same race bias will not occur/same race negative bias will occur.

•    Interaction effects of Rater (Assessor) and Ratee (Assessee) Sex

H1: Raters will rate same-sex ratees higher than they rate other ratees (Same-sex
     positive bias).
     Or: same sex bias will not occur/same sex negative bias will occur.

• Client identity and all client data will be kept anonymous.

• Hope to partner with an existing AC where the
  demographics of both the assessors and assessees are

• Would be ideal to have a sample of +-100 candidates
  evaluated by at least five raters.

• Will use HLM analysis like in Sacco, Scheu, Rayn, & Schmitt
  (2003) to test for interaction and main effect.

• To ensure that selection and evaluation is unbiased and does
  not discriminate against any individual based on demographic
  characteristics one must identify and remove or minimise
  any factors that contribute to error variance in selection

• If the Similarity dyads have a significant impact on dimension
  ratings then one can look at ways to improve the process to
  so increase the validity of the assessment centre process

To top