Desktop version

Home arrow Psychology arrow The Wiley Blackwell handbook of the psychology of recruitment, selection and employee retention

Primary validity studies in Africa, Latin America and the Indian-Pacific Ocean countries

A number of primary validity studies are worth mentioning as they have been carried out in less known countries, such as South Africa, Abu Dhabi, New Zealand, Australia, Singapore, Brazil, Chile, Mexico, Peru and Argentina.

With regard to African countries, an increasing number of studies have been undertaken in South Africa in the last 15 years. Muller and Schepers (2003) carried out a study on the validity of GMA for predicting success in a training course for the South African National Defence Force. They found that a cognitive composite of four tests (Raven’s matrices, conceptualization, reading comprehension and listening comprehension) predicted training proficiency (r= 0.584, N = 96). Kriek and Dowdeswell (2009) reported two concurrent validity studies, with validity coefficients separated for Black and White participants. In the first study, they found an average coefficient of 0.39 (N=66) for Black participants and 0.27 (N=34) for White participants. In the second study, Kriek and Dowdeswell found a validity coefficient of 0.48 (N=47) for Black participants and 0.36 (N=57) for White participants. In another South African study, Strachan (2008) examined the predicted validity of the APIL-SV (a cognitive battery assessing fluid intelligence for predicting training success) in a sample of auditing employees. Strechan found an average uncorrected validity of

Table 7.2 Moderator effects of job complexity on validity size.

Ability

K

N

P

90CV

Training

USA studies’1

Low complexity

8

575

0.54

0.49

Medium complexity

54

3,823

0.57

0.36

High complexity

4

235

0.65

0.65

EU studiesb

Low complexity

29

8,152

0.73

0.60

Medium complexity

66

22,100

0.74

0.60

High complexity

4

596

0.75

0.67

Job Performance

USA studiesa

Low complexity

201

14,403

0.40

0.36

Medium complexity

151

12,933

0.51

0.31

High complexity

17

1,114

0.56

0.52

EU studiesb

Low complexity

12

864

0.51

0.38

Medium complexity

43

4,744

0.53

0.21

High complexity

14

1,604

0.64

0.33

K = number of studies; N=sample size; p = operational validity; 90CV = 90% credibility value.

Source: a = Hunter (1986, p. 344) and Hunter and Hunter (1984, p. 81); b = Salgado et al. (2003); c = Oh, Schmidt, Shaffer & Le (2008; average of two meta-analyses); Oh, Schmidt, Shaffer & Le (2008; average of three meta-analyses).

0.38 (N=69). Fertig (2009) found that cognitive ability predicted managerial performance in a multi-ethnic sample of brand managers (average r= 0.16, N= 124). De Kock and Schlechter (2009) found that a battery of fluid and spatial ability tests predicted flight training (r = 0.35; N= 108). Nicholls, Viviers and Visser (2009) found that cognitive ability predicted supervisory performance ratings, call handling time and call quality in a sample of call centre operators (0.24, 0.28 and 0.33, respectively; N = 140). In another study, Pelser, Berg and Visser (2005) found that GMA predicted supervisor job performance ratings in a sample of truck operators (r = 0.24; N=104). Dale (2010) found that a test of mental alertness predicted work performance among leaners in the clothing industry (r = 0.33, N=200). The correlations were meta-analysed for this chapter and the results show that, on average, cognitive abilities have an observed validity of 0.30 (K = 9, N = 841), and 0.44 (K= 3, N = 273) for predicting job performance and training, respectively. These validities corrected for indirect range restriction (using the u values found by Salgado & Anderson, 2003a), and criterion unreliability (using 0.52 for job performance and 0.56 for training) resulted in operational validities of 0.66 (90CV = 0.66) and 0.81(90CV = 0.76) for job performance and training, respectively.

In the Indian subcontinent and the South Pacific islands (i.e., Australia, Singapore and New Zealand) a few small-sample studies have been undertaken in the last 15 years. In New Zealand, Mann’s (2011) small-sample study found that a battery of three cognitive tests predicted overall job performance (average r = 0.20, N=43). Mann (2011) also found that general cognitive ability predicted task performance efficiently (r = 0.30), but predicted neither contextual performance (r = 0.03) nor team performance (r = -0.12). Black (2000) found that GMA predicted training performance in a sample of New Zealand police recruits (r = 0.33, N=284). In Australia, Green and Macqueen (2008) reported a small- sample study (N=37) in which they found that cognitive ability predicted job performance in managers and supervisors (r = 0.37). In Singapore, Chan and Schmitt (2002) found that cognitive ability predicted task performance (r = 0.25), but not OCB (r = 0.03) or overall job performance (r=-0.02) in a sample of civil servants (N=160), and Roberts, Harms, Caspi and Moffit (2005) found that IQ predicted occupational level (r = 0.44, N=838) and showed a very small correlation with counterproductive work behaviour (r = 0.08, N=838). In Abu Dhabi, Al-Ali, Gamer and Magadley (2009) found that GMA predicted both objective CWB (r = -0.14) and self-rated CWB (r = -0.20) in a sample of police officers (N=310). These studies were also meta-analysed for this chapter and the results showed that cognitive abilities had an average validity of 0.11 (K= 3, N = 240) for predicting job performance. This validity corrected for indirect range restriction (using the u values found by Salgado and Anderson, 2003a), and criterion unreliability (using 0.52 for job performance) resulted in an operational validity of 0.27 (90CV = 0.07).

A few studies have been carried out in the last 20 years in Latin American countries. In Brazil, Baumgartl and Primi (2006) found that a cognitive ability test was a predictor of job accidents in an electrical power company (r=-0.39), Thadeu and Ferreira (2013) found that cognitive ability predicted training in a large sample of police officers (r =0.14, N= 1,177), and Cobero, Primi and Muniz (2006) found that GMA predicted supervisory job performance in a sample of heterogeneous workers (r = 0.39, N=119). In Argentina, Castro-Solano and Casullo (2005) found that Raven’s test predicted training in an Army sample (r = 0.23, N= 137). In Peru, Rosales-Lopez (2012) found that a mechanical aptitude test predicted performance in skilled workers (r= 0.78; N=88). In Mexico, Moreno-Garcia (2011) found that GMA predicted job performance ratings in managers of a financial company (r = 0.22, N=87). In Chile, Cuadra-Peralta (1990) found that cognitive ability, as measured by the GATB, predicted performance in a sample of 80 workers of a copper mining company (r = 0.33), and Barros, Kausel, Cuadra and Diaz (2014) found observed validities of 0.16 (N=253), -0.04 (N=156) and 0.35 (N=103) in three independent studies. I carried out a small-scale meta-analysis with these studies. The present meta-analysis shows that, on average, cognitive ability predicted job performance with an observed validity of 0.26 (K = 7, N = 886) and training with a validity of 0.15 (K= 2, N = 1,314). These validities, corrected for indirect range restriction using the u values found by Salgado and colleagues (2003a) and criterion unreliability using 0.52 for job performance and 0.56 for training, resulted in operational validities of 0.53 (90CV = 0.09), and 0.36 (90CV = 0.36) for job performance and training, respectively.

In summary, primary validity studies in countries of Africa, the Asia-Pacific region and Latin America in the last 12 years show that GMA predicted both job performance and training criteria. Small-scale meta-analyses carried out for this chapter with these validity studies show operational validities similar to those found in the US and Europe, confirming that GMA consistently predicts performance across cultures. The results of these small-scale meta-analyses are shown in Table 7.3.

 
Source
< Prev   CONTENTS   Source   Next >

Related topics