Safety climate– cross-validation, strength and prediction of safety behaviour.

22
Safety climate– cross- validation, strength and prediction of safety behaviour. Anders Pousette, Susanna Larsson and Marianne Törner National Institute for Working Life, Sweden anders . pousette @ arbetslivsinstitutet .se

description

Safety climate– cross-validation, strength and prediction of safety behaviour. Anders Pousette , Susanna Larsson and Marianne Törner National Institute for Working Life , Sweden [email protected]. Safety climate. Dimensionality of safety climate - PowerPoint PPT Presentation

Transcript of Safety climate– cross-validation, strength and prediction of safety behaviour.

Page 1: Safety climate– cross-validation, strength and prediction of safety behaviour.

Safety climate– cross-validation, strength and prediction of safety behaviour.

Anders Pousette, Susanna Larsson

and Marianne Törner

National Institute for Working Life, Sweden

[email protected]

Page 2: Safety climate– cross-validation, strength and prediction of safety behaviour.

2 A. Pousette

Safety climate

•Dimensionality of safety climate

•Sharedness as a property of safety climate

•Longitudinal prediction of safety behaviour

Page 3: Safety climate– cross-validation, strength and prediction of safety behaviour.

3 A. Pousette

Aims•Cross-validate a factor structure reported by Cheyne et al (1998), as well as to test the existence of a hypothesised second order safety climate factor;

•Explore whether degree of agreement (sharedness) differentiated between safety climate factors and measures of individual attitudes towards safety

•Test the predictive validity of safety climate (at time 1) on self-reported safety behaviour (at time 2).

Page 4: Safety climate– cross-validation, strength and prediction of safety behaviour.

4 A. Pousette

Safety management in large construction projects with a complex organisation National Institute for Working Life, Göteborg, SwedenFinancial support: National Road Administration, Vinnova, AFA-TFA, SBUF

Map of central parts of Gothenburg

The Götatunnel Göteborg, Sweden

Page 5: Safety climate– cross-validation, strength and prediction of safety behaviour.

5 A. Pousette

The GötatunnelOpening ceremony 16th June 2006

Page 6: Safety climate– cross-validation, strength and prediction of safety behaviour.

6 A. Pousette

The present study

•Three samples

•7 months interval

•N1 = 242, N2 = 275, N3 = 284

•Matched obs. N1-2 = 166, N2-3 = 174

•Four different companies and their subcontractors

•Response rate 95%, 87% and 84%

Page 7: Safety climate– cross-validation, strength and prediction of safety behaviour.

A. Pousette7

Instrument - overviewSafety climate organizational level

Safety behaviour

Individual attitudes to safety

”How do managers handle safety?”

”What is your opinion with regard to safety? ”

”How do you do with regard to safety?”

Safety climate group level

”How is it in your workgroup with regard to safety?”

Page 8: Safety climate– cross-validation, strength and prediction of safety behaviour.

8 A. Pousette

Safety climate•Management safety priority 4 items•Safety management 16 items•Safety communication 8 items

•Workgroup safety involvement 5 items

Organizational level

Group level

Page 9: Safety climate– cross-validation, strength and prediction of safety behaviour.

9 A. Pousette

Individual attitudes to safety•Safety motivation 7 items, alfa=0,78

•Safety knowledge 4 items, alfa=0,82

Page 10: Safety climate– cross-validation, strength and prediction of safety behaviour.

10 A. Pousette

Safety behaviour(self rated) Average score of three measures

• Structural safety behaviour 5 items, alfa=0,88

• Interactional safety behaviour items, alfa=0,79

• Personal safety behaviour 6 items, alfa=0,86

Page 11: Safety climate– cross-validation, strength and prediction of safety behaviour.

11 A. Pousette

Analysis

•Confirmatory factor analysis (AMOS 4)

•ICC (1,1), ICC (1,k) (Schrout & Fleiss , 1979)

•Hierarchical regression analysis

Page 12: Safety climate– cross-validation, strength and prediction of safety behaviour.

12 A. Pousette

Confirmatory factor analysis

Specified model

Theory

Measurement

Sample covariance

matrix

Model based covariance matrix

Measures of fit

Parameter estimates

Page 13: Safety climate– cross-validation, strength and prediction of safety behaviour.

A. Pousette13

chi2=1287,760df=489p=,000relativ chi2=2,633NFI=,948CFI=,967RMSEA=,082AIC=1497,760

SMgmt

Nv1_22

e1

Nv1_13

e2

Nv1_9

e3

v1_21

e4

,78

v1_20

e5

,73

v1_19

e6

,69

v1_18

e7

,81

v1_17

e8

,83

v1_14

e9

,70

v1_10

e10

v1_8

e11

v1_7

e12

v1_6

e13

v1_5

e14

SGroup

v1_39 e15

,70 v1_40 e16,71

v1_42 e17,66v1_44 e18,63

SComm

v1_15e23v1_25e24v1_27e25v1_28e26v1_29e27

Nv1_30e28Nv1_32e29v1_31e30

,71,73,65,86

,81,60,60,62

SPrio

Nv1_4

e31

Nv1_3

e32

,77

Nv1_2

e33

,83

Nv1_1

e34

Nv1_23

e35

Nv1_24

e36

,72

,85

,73

,66,76

,79

,72,67,55,65,59,68,71,64,68,67

,87 ,82

v1_43 e37

ALLAStandardized estimates

,52

Results aim 1: CFA of four safety climate factors

Page 14: Safety climate– cross-validation, strength and prediction of safety behaviour.

A. Pousette14

Results aim 1: Second order CFA of safety climate dimensions

SPrio SMgmt SComm WSI

1 1 1 1 1

SCg

.77 .96 .92 .79

2 = 2113

df= 491

NFI = .98

CFI = .98

RMSEA = .064

Page 15: Safety climate– cross-validation, strength and prediction of safety behaviour.

A. Pousette15

Results, aim 2: Sharedness. Agreement among raters

Sample 2 Sample 3 Sample 2 Sample 3

Type ICC(1,1) ICC(1,1) ICC(1,k) ICC(1,k)

Management safety priority

C .24 .22 .87 .86

Safety management

C .26 .27 .88 .89

Safety communication

C .28 .30 .89 .90

Workgroup safety involvement

C .12 .13 .74 .77

Safety motivation

A .03 .09 .40 .67

Safety knowledge

A .02 .07 .35 .62

Note: C: Climate A: Attitude

Page 16: Safety climate– cross-validation, strength and prediction of safety behaviour.

A. Pousette16

Results, aim 3: Predicting Safety behaviour at T2 (T1+7months)

Model 1 Model 2

Predictor at T1

B SE B Beta B SE B Beta

Safety behaviour

.74 .05 .74*** .64 .07 .64***

Safety climate

.11 .04 .17*

R2 .55 .57

F for change in R2

196.1*** 6.7*

* p<.05 ** p<.01 *** p.001 N=166

Page 17: Safety climate– cross-validation, strength and prediction of safety behaviour.

A. Pousette17

Results, aim 3: Predicting Safety behaviour at T3 (T2+7months)

Model 1 Model 2

Predictor at T2

B SE B Beta B SE B Beta

Safety behaviour

.79 .05 .76*** .71 .06 .68***

Safety climate

.10 .04 .14*

R2 .58 .59

F for change in R2

222.8*** 5.8*

* p<.05 ** p<.01 *** p.001 N=174

Page 18: Safety climate– cross-validation, strength and prediction of safety behaviour.

18 A. Pousette

Conclusions (aim 1)•Safety climate scales by by Cheyne et al. (1998) was successfully replicated in three samples, in a new context

•Support the construct validity and generalizability of the scales

•Factor loadings were invariant. The scales proved suitable for measuring safety climate in longitudinal studies.

•A higher order safety climate was identified. An overall safety climate measure is meaningful

Page 19: Safety climate– cross-validation, strength and prediction of safety behaviour.

19 A. Pousette

Conclusions (aim 2)•Measures of sharedness (ICC(1,1) and ICC(1,k)) was higher for safety climate factors than for individual safety attitude factors

•Support the theoretically suggested distinction between safety climate (as a property of a social unit) and individual safety attitudes (as a property of the individual person)

•Implications for scale construction: calls for strictness concerning the object of evaluation. Do not mix climate items and attitudinal items within the same scale!

Page 20: Safety climate– cross-validation, strength and prediction of safety behaviour.

A. Pousette20

Foremen

Workgroup

Management

Company /site

R12

3

1 I observe myself

2 I observe my group

3 I observe my company (including the management)

R=Respondent, always the observer

Social unit

Climate= aggregated responses of several R evaluating (2) group climate or (3) organizational climate

Object of evaluation

Page 21: Safety climate– cross-validation, strength and prediction of safety behaviour.

21 A. Pousette

Conclusions (aim3)

• Safety climate predicted change in safety behaviour (self rated) in longitudinal data, 7 months later

•Support the hypothesis of a causal relationship between safety climate and safety behaviour

•Implication for safety interventions: successful intervention in the safety climate area is likely to affect safety behaviour

Page 22: Safety climate– cross-validation, strength and prediction of safety behaviour.

Thank you for your attention!

National Institute for Working Life, Sweden