Mathematical and Statistical Methods for Actuarial Sciences and Finance

80 P. Cerchiello, M. Iannario, and D. Piccolo

0.0 0.1 0.2 0.3 0.4 0.5 0.6

0.0

0.2

0.4

0.6

0.8

1.0

DANGER Degree

pai estimates

csi estimates

SC

SH

MC

EW

CO

SL FE

ST

CU

HI

0.0 0.1 0.2 0.3 0.4 0.5 0.6

0.0

0.2

0.4

0.6

0.8

1.0

FREQUENCY

pai estimates

csi estimates

SC

SH

MCEW

CO

FE

SL

ST

CU

HI

0.0 0.1 0.2 0.3 0.4 0.5 0.6

0.0

0.2

0.4

0.6

0.8

1.0

EXPOSURE

pai estimates

csi estimates SC

SH

MC

EW

CO

FE

SL

ST

CU

HI

0.0 0.1 0.2 0.3 0.4 0.5 0.6

0.0

0.2

0.4

0.6

0.8

1.0

CONTROL

pai estimates

csi estimates

SC

SH

MC

EW

CO FE

SL STCU

HI

0.0 0.1 0.2 0.3 0.4 0.5 0.6

0.0

0.2

0.4

0.6

0.8

1.0

TRAINING

pai estimates

csi estimates

SC

SH MC

EW

FECO

SL

ST

CU

HI

Fig. 1.Assessing risk perception: a map of items. 1=Structural Collapse (SC), 2=Short
Circuit (SH), 3=Moving Machinery Clash (MC), 4=Eye Wound (EW), 5=Collision (CO),
6=Fire/Explosion (FE), 7=Slipping (SL), 8=Strain (ST), 9=Cut (CU), 10=Hit (HI)

3.3 Perception of fire/explosion risk

In this context we analyse thedegree of danger, a principal item in measuring risk
perception and we focus on the responses of samples with respect tofire/explosion
risk. More specifically, we consider the degree of danger that people perceive with
respect tofire riskand we connect it with some covariates.
In this kind of analysis, sensible covariates have to be introduced in the model
by means of a stepwise strategy where a significant increase in the log-likelihoods
(difference of deviances) is the criterion to compare different models. In order to
simplify the discussion, we present only the full model and check it with respect to a
model without covariates.