Impact of Construction Workers' Hazard Identification Skills on Their Visual Attention

Sogand Hasanzadeh, Behzad Esmaeli, Michael D Dodd

Research output: Contribution to journalArticle

12 Citations (Scopus)

Abstract

Eye-movement metrics have been shown to correlate with attention and, therefore, represent a means of identifying and analyzing an individual's cognitive processes. Human errors - such as failure to identify a hazard - are often attributed to a worker's lack of attention. Piecemeal attempts have been made to investigate the potential of harnessing eye movements as predictors of human error (e.g., failure to identify a hazard) in the construction industry, although more attempts have investigated human error via subjective measurements. To address this knowledge gap, the present study harnessed eye-tracking technology to evaluate the impacts of workers' hazard-identification skills on their attentional distributions and visual search strategies. To achieve this objective, an experiment was designed in which the eye movements of 31 construction workers were tracked while they searched for hazards in 35 randomly ordered construction scenario images. Workers were then divided into three groups on the basis of their hazard identification performance. Three fixation-related metrics - fixation count, dwell-time percentage, and run count - were analyzed during the eye-tracking experiment for each group (low, medium, and high hazard-identification skills) across various types of hazards. Then, multivariate ANOVA (MANOVA) was used to evaluate the impact of workers' hazard-identification skills on their visual attention. To further investigate the effect of hazard identification skills on the dependent variables (eye movement metrics), two distinct processes followed: separate ANOVAs on each of the dependent variables, and a discriminant function analysis. The analyses indicated that hazard identification skills significantly impact workers' visual search strategies: workers with higher hazard-identification skills had lower dwell-time percentages on ladder-related hazards; higher fixation counts on fall-to-lower-level hazards; and higher fixation counts and run counts on fall-protection systems, struck-by, housekeeping, and all hazardous areas combined. Among the eye-movement metrics studied, fixation count had the largest standardized coefficient in all canonical discriminant functions, which implies that this eye-movement metric uniquely discriminates workers with high hazard-identification skills and at-risk workers. Because discriminant function analysis is similar to regression, discriminant function (linear combinations of eye-movement metrics) can be used to predict workers' hazard-identification capabilities. In conclusion, this study provides a proof of concept that certain eye-movement metrics are predictive indicators of human error due to attentional failure. These outcomes stemmed from a laboratory setting, and, foreseeably, safety managers in the future will be able to use these findings to identify at-risk construction workers, pinpoint required safety training, measure training effectiveness, and eventually improve future personal protective equipment to measure construction workers' situation awareness in real time.

Original languageEnglish (US)
Article number04017070
JournalJournal of Construction Engineering and Management
Volume143
Issue number10
DOIs
StatePublished - Jan 1 2017

Fingerprint

Hazards
Eye movements
Hazard
Workers
Visual attention
Analysis of variance (ANOVA)
Ladders
Construction industry
Managers
Experiments

Keywords

  • Construction safety
  • Eye-tracking technology
  • Hazard identification skill
  • Human error
  • Labor and personnel issues
  • Multivariate ANOVA (MANOVA)

ASJC Scopus subject areas

  • Civil and Structural Engineering
  • Building and Construction
  • Industrial relations
  • Strategy and Management

Cite this

Impact of Construction Workers' Hazard Identification Skills on Their Visual Attention. / Hasanzadeh, Sogand; Esmaeli, Behzad; Dodd, Michael D.

In: Journal of Construction Engineering and Management, Vol. 143, No. 10, 04017070, 01.01.2017.

Research output: Contribution to journalArticle

@article{0602849724314e449522b01011eacb7a,
title = "Impact of Construction Workers' Hazard Identification Skills on Their Visual Attention",
abstract = "Eye-movement metrics have been shown to correlate with attention and, therefore, represent a means of identifying and analyzing an individual's cognitive processes. Human errors - such as failure to identify a hazard - are often attributed to a worker's lack of attention. Piecemeal attempts have been made to investigate the potential of harnessing eye movements as predictors of human error (e.g., failure to identify a hazard) in the construction industry, although more attempts have investigated human error via subjective measurements. To address this knowledge gap, the present study harnessed eye-tracking technology to evaluate the impacts of workers' hazard-identification skills on their attentional distributions and visual search strategies. To achieve this objective, an experiment was designed in which the eye movements of 31 construction workers were tracked while they searched for hazards in 35 randomly ordered construction scenario images. Workers were then divided into three groups on the basis of their hazard identification performance. Three fixation-related metrics - fixation count, dwell-time percentage, and run count - were analyzed during the eye-tracking experiment for each group (low, medium, and high hazard-identification skills) across various types of hazards. Then, multivariate ANOVA (MANOVA) was used to evaluate the impact of workers' hazard-identification skills on their visual attention. To further investigate the effect of hazard identification skills on the dependent variables (eye movement metrics), two distinct processes followed: separate ANOVAs on each of the dependent variables, and a discriminant function analysis. The analyses indicated that hazard identification skills significantly impact workers' visual search strategies: workers with higher hazard-identification skills had lower dwell-time percentages on ladder-related hazards; higher fixation counts on fall-to-lower-level hazards; and higher fixation counts and run counts on fall-protection systems, struck-by, housekeeping, and all hazardous areas combined. Among the eye-movement metrics studied, fixation count had the largest standardized coefficient in all canonical discriminant functions, which implies that this eye-movement metric uniquely discriminates workers with high hazard-identification skills and at-risk workers. Because discriminant function analysis is similar to regression, discriminant function (linear combinations of eye-movement metrics) can be used to predict workers' hazard-identification capabilities. In conclusion, this study provides a proof of concept that certain eye-movement metrics are predictive indicators of human error due to attentional failure. These outcomes stemmed from a laboratory setting, and, foreseeably, safety managers in the future will be able to use these findings to identify at-risk construction workers, pinpoint required safety training, measure training effectiveness, and eventually improve future personal protective equipment to measure construction workers' situation awareness in real time.",
keywords = "Construction safety, Eye-tracking technology, Hazard identification skill, Human error, Labor and personnel issues, Multivariate ANOVA (MANOVA)",
author = "Sogand Hasanzadeh and Behzad Esmaeli and Dodd, {Michael D}",
year = "2017",
month = "1",
day = "1",
doi = "10.1061/(ASCE)CO.1943-7862.0001373",
language = "English (US)",
volume = "143",
journal = "Journal of Construction Engineering and Management - ASCE",
issn = "0733-9364",
publisher = "American Society of Civil Engineers (ASCE)",
number = "10",

}

TY - JOUR

T1 - Impact of Construction Workers' Hazard Identification Skills on Their Visual Attention

AU - Hasanzadeh, Sogand

AU - Esmaeli, Behzad

AU - Dodd, Michael D

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Eye-movement metrics have been shown to correlate with attention and, therefore, represent a means of identifying and analyzing an individual's cognitive processes. Human errors - such as failure to identify a hazard - are often attributed to a worker's lack of attention. Piecemeal attempts have been made to investigate the potential of harnessing eye movements as predictors of human error (e.g., failure to identify a hazard) in the construction industry, although more attempts have investigated human error via subjective measurements. To address this knowledge gap, the present study harnessed eye-tracking technology to evaluate the impacts of workers' hazard-identification skills on their attentional distributions and visual search strategies. To achieve this objective, an experiment was designed in which the eye movements of 31 construction workers were tracked while they searched for hazards in 35 randomly ordered construction scenario images. Workers were then divided into three groups on the basis of their hazard identification performance. Three fixation-related metrics - fixation count, dwell-time percentage, and run count - were analyzed during the eye-tracking experiment for each group (low, medium, and high hazard-identification skills) across various types of hazards. Then, multivariate ANOVA (MANOVA) was used to evaluate the impact of workers' hazard-identification skills on their visual attention. To further investigate the effect of hazard identification skills on the dependent variables (eye movement metrics), two distinct processes followed: separate ANOVAs on each of the dependent variables, and a discriminant function analysis. The analyses indicated that hazard identification skills significantly impact workers' visual search strategies: workers with higher hazard-identification skills had lower dwell-time percentages on ladder-related hazards; higher fixation counts on fall-to-lower-level hazards; and higher fixation counts and run counts on fall-protection systems, struck-by, housekeeping, and all hazardous areas combined. Among the eye-movement metrics studied, fixation count had the largest standardized coefficient in all canonical discriminant functions, which implies that this eye-movement metric uniquely discriminates workers with high hazard-identification skills and at-risk workers. Because discriminant function analysis is similar to regression, discriminant function (linear combinations of eye-movement metrics) can be used to predict workers' hazard-identification capabilities. In conclusion, this study provides a proof of concept that certain eye-movement metrics are predictive indicators of human error due to attentional failure. These outcomes stemmed from a laboratory setting, and, foreseeably, safety managers in the future will be able to use these findings to identify at-risk construction workers, pinpoint required safety training, measure training effectiveness, and eventually improve future personal protective equipment to measure construction workers' situation awareness in real time.

AB - Eye-movement metrics have been shown to correlate with attention and, therefore, represent a means of identifying and analyzing an individual's cognitive processes. Human errors - such as failure to identify a hazard - are often attributed to a worker's lack of attention. Piecemeal attempts have been made to investigate the potential of harnessing eye movements as predictors of human error (e.g., failure to identify a hazard) in the construction industry, although more attempts have investigated human error via subjective measurements. To address this knowledge gap, the present study harnessed eye-tracking technology to evaluate the impacts of workers' hazard-identification skills on their attentional distributions and visual search strategies. To achieve this objective, an experiment was designed in which the eye movements of 31 construction workers were tracked while they searched for hazards in 35 randomly ordered construction scenario images. Workers were then divided into three groups on the basis of their hazard identification performance. Three fixation-related metrics - fixation count, dwell-time percentage, and run count - were analyzed during the eye-tracking experiment for each group (low, medium, and high hazard-identification skills) across various types of hazards. Then, multivariate ANOVA (MANOVA) was used to evaluate the impact of workers' hazard-identification skills on their visual attention. To further investigate the effect of hazard identification skills on the dependent variables (eye movement metrics), two distinct processes followed: separate ANOVAs on each of the dependent variables, and a discriminant function analysis. The analyses indicated that hazard identification skills significantly impact workers' visual search strategies: workers with higher hazard-identification skills had lower dwell-time percentages on ladder-related hazards; higher fixation counts on fall-to-lower-level hazards; and higher fixation counts and run counts on fall-protection systems, struck-by, housekeeping, and all hazardous areas combined. Among the eye-movement metrics studied, fixation count had the largest standardized coefficient in all canonical discriminant functions, which implies that this eye-movement metric uniquely discriminates workers with high hazard-identification skills and at-risk workers. Because discriminant function analysis is similar to regression, discriminant function (linear combinations of eye-movement metrics) can be used to predict workers' hazard-identification capabilities. In conclusion, this study provides a proof of concept that certain eye-movement metrics are predictive indicators of human error due to attentional failure. These outcomes stemmed from a laboratory setting, and, foreseeably, safety managers in the future will be able to use these findings to identify at-risk construction workers, pinpoint required safety training, measure training effectiveness, and eventually improve future personal protective equipment to measure construction workers' situation awareness in real time.

KW - Construction safety

KW - Eye-tracking technology

KW - Hazard identification skill

KW - Human error

KW - Labor and personnel issues

KW - Multivariate ANOVA (MANOVA)

UR - http://www.scopus.com/inward/record.url?scp=85025673585&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85025673585&partnerID=8YFLogxK

U2 - 10.1061/(ASCE)CO.1943-7862.0001373

DO - 10.1061/(ASCE)CO.1943-7862.0001373

M3 - Article

VL - 143

JO - Journal of Construction Engineering and Management - ASCE

JF - Journal of Construction Engineering and Management - ASCE

SN - 0733-9364

IS - 10

M1 - 04017070

ER -