5
using AIC was used to retain only the most relevant predic-
tors, which were age and ethnicity. Age was measured in
about 15-year age brackets, and participants were on aver-
age 0.36 points on perceived safety for each higher bracket,
p 0.001, beta =0.247. White participants were 0.29
points higher in perceived safety than non-whites, though
this effect failed to reach statistical significance, p =0.10,
beta =0.115.
DISCUSSIONS
The results indicate several of factors evaluated in this study
affect participants’ safety and trust in a gas monitor alert.
First, the amount of information provided to participants
affected their perceived safety and trust with a positive trend
where increasing the information level increases safety and
trust score. System type, whether AI- or human-based sys-
tem did not have a significant effect.
Effect of System Type on Safety and Trust
Based on the analysis from the ANOVA test, we can con-
clude that the type of system sending the alert messages
did not have any significant effect on safety and trust. This
result is consistent with the results from [5] but counter to
the results from [5] that suggest that the channel of com-
munication affects human response [41]. The difference in
results could be due to the fact that each of these studies
was conducted in different settings thereby showing differ-
ent outcomes. For example, work done by [5], showed on
one hand that, when it comes to evacuation decisions made
during fire and explosion scenarios, channels of commu-
nication had no significant effect while on the other hand,
channel of communication had a significant impact on
miners’ evacuation decision in cave-in scenario. This sug-
gests that a person’s decision will not only be affected by
the channel of communication but also by the emergency
presented.
Effect of Information Level on Safety and Trust
The amount of information about the gas monitoring sys-
tem provided to participants had a significant effect on their
perceived safety and trust, with an increase in information
level causing an increase in safety and trust score. This effect
is consistent with the results from studies that looked at
the effect of details in alert messages on people’s responses.
Works by [5] and [37] support such a trend where increas-
ing the level of detail in alert messages tends to have a posi-
tive significant effect on responses. The results also tend to
reecho the warning given by [37] on the misconception
that during emergencies, shorter information is preferred
over longer, when most studies including this, have shown
that more information works better.
Effect of Demographics on Safety and Trust
Based on the analysis, we can conclude that demographic
factors such as age, number of children and ethnicity had
a significant effect on safety and trust with a positive trend
where safety and trust scores increase with age and number
of children. Literature shows the work of other research-
ers also indicate that demographics such as age, gender,
ethnicity have an influence on decision-making and safety.
Notably, the work done by [5, 42, 7, 44 and 45] demon-
strated the effects of participants’ demography on their
safety performance. The results show age, ethnicity, and
number of children are significant towards safety, and trust
measurement is consistent with [7, 42 and 5]. Though our
study did not demonstrate the U-shaped age-safety rela-
tionship identified by [43], it shows workers above age 50
have the highest levels of safety and trust, which is consis-
tent with the result from the study by [45]. Participants
aged below 40 tend to be complacent [46], resulting in
lower safety attitude and awareness [42, 47].
CONCLUSIONS AND FUTURE WORK
This study explored how system type, information level,
and individual demographics impact safety, trust, and
decision-making during underground mine gas emergen-
cies. The findings reveal that, while the type of system used
did not significantly affect perceptions of safety and trust in
this experiment, the amount of information provided did.
Additionally, demographic factors including age, number of
children, and ethnicity also play a role in shaping safety and
trust perceptions. These results highlight the importance of
offering detailed information in emergency situations and
suggest that human-centered engineering systems can bet-
ter support miner safety during gas-related incidents.
This study has some limitations, such as potential biases
from relying on self-reported data in scenario-based surveys,
which underscores the need for more objective measure-
ment methods. Additionally, scenario-based surveys may
not fully capture the realism of real-world responses, point-
ing to the value of near-reality or virtual reality simulations
to better simulate emergency conditions. The participant
group also did not include actual miners, which suggests
that future research should recruit real miners to enhance
both ecological validity and generalizability.
The study reinforced the importance of offering
detailed information during gas emergencies but left open
the question of exactly what information should be pro-
vided about the gases. To address this, future research could
using AIC was used to retain only the most relevant predic-
tors, which were age and ethnicity. Age was measured in
about 15-year age brackets, and participants were on aver-
age 0.36 points on perceived safety for each higher bracket,
p 0.001, beta =0.247. White participants were 0.29
points higher in perceived safety than non-whites, though
this effect failed to reach statistical significance, p =0.10,
beta =0.115.
DISCUSSIONS
The results indicate several of factors evaluated in this study
affect participants’ safety and trust in a gas monitor alert.
First, the amount of information provided to participants
affected their perceived safety and trust with a positive trend
where increasing the information level increases safety and
trust score. System type, whether AI- or human-based sys-
tem did not have a significant effect.
Effect of System Type on Safety and Trust
Based on the analysis from the ANOVA test, we can con-
clude that the type of system sending the alert messages
did not have any significant effect on safety and trust. This
result is consistent with the results from [5] but counter to
the results from [5] that suggest that the channel of com-
munication affects human response [41]. The difference in
results could be due to the fact that each of these studies
was conducted in different settings thereby showing differ-
ent outcomes. For example, work done by [5], showed on
one hand that, when it comes to evacuation decisions made
during fire and explosion scenarios, channels of commu-
nication had no significant effect while on the other hand,
channel of communication had a significant impact on
miners’ evacuation decision in cave-in scenario. This sug-
gests that a person’s decision will not only be affected by
the channel of communication but also by the emergency
presented.
Effect of Information Level on Safety and Trust
The amount of information about the gas monitoring sys-
tem provided to participants had a significant effect on their
perceived safety and trust, with an increase in information
level causing an increase in safety and trust score. This effect
is consistent with the results from studies that looked at
the effect of details in alert messages on people’s responses.
Works by [5] and [37] support such a trend where increas-
ing the level of detail in alert messages tends to have a posi-
tive significant effect on responses. The results also tend to
reecho the warning given by [37] on the misconception
that during emergencies, shorter information is preferred
over longer, when most studies including this, have shown
that more information works better.
Effect of Demographics on Safety and Trust
Based on the analysis, we can conclude that demographic
factors such as age, number of children and ethnicity had
a significant effect on safety and trust with a positive trend
where safety and trust scores increase with age and number
of children. Literature shows the work of other research-
ers also indicate that demographics such as age, gender,
ethnicity have an influence on decision-making and safety.
Notably, the work done by [5, 42, 7, 44 and 45] demon-
strated the effects of participants’ demography on their
safety performance. The results show age, ethnicity, and
number of children are significant towards safety, and trust
measurement is consistent with [7, 42 and 5]. Though our
study did not demonstrate the U-shaped age-safety rela-
tionship identified by [43], it shows workers above age 50
have the highest levels of safety and trust, which is consis-
tent with the result from the study by [45]. Participants
aged below 40 tend to be complacent [46], resulting in
lower safety attitude and awareness [42, 47].
CONCLUSIONS AND FUTURE WORK
This study explored how system type, information level,
and individual demographics impact safety, trust, and
decision-making during underground mine gas emergen-
cies. The findings reveal that, while the type of system used
did not significantly affect perceptions of safety and trust in
this experiment, the amount of information provided did.
Additionally, demographic factors including age, number of
children, and ethnicity also play a role in shaping safety and
trust perceptions. These results highlight the importance of
offering detailed information in emergency situations and
suggest that human-centered engineering systems can bet-
ter support miner safety during gas-related incidents.
This study has some limitations, such as potential biases
from relying on self-reported data in scenario-based surveys,
which underscores the need for more objective measure-
ment methods. Additionally, scenario-based surveys may
not fully capture the realism of real-world responses, point-
ing to the value of near-reality or virtual reality simulations
to better simulate emergency conditions. The participant
group also did not include actual miners, which suggests
that future research should recruit real miners to enhance
both ecological validity and generalizability.
The study reinforced the importance of offering
detailed information during gas emergencies but left open
the question of exactly what information should be pro-
vided about the gases. To address this, future research could