6
significant relationship was found, these covariates were
excluded from further analyses. Secondly, a Chi-square test
of independence was performed to determine if the par-
ticipants were generally interested in VR (i.e., yes, maybe,
no). General VR interest was determined by the highest
level of interest in VR from both implementation strategy
questions. For example, if a participants responded “no” to
“instead of,” but “yes” to “in addition to” it would be a “yes”
for general VR interest. Lastly, a marginal homogeneity
test was used to assess whether participants preferred one
implementation strategy over the other. This within-sub-
jects comparison examined the change in responses from
“instead of” to “in addition to.” Missing data were excluded
on a case-by-case basis (N =107). Significance was set at α
≤ 0.05 for all tests. Because the qualitative survey responses
were brief and sometimes incomplete, no formal analysis
was conducted. Direct quotes are included as anecdotal
support.
Interviews
Two researchers analyzed the interview notes from all 20
interviews using a combination of deductive and inductive
strategies. For this study, researchers deductively applied
the five domains and 45 constructs (Table 1) of the CFIR
2.0 framework [40] to the data. The researchers used the
framework guidance from https://cfirguide.org [47] and
the planning questions from a previous study [39] as a basis
for the codebook. From there, the researchers used a con-
stant comparison method [48] to identify “themes” made
up of actionable barriers and facilitators for each construct.
To do so, the two researchers first coded two of the inter-
views together—one mine rescue trainer and one organiza-
tional representative. The researchers then divided up the
remaining interviews and coded them independently. The
researchers met periodically while coding to refine the con-
struct and theme definitions. One hundred percent agree-
ment was achieved through discussion.
Other Pre-Implementation Activities
An important but sometimes overlooked aspect of IS is the
criticality of ensuring that the innovation is well designed
for implementation prior to deployment and dissemina-
tion at scale. In addition to the data collection described
above, NIOSH has been engaging in additional antecedent
assessment activities as defined in the CFIR 2.0 framework.
These activities align with the constructs of acceptability
and appropriateness proposed by Procter et al. [45, 46].
Since the project’s inception, NIOSH researchers have
relied on internal and external subject matter experts to
conceptualize and guide the development of VR-MRT.
Development began as a partnership with the Mine Safety
and Health Administration (MSHA) in 2020, and the
two organizations have maintained a cooperative relation-
ship throughout the development process. Additionally,
NIOSH has developed and sustained strong relationships
with multiple state agencies, mine operators, universities,
mine rescue teams, and myriad other stakeholders.
Specifically, researchers engaged in several pilot activities
to understand the feasibility of long-term implementation
including a VR mine rescue contest, a university implemen-
tation, and a general safety training. Researchers recently
co-hosted a VR mine rescue competition with collabora-
tors from the Pennsylvania Department of Environmental
Protection. The contest involved six teams completing at
least one 90-minute mine rescue problem in VR-MRT. This
contest was held at NIOSH’s Bruceton research facility and
took place over a three-day period in September 2024. All
teams were able to successfully complete the problem, and
four of the six teams chose to extend their visit to com-
plete a second 90-minute problem. NIOSH is also working
with the West Virginia University Department of Mining
Engineering to incorporate VR-MRT into their academic
curriculum and outreach activities. University staff have
helped improve the hardware specification, documenta-
tion, and software design through their testing and sugges-
tions. The staff have even become proficient at running the
system on their own. Lastly, researchers collaborated with
an underground stone mine to develop and run a VR haz-
ard identification training for use during the mine’s “Stand-
Down for Safety” day. The simulation demonstrated the
utility of VR-MRT beyond mine rescue and the ability to
provide site-specific training incorporating LiDAR scans of
the mine.
RESULTS AND DISCUSSION
Antecedent Assessment: Acceptability, Appropriateness
By talking directly to the mine rescue team members—the
innovation recipients—researchers were able to gain a bet-
ter understanding of the acceptability and appropriateness
of VR-MRT for mine rescue training. One hundred per-
cent of the respondents indicated that they enjoyed the
experience, suggesting that the demonstrated Simulation
Module of VR-MRT was acceptable.
Results suggest that opinions about VR implementa-
tion strategy did not vary with mine rescue or VR expe-
rience. The Chi-square tests of independence revealed no
significant association between participants’ mine rescue
experience and their responses to either of the implemen-
tation strategies (Addition to: χ2(2) =4.331, p =0.115
Instead of: χ2(2) =5.066, p =0.079). This was also true of
significant relationship was found, these covariates were
excluded from further analyses. Secondly, a Chi-square test
of independence was performed to determine if the par-
ticipants were generally interested in VR (i.e., yes, maybe,
no). General VR interest was determined by the highest
level of interest in VR from both implementation strategy
questions. For example, if a participants responded “no” to
“instead of,” but “yes” to “in addition to” it would be a “yes”
for general VR interest. Lastly, a marginal homogeneity
test was used to assess whether participants preferred one
implementation strategy over the other. This within-sub-
jects comparison examined the change in responses from
“instead of” to “in addition to.” Missing data were excluded
on a case-by-case basis (N =107). Significance was set at α
≤ 0.05 for all tests. Because the qualitative survey responses
were brief and sometimes incomplete, no formal analysis
was conducted. Direct quotes are included as anecdotal
support.
Interviews
Two researchers analyzed the interview notes from all 20
interviews using a combination of deductive and inductive
strategies. For this study, researchers deductively applied
the five domains and 45 constructs (Table 1) of the CFIR
2.0 framework [40] to the data. The researchers used the
framework guidance from https://cfirguide.org [47] and
the planning questions from a previous study [39] as a basis
for the codebook. From there, the researchers used a con-
stant comparison method [48] to identify “themes” made
up of actionable barriers and facilitators for each construct.
To do so, the two researchers first coded two of the inter-
views together—one mine rescue trainer and one organiza-
tional representative. The researchers then divided up the
remaining interviews and coded them independently. The
researchers met periodically while coding to refine the con-
struct and theme definitions. One hundred percent agree-
ment was achieved through discussion.
Other Pre-Implementation Activities
An important but sometimes overlooked aspect of IS is the
criticality of ensuring that the innovation is well designed
for implementation prior to deployment and dissemina-
tion at scale. In addition to the data collection described
above, NIOSH has been engaging in additional antecedent
assessment activities as defined in the CFIR 2.0 framework.
These activities align with the constructs of acceptability
and appropriateness proposed by Procter et al. [45, 46].
Since the project’s inception, NIOSH researchers have
relied on internal and external subject matter experts to
conceptualize and guide the development of VR-MRT.
Development began as a partnership with the Mine Safety
and Health Administration (MSHA) in 2020, and the
two organizations have maintained a cooperative relation-
ship throughout the development process. Additionally,
NIOSH has developed and sustained strong relationships
with multiple state agencies, mine operators, universities,
mine rescue teams, and myriad other stakeholders.
Specifically, researchers engaged in several pilot activities
to understand the feasibility of long-term implementation
including a VR mine rescue contest, a university implemen-
tation, and a general safety training. Researchers recently
co-hosted a VR mine rescue competition with collabora-
tors from the Pennsylvania Department of Environmental
Protection. The contest involved six teams completing at
least one 90-minute mine rescue problem in VR-MRT. This
contest was held at NIOSH’s Bruceton research facility and
took place over a three-day period in September 2024. All
teams were able to successfully complete the problem, and
four of the six teams chose to extend their visit to com-
plete a second 90-minute problem. NIOSH is also working
with the West Virginia University Department of Mining
Engineering to incorporate VR-MRT into their academic
curriculum and outreach activities. University staff have
helped improve the hardware specification, documenta-
tion, and software design through their testing and sugges-
tions. The staff have even become proficient at running the
system on their own. Lastly, researchers collaborated with
an underground stone mine to develop and run a VR haz-
ard identification training for use during the mine’s “Stand-
Down for Safety” day. The simulation demonstrated the
utility of VR-MRT beyond mine rescue and the ability to
provide site-specific training incorporating LiDAR scans of
the mine.
RESULTS AND DISCUSSION
Antecedent Assessment: Acceptability, Appropriateness
By talking directly to the mine rescue team members—the
innovation recipients—researchers were able to gain a bet-
ter understanding of the acceptability and appropriateness
of VR-MRT for mine rescue training. One hundred per-
cent of the respondents indicated that they enjoyed the
experience, suggesting that the demonstrated Simulation
Module of VR-MRT was acceptable.
Results suggest that opinions about VR implementa-
tion strategy did not vary with mine rescue or VR expe-
rience. The Chi-square tests of independence revealed no
significant association between participants’ mine rescue
experience and their responses to either of the implemen-
tation strategies (Addition to: χ2(2) =4.331, p =0.115
Instead of: χ2(2) =5.066, p =0.079). This was also true of