The Impact of Avalanche Education on Risk Perception, Confidence, and Decision-Making among Backcountry Skiers

Abstract

With the burgeoning popularity of winter backcountry leisure, there has been an increased interest in implementing education to prevent avalanche related accidents. Yet little research has explored whether avalanche education actually improves avalanche risk awareness. The current study examined why participants enrolled in avalanche education, what they learned, and whether participation led to changes in risk perceptions, risk perception confidence, and decision-making. Backcountry skiers and snowboarders enrolled in an introductory avalanche course in Montana (N = 184, 47.28% female) completed surveys before and after the course that included randomized photos of regional avalanche terrain. Multi-level models demonstrated that participants became more conservative in their risk perceptions and their stated willingness to ski uncertain terrain between pre- and post-test. Perception confidence increased, with larger gains for women, novices, and risk averse skiers. The results provide preliminary evidence that avalanche education may impact risk perceptions and provide useful hands-on skills for winter backcountry recreationalists.

Keywords:

• avalanche education
• decision-making
• outdoor recreation
• risk perceptions
• skiing

Introduction

A growing number of people are venturing out into the backcountry for winter recreation. Although data on the topic are sparse and precise quantification is challenging, Birkeland et al. (2017) conservatively estimated an eightfold increase in winter recreationalists between 1995 and 2017 in the western US; the Covid-19 pandemic appears to have accelerated that growth (Peipert, 2021). As a leisure activity, backcountry skiing or “free-riding” as it is sometimes known, is appealing to those desiring an exciting form of winter recreation. Qualitative research has documented the inherent challenge and freedom involved in the sport, the social connections it facilitates, and the calming focus it brings which enables individuals to “reset” for their everyday lives (Frühauf et al., 2020; Mannberg et al., 2018). Furthermore, backcountry skiing or snowboarding¹ may have physical health benefits. Previous literature suggests that downhill skiing can increase physical activity and has the potential to positively impact motor control as well as the cardiovascular and musculoskeletal system (Burtscher et al., 2018). Although backcountry riding can have numerous physiological, psychological, and social benefits, it can be considered a high-risk sport or adventure activity because of its potential for injury or even death (Breivik, 2010; Brymer & Feletti, 2020). One notable and dangerous aspect of the sport is that it exposes riders to avalanche terrain and consequently, fatal avalanche accidents.

Although rare, avalanches can have devastating and tragic consequences. Over the last ten years, on average 27 people are killed each year in avalanches in the US and many more are injured (Colorado Avalanche Information Center (CAIC), 2021); estimating non-fatal injuries, as well as partial or complete burials, is challenging because the vast majority of these non-fatal incidents go unreported (CAIC, 2021). The leading cause of death in US National Forests during wintertime, avalanches are caused by snowpack failures combined with decision-making errors. In 90% of avalanche accidents, either the victim themselves or a member of their party triggers the avalanche (Tremper, 2008). This direct connection between individuals’ actions and the avalanche event indicates that avalanche education could potentially reduce avalanche accidents by improving avalanche awareness among those with less technical knowledge of terrain and snowpack hazard.

Background: perceiving risk and making decisions in avalanche terrain

Avalanches are difficult to predict, relatively uncommon, and sometimes fatal. Riders are rarely provided opportunities to learn from their mistakes. As such, avalanche terrain has been termed a wicked learning environment due to the lack of direct feedback (see, e.g., Hogarth et al., 2015). Four factors must be present to precipitate a slab avalanche. First, the snowpack, which has fallen in layers and undergone metamorphism over time, must develop a layer of relatively weaker snow. Second, this layer of weaker snow needs to be located below a relatively stiffer snow layer above (i.e. the slab). This creates the required layer combination and interface on which the slab avalanche occurs. Third, the slope must be steep enough for the snow to overcome the friction within the snowpack. Generally those slopes are between 30° and 45° with 38° being the “bullseye” slope angle where avalanches are the most common (Tremper, 2008, p. 69). The fourth and final criteria is a “triggering event” that causes the slope to release. While this trigger can be a natural event (e.g. new snowfall, wind loading), in the case of accidents, often the trigger is the accident victim or a member of their party.

Backcountry riders face a tradeoff between the lure of untracked powder and the potential for an avalanche. When recreating in avalanche terrain, individuals must assess the perceived risk of a situation. Multiple frameworks across various disciplines explore the interplay between risk perception and decision-making. Decision heuristics—a form of mental process used in decision-making—are a critical way for people to reduce effort while making a decision to arrive at a satisfactory outcome or solution while conserving cognitive resources (Arkes & Blumer, 1985; Arnott, 1998; De Martino et al., 2006; Frame, 2012; Kahneman et al., 1982; Kahneman & Tversky, 1973; Slovic et al., 2005; Tversky & Kahneman, 1974, 1986). In the field of avalanche risk, the decision heuristics framework has dominated as found in post hoc accident analysis carried out by McCammon (Furman et al., 2010; Grímsdóttir & Mcclung, 2006; Haegeli et al., 2010, 2012; Johnson et al., 2016; Landrø et al., 2020; Landrø et al., 2020; McCammon, 2004, 2002, 2000, 2009).

Two research streams dominate the field of heuristics: the “heuristic and biases” approach - the study of systematic biases in decisions and, the “fast and frugal” program where heuristics are thought of as an effective and precise decision-making tool (Kahneman, 2011). Both recognize the limits of available information and mental processing in the effort to seek an acceptable decision (Gigerenzer & Gaissmaier, 2011; Grüne-Yanoff & Hertwig, 2016). Whereas the heuristics and bias program demonstrates circumstances under which heuristics trade off some accuracy for less effort (Tversky & Kahneman, 1974), the study of fast and frugal heuristics focus on the accuracy (or lack of) compared to rational comprehensive decision-making. The heuristics frame is appealing to avalanche educators and students because it reflects the real world of snowpack analysis. That is, risk of snowpack failure and a resultant avalanche is largely hidden and widely variable across space and time so complete and perfect information is elusive.

Photos are a useful tool to understand how individuals perceive of risks and make decisions in the backcountry. There is a broad literature in the field of outdoor recreation that uses a variety of visual methods to understand participants’ opinions, beliefs and behaviors (e.g., Cherem & Driver, 1983; Gill et al., 2015; Roberts et al., 2021; Zajchowski et al., 2019). Compared to long written descriptions, the use of photographs can increase standardization and improve ecological validity (Manning & Freimund, 2004), but attending to potential measurement biases is important (e.g., Gibson et al., 2014). With regard to avalanche-related decision-making, photos of backcountry riding capture some of the allure of the activity and, as such, they may be accompanied by feelings (e.g., of joy, excitement, anticipation, trepidation). Because these feelings may influence decision-making in avalanche terrain, these photos capture an important dimension that would be lost if the scenarios were just depicted using words. In avalanche science, Mannberg et al. (2018, 2021) and Haegeli and Strong-Cvetich (2020) and Haegeli et al. (2020) used photos for discrete choice analysis, while Stephensen et al., 2021 examined question framing in avalanche terrain (Stephensen et al., 2021). The current study follows the approach of Stephensen et al. (2021) that uses photos to examine how participants rate a photo from safe to risky, their confidence in their risk perceptions, and their stated willingness to ski a slope.

Understanding the impacts of avalanche education

In an effort to increase avalanche-related knowledge and promote safe decisions and behaviors in avalanche terrain, local and national organizations offer avalanche education courses. These vary in duration and depth of content that ranges from brief 1 to 2 hour avalanche awareness courses (e.g., Evening Awareness Class, Know Before You Go (https://kbyg.org/)) to the more time-intensive recreational and professional Avalanche Level 1 or Avalanche Level 2 courses. Individuals may enroll in these courses for various reasons (Balent et al., 2016) but the overarching goal of the recreational-focused avalanche education courses is to learn to recognize and travel safely through potentially hazardous terrain while recreating in the backcountry. These avalanche education courses represent a type of short intervention program (SIP); their purpose is to disseminate knowledge so that participants learn how to interpret relevant information about avalanches, and thereby make better choices in avalanche terrain. This approach differs from SIPs targeting other risky behaviors (e.g., drug use) that focus on eliminating all user behavior. Substantial evidence demonstrates the ineffectiveness of some of these programs that tell participants to “just say no” (e.g., D.A.R.E.) or engage in “scared straight” tactics (Petrosino et al., 2000, 2005; West & O’Neal, 2004). The goal of avalanche education differs markedly from these approaches as the goal is not to prevent use of backcountry terrain. Instead, by emphasizing when and under what conditions individuals can recreate in possible avalanche terrain, education may actually facilitate increased opportunity to recreate safely in this setting (see Haegeli et al., 2020 who make a similar point when discussing avalanche airbags). Therefore, one of the goals of avalanche education is to provide tools to enable, rather than restrict, backcountry leisure activities.

Relatively little research has explored whether and how avalanche education courses may impact participants. Some researchers have argued that despite the increased number of winter recreationalists in the backcountry, the work of avalanche educators and the availability of avalanche information and education to the public have contributed to the declining avalanche fatality rate over the past few decades (Birkeland et al., 2017). While this seems plausible, there is limited quantitative evidence to support this, or to provide insight on the relative contributions from these various sources. Avalanche education courses may influence participants’ risk perceptions, beliefs, or behaviors; the assumption is that individuals become more risk averse and safety-oriented, but this is uncertain. Supporting this possibility, one study found that backcountry riders who had participated in an avalanche safety course were more likely to carry avalanche safety gear while recreating than those who had not taken a course (Nichols et al., 2018). Another cross-sectional study found that individuals who had participated in formal avalanche education were less likely to accept to ski a risky route in a hypothetical case, than those who had not received formal education (Mannberg et al., 2018). By identifying differences between individuals with formal avalanche education and those without, these studies suggest the possibility that avalanche education may have a positive impact on participants with respect to recognizing and minimizing risk. Yet it is also possible that selection factors or preexisting differences may be responsible for these associations. That is, people who enroll in avalanche education courses may hold qualitatively different risk behaviors from those who do not enroll (e.g., those who seek out education may be more risk averse or engage in safer behaviors in the first place, or vice versa). One study did use a pre-post test design with follow-up to examine the impact of the Avaluator decision support tool combined with avalanche education curriculum (Haegeli & Haider, 2008). The results documented increases in assessment confidence and evidence of improved decision-making skills as indicated by a higher correlation between participants’ seriousness assessment and the obvious clue method at post-test.

Although education programs may have positive benefits, it is also important to recognize that a short-term avalanche education program may be insufficient to have any measurable impact on participants. Just because a curriculum or program is widespread or accepted by educators does not mean that it works, or that it is as effective as it could be. Indeed, the educational program might even have iatrogenic—or inadvertent deleterious—effects on participants (Petrosino et al., 2000). With regard to avalanche education, an example of an iatrogenic effect would be if program participation caused individuals to be excited about skiing more risky slopes or overconfident in their abilities to mitigate the avalanche hazard, thereby causing them to make riskier decisions in the backcountry. Empirically testing and describing what individuals learn in avalanche education is needed to build the evidence-base for the avalanche education field, and document any possible beneficial or detrimental impacts.

Study aims

Drawing from a sample of participants in the US state of Montana, the current study sought to better understand the possible impacts of avalanche education (see Figure 1 for the related logic model).

Figure 1. Logic model for avalanche education.

• Aim 1: Examine reasons for enrolling in the course and identify what participants learned from the course.

• Aim 2: Test whether avalanche risk perception, related confidence, or decision-making changed as a result of participating in the course.

• Aim 3: Identify whether risk perception and perception confidence predict avalanche-related decision-making.

• Aim 4: Explore whether changes (in perception, confidence, or decision-making) differed by gender, backcountry experience, risk attitudes, and program engagement.

Methods

Procedure

Eligible participants were enrolled in Avalanche Fundamentals, an avalanche safety course offered in southwestern Montana delivered by a local nonprofit organization (Friends of the Gallatin National Forest Avalanche Center). The course covers avalanche terrain, weather and snowpack, decision-making, and rescue. It is accessible to individuals with no experience in the backcountry as well as more experienced riders. Content was delivered online via Zoom^® through two classroom sessions (2 hours each) as well as 1 in-person field day (6 hours) for a total of 10 instructional hours. Thus, the course provided more instructional time than a typical 2–3 hour awareness course, but much less than the 20–24 hours for a Level 1 recreational course. The field day for Avalanche Fundamentals was socially distanced due to the Covid-19 pandemic and it occurred on a weekend day within 1–2 weeks of the classroom sessions. During the 2020–2021 season, Avalanche Fundamentals was offered twice, once in December 2020, and once in January 2021.

After obtaining an email list of enrollees, our research team sent personalized emails to participants during the week prior to the start of the course. We described the study and included a link to the survey (the “pre-survey”) via the online survey platform Qualtrics^®. Approximately 3–5 days after their scheduled field day (which represents the culmination of the course), participants were emailed a link to the second survey (the “post-survey”). A subsequent follow-up email was sent to encourage participation. To incentivize participation, respondents were entered into a raffle drawing for one of 15 $100 gift certificates to a local ski and outdoor retail shop. All procedures were approved and deemed exempt by the university’s Institutional Review Board (IRB #KG111620-EX). Surveys took about 10 minutes to complete and covered a variety of topics including risk perceptions, risk attitudes, background characteristics, reasons for enrolling (pre-survey), and information learned (post-survey).

The initial pre-survey was completed by 300 people. Because of our interest in how course participation impacted participants, we limited our analysis sample to those participants who provided their email address and completed both pre- and post-surveys (N = 198). Multivariate logistic regression analyses demonstrated that those who completed both the pre and post surveys did not differ from those who completed only the presurvey in terms of gender, race, preferred mode of travel in the backcountry or backcountry experience; however, age was negatively associated with attrition (OR = .96). Older participants were more likely to complete the post-survey (with a median age of 31.8 in the retained sample, compared with 30.9 in the larger participant cohort).

Of the participants who completed both the pre- and post-surveys, we limited our sample to individuals who reported at the pretest that their primary mode of backcountry travel was skiing or snowboarding (omitting 10 snowmobilers who viewed different photos and 3 snowshoers who viewed no snowshoe-specific photos). After omitting one participant for missing data on backcountry experience, our sample consists of one hundred and eighty-four nonmotorized backcountry riders who completed both surveys and were not predominantly snow shoers.

The sample was primarily comprised of backcountry skiers (83.7%); however, 16.3% identified snowboarding as their primary mode of winter travel. Participants ranged in age from 12 to 69 years of age (M = 31.7, SD = 11.5) and slightly over half of participants (52.7%) were male. Most participants were non-Hispanic White (94.0%) while smaller percentages self-identified as Asian American (1.6%), multiracial (2.7%) or Hispanic/Latinx (1.6%). In addition, 24.5% reported no backcountry experience, 52.7% reported being novices, 17.9% were intermediate backcountry skiers and 4.9% self-identified as experts. Previous studies have surveyed the wider population of backcountry users resulting in more experienced respondents (Hendrikx et al., 2016; Johnson & Hendrikx, 2021; Mannberg et al., 2018; Mannberg et al., 2021) whereas we targeted students in an introductory course, many of whom have little backcountry experience.

Measures

Risk perception, perception confidence, and decision-making

At both pre- and post-test a series of seven photos of regional avalanche terrain were presented in a randomized order. The photos were chosen by the authors and designed to capture varying levels of avalanche risk. Each photograph was accompanied by a brief statement describing the slope angle, the regional avalanche danger rating (held constant at “moderate”), and the snow conditions. For example, if the caption indicated a 42° slope, and current conditions indicated 2–4” of new snow, then the photo was consistent with this description with a slope that looked to be at about 42°, and for instance not snowing spring/firm conditions. Where photos were from the authors’ personal collection, recorded slope angles, and snow conditions were used to guide the photo caption to ensure that the photo captions matched the observed conditions as closely as possible. Furthermore, photos were independently calibrated on the scale from 0 (Totally Safe) to 7 (Very Dangerous) by the authors prior to the initiation of data collection.

For each scenario, participants rated the risk of the slope, their confidence in their rating, and whether they would ski the slope. For risk perception, participants were asked, “How would you rate the risk of this slope?” on a scale from 0 (Totally Safe) to 7 (Very Dangerous). Subsequently, they were asked “How confident are you in your rating?” and responded a scale from 1 (Not Confident) to 7 (Very Confident). Then participants were asked “Would you ski this slope?” with yes and no response options. Figure 2 presents an example photo – which corresponded to scenario 2. The other 6 photos that were used appear in Appendix A.

Figure 2. Example photo for survey risk perception: scenario 2.

Scenario type

The photos were classified into three categories: risky (2 photos), safe (2 photos) and uncertain (3 photos). In the multivariate models, uncertain and risky photos are compared to safe photos (the reference category). The word “uncertain” captures the participants’ perspective, as these photos are not clearly at either end of the safer or riskier continuum; they were chosen because of their ambiguity in regards to avalanche risk.

Reasons for taking the course

At pretest, participants were asked “What are your reasons for taking Avalanche Fundamentals?” Following Balent et al. (2016), a list of eight possible reasons for taking the course was presented to participants (see Table 1). The order of the reasons was randomized and participants could choose all that applied. An additional write-in option allowed respondents to identify other reasons for enrolling.

Table 1. Reasons for enrolling in avalanche fundamentals (N = 184).

Gain confidence in avalanche terrain	78.80
Learn about terrain management	75.54
Improve decision-making skills	76.63
Learn more about the snowpack	72.83
Owed it to partners to be a stronger member of my ski/riding group	69.57
To meet like-minded people or backcountry partners	21.20
Had experience; felt like it was time for a formal course	9.78
My parents wanted me to prior to backcountry skiing/riding	9.73
Other	6.52

Note: Percentages do not add to 100 because participants were instructed to “select all that apply.”

Skills learned

Participants were asked, “What are three important things that you learned from this course?” Open-ended responses were reviewed by two researchers and categorized into six groupings.

Background characteristics

Participants reported their gender identification (female, male or nonbinary/another gender). Prior work documents that most avalanche fatalities are men (Jekich et al., 2016; Peitzsch et al., 2020), and women have less confidence in their avalanche-related decision-making (McCammon, 2009; Volken et al., 2007). Furthermore, examining backcountry experience is important because novices and experts weigh avalanche information differently (Burkeljca, 2013; Hallandvik et al., 2017). Participants identified their experience by selecting from four options: No experience backcountry skiing; novice with a little experience traveling in backcountry avalanche terrain; intermediate with a few years (i.e. <5 years) experience; or expert with many years (i.e. >5 years) of backcountry experience. These categories were combined and dichotomized such that participants with five or more years of experience were compared to those with fewer than five years of experience.

Risk attitudes

Risk-taking in skiing and snowboarding has been examined through a variety of survey methods including single-item measures, scales related to the sport, general scales (i.e., not sport specific) and even person-centered analyses (Haegeli et al., 2012; Thomson et al., 2012). In the current study, we used a single item that has shown to influence decision-making across a variety of settings (Dohmen et al., 2011). We included the context-specific risk question because research demonstrates that context-specific questions have more predictive power than a general question when interested in a specific context such as skiing. Participants were asked, “When it comes to riding or snowmobiling, how willing or unwilling are you to take risks? 0 means that you are completely unwilling to take risks, and 10 means that you are very willing to take risks.” This question was used because of our interest in risk-taking during backcountry skiing; research indicates that risk-seekers prefer to ski riskier routes (Mannberg et al., 2021).

Participant engagement

Participant engagement is a multi-dimensional construct that includes the emotional affect, cognitions, and behaviors connecting the participants to a program (Mackenzie et al., 2018; Roth & Brooks-Gunn, 2016). Outdoor education can bring about optimal “flow” experiences and promote intrinsic motivation because participants are challenged, focused, and enjoying their activities (Mackenzie et al., 2018). In the current study, to measure engagement, participants read the following instructions, “These next questions are about the Avalanche Fundamentals program. Please tell us your honest opinions.” They were subsequently presented with three statements “I learned new things from the program,” The program kept my attention”, and “I’m glad I participated in the program,” each of which they rated on a scale from 1 (Strongly Agree) to 7 (Strongly Disagree). These items were reverse-coded and averaged such that higher scores indicated more engagement (Chronbach’s alpha = .77).

Analytical approach

To address aim 1, we used content analysis to examine why participants enrolled in the program and the most important lessons or skills that they learned from participation. Content analysis is a flexible analytical strategy compatible with quantitative and qualitative as well as inductive and deductive approaches (Cho & Lee, 2014; Krippendorff, 2018; Neuendorf, 2017). To analyze what the participants learned, two researchers read the responses, created a set of six categories, coded each response independently, and came to consensus. A third coder provided input on unclear discrepant responses. Code frequencies were tabulated.

To address aim 2, we conducted multivariate multilevel models examining the association between time (pre- versus post-test), gender, backcountry experience, risk attitudes, and the outcomes of interest. Because we expected that changes in perception and willingness to ski might depend on the level of danger depicted in the scenario, we tested the interaction between time and scenario danger (i.e., safe, uncertain, risky) and included it whenever it was statistically significant. Multi-level models are appropriate for the current clustered dataset. These models allow for the clustering of photos (at Level 1) within individual participants (at Level 2) and account for the associated correlated errors (Hoffman, 2015; Raudenbush & Bryk, 2002).² We calculated effect size (f²) for risk perception and confidence after using the variance explained (R²) equation for multi-level models provided by Snijders and Bosker (Lorah, 2018; Snijders & Bosker, 2012). Subsequently we included risk perception and perception confidence into the model predicting decision-making (i.e., willingness to ski the slope), to address aim 3. Finally, to address aim 4 we tested interactions between time and possible moderators (i.e., gender, backcountry experience, risk attitudes, program engagement) by including them separately in models. Linear regressions were computed for risk perception and perception confidence and logistic regressions were computed when the outcome was dichotomous (i.e., to ski/not ski a slope). We considered probability values of <.05 to be statistically significant.

Results

Reasons for participating and skills learned

The various reasons why individuals enrolled in the program are described in Table 1. About three quarters of participants noted that they hoped to gain confidence in avalanche terrain (78.8%), improve decision-making (76.6%), learn more about terrain management in avalanche terrain (75.5%), and learn about the snowpack (72.8%). Many also wanted to be a stronger member of their ski/riding group (69.6%).

After completing the course, participants reported three important things they learned in an open-ended response. 143 individuals (78% of the sample) answered this question and reported learning one or more items. Descriptions were segmented into individual items that represented a singular idea (N_items = 416). Responses fell into six categories: Terrain Assessment (e.g., slope angle/measurement, route management), Snowpack Assessment and Forecast (e.g., digging/testing snowpits, layers), Rescue Skills (e.g., using beacons, searching, digging), Human Factors and Decision-making (partners, communication, decision-making), Weather, and Other Skills/Terminology/General Statements.

Table 2 displays the proportions and frequencies of these six categories and provides corresponding example quotes from participants. Participants frequently pointed to hands-on skills that they had acquired through participation in the field day. For example, they not only learned about the slope at which avalanches are likely to occur (30–45 degrees), but they also learned the hands-on skill of using an inclinometer to estimate a slope prior to skiing it. Likewise, participants described learning to test the layers of the snow and check for instability before skiing or riding by digging a snowpit and executing a “compression test” most commonly the Extended Column Test (ECT). Supporting the value of the practical and applied nature of the field day, one participant noted, “I wanted and received hands on experience digging snow pits and evaluating the snowpack.” Furthermore, participants not only learned skills and knowledge (through both traditional and experiential methods), but some acquired new values and articulated their intent to participate in particular safety behaviors in the future. As one participant described, “I will need to continually practice beacon rescue so that I will be able to do it by muscle memory if in a scenario in which performing a beacon search is necessary.” This quote indicates that the participant has internalized the importance of a particular behavior (beacon practice) and views this future behavior as a necessary component of their backcountry skiing experience because it will be useful should they ever have a ski partner that has been buried by an avalanche and needs to be rescued. Taken together, the responses indicate that participants felt as though they had gained knowledge and skills related to perceiving avalanche risk as well as how to use their gear to locate and dig out partners should an avalanche occur.

Table 2. What individuals learned from avalanche fundamentals.

Category	N/%	Description	Example quotations
Terrain assessment	92 (60.14%)	Slope angle/measurement, route management, slope aspect, terrain features, terrain based risk recognition.	Slope Angle plays a big role in avalanche terrain. Traveling safe and away from avalanche terrain will keep you away from avalanches.
Snowpack assessment and forecast	126 (70.63%)	Digging snowpack test pits, Snowpack compression tests (ECT), understanding the avalanche forecast, making snowpack observations.	If a slope looks unstable, take some tests, and if in doubt don’t ski on it. How to dig a pit and do an extended column test to look for instability and propagation.
Rescue skills	105 (63.64%)	Using avalanche safety gear (beacon/shovel/probe), victim search patterns, emergency procedures.	How long I have to save a friend. How to better use rescue equipment.
Human factor/decision-making	59 (32.17%)	Decision-making skills, group communication, partner considerations, trip planning, habit and value acquisition.	How to make decisions on how/if to proceed in the backcountry. The importance of communication with group members.
Weather	6 (4.20%)	Weather indicators, assessment, and impact	How weather influences avalanche danger.
Other skills/ terminology/general statements	28 (18.88%)	Avalanche terminology, general statements, skills or knowledge not captured in other categories.	“Bullseye” data Definitions of avalanche terms

Notes: N is the frequency with which a category was mentioned across all responses and participants who answered the question. % indicates the percentage of respondents who described learning an item within the category. Participants could mention multiple items in a single category. Percentages do not add to 100 because participants reported three things they learned.

Risk perceptions, perception confidence, and decision-making

Central to safe riding in avalanche terrain is the ability to correctly recognize hazardous slopes under specific conditions, such that one’s risk perception is in line with the objective level of risk posed by the slope. Further, in the face of anticipation, one must be able to express the self-control needed to weigh the risk/reward and make a prudent decision based on the risk perception, and have confidence in that perception. Introductory avalanche courses introduce attendees to the basics of terrain awareness and judicious decision-making, and they aim to improve risk perception.

The results demonstrated that with respect to risk perception, participants identified uncertain and risky scenarios as more dangerous than safe scenarios (see Table 3, column 1). Furthermore, there was a Time X Scenario interaction. The change from pre- to post test in risk perception depended on the scenario type. For safe scenarios, participants reported lower perceived risk at post-test whereas for uncertain and risky scenarios participants reported higher perceived risk at post-test. Approximately 29% of the variance in risk assessment could be attributed to the time (pre-post), scenario type, and the interaction between the two. This indicates a medium-to-large effect size, f² (Cohen, 1992).

Table 3. Multi-level linear regression models predicting risk perception, perception confidence, and willingness to ski.

	Risk perception	Confidence	Would ski
	b (se)	b (se)	b (se)
Fixed effects
Time	−0.37 (0.09)***	1.55 (0.04)***	−0.28 (0.18)
Scenario type
Uncertain	0.54 (0.08)***	−0.15 (0.05)***	−0.38 (0.17)*
Risky	1.55 (0.09)***	−0.10 (0.05)*	−1.62 (0.21)***
Time X scenario type
X uncertain	0.82 (0.12)***
X risky	0.51 (0.13)***
Female	0.08 (0.09)	−0.16 (0.14)	−0.77 (0.27)**
Experienced	−0.15 (0.11)	0.72 (0.16)**	0.71 (0.31)*
Risk attitudes	0.02 (0.03)	0.11 (0.04)**	0.19 (0.08)*
Risk perception			−2.04 (0.10)***
Perception confidence			0.21 (0.07)**
Constant	3.24 (0.22)***	3.12 (0.31)***	8.62 (0.75)***
Random effects
Constant variance	0.27 (0.04)	0.73 (0.08)	2.14 (0.38)
Error variance	1.45 (0.04)	0.97 (0.03)
Nindividuals	184	184	184
Nobservations	2465	2515	2426

Notes: N_observations is the combined total number of photos scored for all of the participants in each model. The residual variance for multi-level logistic regression is π²/3 or 3.29. *p<.05, **p<.01, ***p<.001

In addition to changing how they assessed risk, participants altered their confidence in their risk perceptions of the photographed avalanche terrain. Across all scenario types, participants reported being substantially more confident in their perception at post-test than they were at pretest. Time (i.e., pre versus post-test) explained 35% of the variance in confidence; the f² value of .35 indicates that the program had a “large” effect on confidence according to standard guidelines (Cohen, 1992). In addition, participants were less confident in their perception of uncertain and risky scenarios compared to safe scenarios (Table 3, column 2).

Backcountry experience was also strongly associated with perception confidence; those who had 5 or more years in the backcountry were more confident than those who had little or no backcountry experience. Furthermore, people who self identified that they were more likely to take risks while skiing also reported being more confident in their risk perception.

Participants were less likely to report that they would ski uncertain or risky slopes depicted in the photographs. However, there was a Time X Scenario interaction (Table 3, column 3). Participants decreased their willingness to ski uncertain slopes over time, but there was evidence that they increased their willingness to ski safe slopes (model not shown). This interaction was no longer significant when risk perception and perception confidence were included in the model. As shown in Table 3, column 3, a higher perceived risk of the slope was associated with a lower likelihood of being willing to ski the slope. Confidence in one’s risk perception was associated with a higher likelihood of intending to ski the slope. Gender and experience were associated with the decision to ski the slope. Men and more experienced riders were more likely to decide to ski the slope than women and less experienced riders.

Examining differential program impact

To identify whether program impacts were consistent or varied by individual characteristics, we tested a series of interactions. These results demonstrated that increases in perceived risk from pretest to post-test were larger for individuals who were more engaged in the program (b = .427, s = .155, p < .01). Furthermore, perception confidence depended on a number of individual characteristics (see Table 4). Novices gained more confidence than more experienced riders and women gained more confidence than men. Those who were more risk averse gained more confidence than risk takers and those who were more engaged in the program gained more confidence than those who were less engaged.

Table 4. Multi-level linear regression models predicting perception confidence.

	Gender X time b (se)	Experience X time b (se)	Risk attitudes X time b (se)	Engagement X time b (se)
Fixed effects
Time	1.35 (0.05)***	1.82 (0.04)***	2.09 (0.11)***	−3.24 (.59)***
Female	−0.38 (0.14)*	−0.08 (0.14)	−0.16 (0.14)	−0.16 (0.14)
Experienced	0.72 (0.16)**	1.30 (0.16)***	0.72 (0.16)***	0.73 (0.16)***
Risk attitudes	0.11 (0.04)**	0.11 (0.04)**	0.17 (0.04)***	0.11 (0.04)**
Time X female	0.43 (0.08)*
Time X experienced		−1.16 (0.09)***
Time X risk attitudes			−0.12 (0.02)***
Time X engagement				0.71 (0.09)***
Engagement				−0.30 (0.15)*
Constant	3.06 (0.21)***	2.98 (0.31)***	2.85 (0.31)***	5.09 (1.05)***
Random effects
Individual variance	0.73 (0.08)	0.74 (0.08)	0.73 (0.08)	0.72 (0.08)
Error variance	0.96 (0.03)	0.91 (0.03)	0.96 (0.03)	0.95 (0.03)
Nindividuals	184	184	184	184
Nobservations	2515	2515	2515	2515

Notes: N_observations is the combined total number of photos scored for all of the participants in each model. Models also include scenario type. *p<.05, **p<.01, ***p<.001

Specification tests

We conducted specification tests and recomputed the main risk perception and confidence models using ordinal logistic regressions and the substantive findings remained the same. In addition, because one participant was nonbinary, there was insufficient sample size for this category. Classifying this individual as another gender did not impact the substantive results.

Discussion

Drawing from a community sample of participants attending Avalanche Fundamentals an avalanche education course in Montana, the current study suggests that an introductory program may have some important impacts on how participants evaluate terrain and perceive risk in the backcountry, and how they make decisions aimed at mitigating that risk. Furthermore, participants reported learning a variety of hands-on skills that may be useful to them during their backcountry pursuits. These are positive outcomes for avalanche education where the aim is to enable safe and appropriate use of terrain rather than simply restrict all use.

A primary goal of the current study was to understand whether participants perceived photographs of terrain-based risk with limited snowpack information differently after participating in the course. We found clear evidence that participants’ perceptions changed. For scenarios in which the avalanche danger was uncertain or in which there was a high risk of avalanches, participants evaluated the situations as more dangerous after the course. Thus, course attendance was associated with more conservative ratings of the risk of those photos. For safe scenarios, participants correctly assessed the scenario as safer after the class than they had before the class. Taken together, these findings suggest that participants learned new information about potentially risky terrain and the related avalanche hazard, and they could apply this information to novel scenarios. Further, confidence in their decision-making was enhanced after participation in the course. By exploring change over time, the current findings extend the existing literature that has primarily looked at associations between level of avalanche education and avalanche safety practices (Balent et al., 2016; Fitzgerald et al., 2016; Hendrikx et al., 2016; Hendrikx & Johnson, 2014, 2016; Mannberg et al., 2018; McCammon, 2004, 2000; Sykes et al., 2020; Tremper, 2008).

In general, the decision to ski the slope corresponded to changes in risk perception. Participants were less likely to report that they would ski uncertain scenarios and more likely to ski safe scenarios than they were at pretest. These are mostly reassuring results, given the possibility of iatrogenic effects associated with avalanche education. It would be problematic if course attendance gave participants a false sense of security or protection. This relates to the idea of risk compensation – that is, when individuals perceive themselves to have lower risk (due to education, experience, safety equipment) they may engage in additional risks to compensate for the perceived increase in their margin of safety (Hedlund, 2000; Thompson et al., 2001). A large body of literature has examined risk compensation (Thompson et al., 2001; Wilde, 1998), and support for this idea is mixed. Within the field of avalanche science, research demonstrates that a specific segment of backcountry and out-of-bounds skiers—the risk seekers—may actually engage in riskier behaviors if they are given airbags, a tool to reduce avalanche deaths (Haegeli et al., 2020). In our study, we do not find evidence that education results in riskier behaviors. On the whole, participants rated riskier and uncertain scenarios as more dangerous after having attended the course and they were more conservative/less likely to ski the slopes in these scenarios.

Before participants enrolled in the course, they hoped to gain confidence in avalanche terrain. They wanted to be able to recognize hazards, mitigate them, and continue on their tour with the confidence that they had made a good decision. In line with this reason for attending, our results indicated substantial and consistent change in participants’ confidence in their risk perception. Across all scenarios, individuals reported that they had more confidence after having attended the course. In some ways this could be considered a positive outcome. Yet, it is important to remember – especially in any extreme sport or leisure activity – that confidence can be a double-edged sword. Some confidence is a good a thing, but too much – or overconfidence—could be problematic (Groves & Varley, 2020). Related to this idea is the finding that risk attitudes predicted perception confidence. This may be because people who are willing to take risks on the mountain are confident (and perhaps overconfident) in their abilities; they think they are good at assessing terrain and avalanche risk and they believe in their likelihood of successfully skiing a slope.

Interestingly, beginners gained much more confidence in their perception than experienced backcountry enthusiasts. At pre-test, less experienced backcountry riders reported less confidence than experienced backcountry riders. By post-test, both groups reported similar confidence in their perceptions. On one hand this is an expected outcome. If students started with little confidence, then new information and training should boost confidence to new levels. The downside is that such confidence may be a manifestation of the Dunning Kruger effect whereby inexperienced participants do not realize how much knowledge is required to make quality decisions. For instance, Balent et al. (2016) used personal interviews with participants who completed an avalanche Level 1 course and found that attendees gained confidence but also expressed uncertainty about their ability to apply lessons from the classroom into the field (Balent et al., 2016). Importantly, those who completed the course, but had not yet ventured into the backcountry afterward, demonstrated greater confidence than those who had been out in the backcountry and thus better recognized the challenges involved in applied decision-making.

In terms of who changed the most as a result of participation, our findings suggest that individuals who were more engaged benefited the most. Those who were engaged in the course material – enjoying it and paying attention—gained more confidence than those who were less engaged, emphasizing the importance of providing programming material that keeps participants’ attention and is appropriately challenging. Furthermore, women and novices gained the most confidence in their ability to perceive avalanche risk. For instance, female participants started from a lower level of confidence before the class, but reported confidence similar to men after taking the course. These findings may be useful for avalanche educators seeking to target specific groups and develop tailored and evidence-based programming.

Limitations

Our findings should be interpreted in light of some limitations. The sample was drawn from attendees of a class in Southwestern Montana and therefore results may not generalize to participants attending other classes or classes in other regions utilizing a different curriculum. Nonetheless, because the majority of avalanche deaths in the US occur in the Mountain West region (i.e., Colorado, Idaho, Montana, Utah and Wyoming), this is an appropriate location for an initial study (CAIC, 2021). Another limitation relates to the photos. We purposefully used regional photos so that the scenery and topography would be context specific, and similar to what participants could experience on their backcountry excursions. Nonetheless, these photos differed in a variety of ways that might influence perceptions including photo angle (top of slope, mid slope, etc.) and weather conditions (clear blue sky, cloudy, etc.). Studies in outdoor recreation often control for as many variables as possible to isolate individual factors that are hypothesized to be important (Gibson et al., 2014); however, we did not isolate different aspects of the photo scenarios. Additional research is thus needed using a more elaborate design and more participants that systematically varies incline, slope characteristics, danger rating, and new snow to help researchers better understand the most salient factors that drive risk perception and decision-making. Furthermore, we used the same photo scenarios twice to measure change. Future research could consider substituting new, comparable images at pre- or post-test to determine change in risk perceptions; however, this approach would require additional testing to establish the measurement equivalence of the scenarios. In addition, the avalanche research community could cooperate to build a comprehensive library of representative photos to standardize the representation of risky terrain. Another limitation of the current study is the pre- post-test design. Although this design enables us identify change, randomized controlled trials or quasi-randomized designs that include a control group would be necessary to provide clear evidence of program effectiveness (Mihalic & Elliott, 2015). However, it would have been premature to conduct a randomized controlled trial or a more rigorous longitudinal evaluation if students do not demonstrate any short-term gains from the program.

Conclusions

Researchers have noted that the increase in backcountry use combined with a stable rate of avalanche fatalities should be considered a “big win” for avalanche educators and education in this domain. Yet, to date, avalanche education has lacked systematic evidence-based information with respect to efficacy. Our study is one of the first to examine what people learn in avalanche education and whether participation is associated with beneficial or harmful outcomes. We demonstrate that individuals gain skills and knowledge that may lead to improvements in risk perception, increased confidence, and safer decisions. Yet, this research is still in its infancy and program evaluation across a range of formats (e.g., awareness, Avalanche Level 1) is needed. Furthermore, important questions remain about the role of emotion in decision-making and whether program participation may lead individuals to become overconfident in their ability to assess the subtleties of terrain and snowpack. Likewise, it is possible that education may increase cumulative exposure to avalanche terrain by enabling individuals to spend more time in the backcountry and seek out more advanced terrain. Thus, the question remains whether the improvements in skills and risk perceptions obtained in avalanche education are sufficient to offset any additional exposure over time. As winter backcountry recreation becomes more popular, adventure sport participants are increasingly enrolling in avalanche education courses. Continued research is needed exploring how these avalanche education courses shape cognitive processes, risk-taking propensity, and behaviors during backcountry leisure pursuits.

Acknowledgments

The authors would like to thank the Friends of the Gallatin National Forest Avalanche Center as well as the Avalanche Fundamentals participants who completed our surveys.

Disclosure statement

The authors declare that they have no potential conflict of interest.

Additional information

Funding

This study was supported by a Scholarship and Creativity grant from Montana State University.

Notes

1 Winter backcountry recreation may include skiing, snowboard riding, snowmobiling, and those traveling on snowshoes. The current study focuses on backcountry skiers and snowboarders who we refer to as “riders.”

2 We did not include a random intercept for photo because the variance component was extremely close to zero, suggesting that the addition of a random intercept was not necessary.