What Predicts Success At Special Forces Assessment & Selection? A Comprehensive Guide

Success in the U.S. Army Special Forces Assessment and Selection (SFAS) course is no small feat. This grueling 19-20-day evaluation challenges your physical endurance, mental resilience, and ability to adapt to high-stress, unpredictable environments.

As such, a research project was conducted on 800 candidates between 2015 and 2017  by a team of researchers under a memorandum of agreement between the United States Army Special Operations Command and the United States Army Research Institute of Environmental Medicine (USARIEM). 

The selection rate during this research was a staggering 31% (247 out of 800). Many candidates had previously gone through Ranger School before attempting SFAS. 

E.K. Farina, et al. (the research team) set out to answer some very simple yet crucial questions, starting with: what separates those selected from those not? Is it purely physical fitness? Does age, geography, or life experience play a role? What about psychological traits like grit and resilience?

In this article, I will lay out the findings of that research paper to help you answer some of these questions and figure out where you need to work before attempting SFAS.

Update July (2025): I recently found another research article published in the Military Medicine journal by the same team of researchers (E.K. Farina, et al.). This time, they researched how anthropometrics and body composition predict physical performance and selection in Special Forces Assessment and Selection (SFAS) in the United States Army. I have tried to include the important metrics from this study without disrupting the flow of the original article.

1. Understanding the research methodology, Study Design, and Sample metrics

Let’s start with the methodology of this research study and what metrics were used to calculate these probabilities of success.

Update July 2025: Both studies used the same research methodology and sample space.

1.1 Sample Space

The study involved 800 active-duty male U.S. Army Soldiers who voluntarily participated in the SFAS course. Participants were recruited across 12 SFAS iterations between May 2015 and March 2017. All participants had to meet stringent eligibility criteria, including:

• U.S. citizenship.
• Minimum age of 20.
• Armed Services Vocational Aptitude Battery General Technical (ASVAB GT) score 110 or higher.
• Possession or eligibility for a security clearance and Airborne training.

The 19-20 day SFAS course evaluates candidates’ suitability for the Special Forces based on physical, psychological, and teamwork capabilities. The course comprises:

Candidates were subject to sleep deprivation (<6 hours per day on average) and negative energy balance due to reliance on military rations. Physical and psychological screenings were conducted before and during the course to gather data by the research team and United States Army John F. Kennedy Special Warfare Center and School (USAJFKSWCS) personnel.

1.2 Metrics of Success

Selection outcomes were categorized as “Selected,” “Involuntarily Withdrawn,” or “Voluntarily Withdrawn.” Key metrics included:

• Physical Performance: Road march times, run times, pull-ups, and fitness scores.
• Demographics: Age, education, marital status, geographic origin, years of service, and Ranger School experience.
• Psychological Traits:
  • Grit (measured using the Duckworth short grit scale).
  • Resilience (measured using the Connor-Davidson Resilience Scale).
  • General intelligence (IQ) and aptitude (ASVAB GT and TABE scores).
• Physiological Markers: Baseline levels of cortisol, testosterone, sex hormone binding globulin (SHBG), C-reactive protein (CRP), and neurotransmitters (e.g., epinephrine).

1.3 Data Collection and Analysis

Fasted blood samples were collected from candidates before the course to assess physiological markers. USAJFKSWCS personnel recorded physical performance and psychological measures throughout the course.

Statistical analyses were conducted using logistic regression to identify predictors of selection. The likelihood of selection was expressed as odds ratios (ORs) with confidence intervals (CIs).

Performance metrics were analyzed in quartiles, while psychological and physiological markers were treated as continuous variables. This was done, as instructed by USAJFKSWCS, to ensure data privacy and security. And to not hinder future SFAS procedures.

Next, let’s talk about the results. The overall probability of selection was 31% (247 out of 800).

2. Physical Predictors of Success in SFAS

Since we’re talking about physical predictors of success, it’ll come as no surprise that higher performance equates to better success rates. Duh!

Of course, there were some anomalies such as candidates in the third quartile of pull-ups performance had a lower success rate than those in the second quartile. However, it’s just that. An anomaly. And the p-value proved that it’s quite improbable as well.

Let me quickly explain the metrics before we dive into the statistics:

• Quartile: To protect data privacy for candidates and USAJFKSWCS, scores were categorized into quartiles which helped generalize the results. For example, people in Q1 will be in the bottom 25 percentile and candidates in Q4 are above 75% percentile (top 25%).
• % (N/Total N): This is the distribution of quartiles based on number of candidates.
• Probability of Selection: This was calculated based on 95% confidence interval (CI), that is, the statistics can ascertain with 95% surety that the probability is true.
• Odds: The Odds Ratio (or Odds) measures how much more (or less) likely selection is for one group compared to the reference group (Q1 here).
• p-value: The p-value quantifies how likely it is that the observed result occurred by chance alone. A small p-value (<0.05) will suggest that the result is statistically significant and unlikely due to chance. In contrast, a large p-value (>0.05) suggests the result is not statistically significant.

2.1 Number of pull-ups

Selection probability increases as the number of pull-ups increases. Candidates in the highest quartile (Q4) had nearly 47.5% probability of selection, compared to only 19% for the lowest quartile (Q1).

Sure, 47.5% doesn’t sound too high, does it? It’s even less than 50%. But that’s just how tough selection at SFAS actually is. 47.5% is still far higher than the total average of 31%. Every percentage counts in SFAS. 

As such, managing to rank in the Q4 (fourth quartile) would have made odds of selection 3.49 times higher than in Q1.

Let’s talk about the anomaly now. Does managing to rank in the third quartile actually lower your chances of selection? Well, no. If you look at the p-value (0.284), it’s far too high compared to our 95% confidence level (0.05), which suggests that the anomaly occurred due to chance alone.

Note: While pull-up performance correlated with the selection probability, its proportion of variation (R ² = 4.4%) suggests it plays a smaller role than other metrics like road marches or land navigation.

2.2 Land navigation points

Selection probability improves dramatically with better land navigation scores.

Candidates in the highest quartile (Q4) had an impressive 74% probability of selection, compared to 24% for Q1. This means candidates in Q4 had odds of selection that were 8.68 times higher than those in Q1.

But here’s where it gets interesting: moving from Q2 to Q3 (from 58% to 64%) or from Q3 to Q4 (64% to 74%) shows substantial improvement in selection chances.

With p-values <0.001 for all quartiles, these results are highly significant and not due to random chance.

What does this mean? Land navigation is a critical skill in SFAS. Its strong predictive power (R ² = 19.4%) makes it one of the most important metrics for selection. In fact, it has the second-highest predictive power in physical metrics.

2.3 Obstacle Course Score

Selection probability increases as obstacle course performance improves, peaking at 50% in Q3. However, it slightly dropped to 41% in Q4. Candidates in Q3 had odds of selection that were 3.45 times higher than in Q1.

Does Q4’s drop to 41% mean you should aim for the middle? Not really. The p-values indicate that Q3’s performance is significantly better (p < 0.001), but the slight decline in Q4 could be due to noise or other variables not captured here.

Overall, the obstacle course shows a moderate correlation with selection probability. Still, its proportion of variation (R ² = 6%) suggests that while it’s important, it doesn’t carry as much weight as road marches or land navigation.

2.4 APFT Score

Candidates who ranked in the top 25% (Q4) in terms of APFT scores had a 45% probability of selection, compared to just 16% in Q1. The selection odds in Q4 were 4.40 times higher than in Q1.

And the best part is that Q2 (p=0.001) and Q4 (p<0.001) show statistically significant results. Add that to their predictive value (R ² = 10.6%), and you’ll understand why APFT scores are so important.

2.5 Run Time

I have combined the tables for Run Time 1 and Run Time 2 as the statistics were very close to each other.

Faster run times strongly correlate with higher selection probabilities for obvious reasons.

Candidates in Q1 (fastest runners) had a 51-56% probability of selection, compared to only 15-19% for Q4 (slowest runners). This means Q1 candidates had odds of selection of 4.64-6.99 times higher than Q4.

Impressive, right? And with p-values all <0.001, this relationship is highly significant.

Running speed reflects endurance, strength, and cardiovascular fitness, which are crucial for SFAS success. Its proportion of variation (R ² = 11.1-14%) also shows it’s one of the stronger selection predictors.

2.6 Road March

I have combined the tables for Road March 1 and Road March 2 as the statistics were very close to each other.

Road march performance is by far the most critical predictor of selection. In Q1 (fastest marchers), candidates had a 66-67% probability of selection, compared to a mere 5-6% for Q4 (slowest marchers).

Think about it: 66-67% means two-thirds of the top-performing candidates made it through, which is extraordinary given the 31% average selection rate. The odds of selection in Q1 were 31.85-41.37 times higher than in Q4, which is massive.

And the p-values? All <0.001 for Q1 and Q2, meaning these results are statistically significant and not due to chance. The proportion of variation (R ² = 32.6%) shows that road march performance explains nearly one-third of the variance in selection outcomes, making it the strongest predictor of success in SFAS.

3. Demographic Predictors of Success in SFAS

In case you jumped directly into this section, let me quickly explain the metrics before we dive into the statistics:

• Quartile: To protect data privacy for candidates and USAJFKSWCS, scores were categorized into quartiles which helped generalize the results. For example, people in Q1 will be in the bottom 25 percentile and candidates in Q4 are above 75% percentile (top 25%).
• % (N/Total N): This is the distribution of quartiles based on number of candidates.
• Probability of Selection: This was calculated based on 95% confidence interval (CI), that is, the statistics can ascertain with 95% surety that the probability is true.
• Odds: The Odds Ratio (or Odds) measures how much more (or less) likely selection is for one group compared to the reference group (Q1 here).
• p-value: The p-value quantifies how likely it is that the observed result occurred by chance alone. A small p-value (<0.05) will suggest that the result is statistically significant and unlikely due to chance. In contrast, a large p-value (>0.05) suggests the result is not statistically significant.

3.1 Age

Age may influence selection, but it isn’t a decisive factor at SFAS.

Candidates aged 25 years or older had a 34% probability of selection, compared to 27% for those aged 18-24. However, the p-value (0.096) indicates this result isn’t statistically significant.

3.2 Education

Your level of education obviously has a good impact on your probability of selection. Candidates with a Bachelor’s degree or higher had the highest probability of selection (43%), compared to 23% for high school graduates.

The odds of selection for candidates with a Bachelor’s degree were 2.79 times higher than those with only a high school diploma, and the p-value (<0.001) indicates these results are statistically significant.

3.3 Officer/Enlisted Status

Officers had the highest probability of selection (46%), followed by 18X Enlisted candidates (40%), and finally, Active Duty enlisted soldiers (19%).

As indicated by the higher odds of selection (3.63 and 2.85 respectively) and p-value (0.001), the research results are statistically significant. It can be concluded that officer training and 18X preparation programs likely give candidates an edge at SFAS.

3.4 Marital Status

Unmarried candidates had a 34% probability of being selected, compared to 26% for married candidates.

Implication? While being married isn’t a disqualifier, candidates without family obligations might better handle the demands of SFAS.

With the p-value being 0.016, the results did not occur due to chance. However, the gap itself (8%) and the proportion of variation (R ² = 1.0%) imply that the difference carries little to no weight in SFAS selection.

3.5 Number of Children

Candidates without children were far more likely to be selected (34% probability) than those with children (20% probability).

Something similar to marital status can be concluded here. On one side of the coin, the p-value is less than 0.001 (great statistical significance) – it might mean that family responsibilities could impact a candidate’s ability to perform at SFAS.

However, the proportion of variation (R ² = 2.6%) suggests that the difference in probability holds little to no weight in SFAS selection.

3.6 Ranger School

Ranger School graduates had a significantly higher probability of selection (50%) than non-graduates (29%). The odds of selection for Ranger School graduates were 2.45 times higher, and the result is statistically significant (p < 0.001).

3.7 Geographic Area of Origin

Selection probability varies by region, well, sort of.

Candidates from the Mountain region had the highest probability of selection (46%), followed by New England (40%) and South Central (34%).

But all the results aren’t equally important. Except for the Mountain region (p = 0.009), every other region did not showcase statistically significant results.

3.8 Community of Origin

Community type doesn’t significantly impact selection probability. Candidates from urban, suburban, and rural backgrounds all had an average 32% probability of selection.

The selection odds for suburban and rural candidates were slightly higher (1.09-1.10), but the p-values (0.699-0.725) indicate these differences are not statistically significant.

My takeaway is that success at SFAS is not tied to where you grew up-it’s about how you perform.

3.9 Years of Service

Selection probability was highest for candidates with less than 1 year of service (41%), followed by those with 3-4 years of service (32%). 

Candidates with 5 or more years of service had the lowest probability (20%).

The selection odds were 2.68 times higher for candidates with less than 1 year of service, and this result is statistically significant (p < 0.001). Conversely, candidates with 1-2 years of service had reduced odds (0.69) but the results might have occurred due to chance (p = 0.236).

What does this mean? Fresh recruits and moderately experienced soldiers outperform their more seasoned counterparts at SFAS, possibly due to physical readiness or adaptability.

This aligns with our prior research findings that civilians on 18X contracts performed best at selection.

4. Psychological Predictors of Success in SFAS

In case you jumped directly into this section, let me quickly explain the metrics before we dive into the statistics:

• Quartile: To protect data privacy for candidates and USAJFKSWCS, scores were categorized into quartiles which helped generalize the results. For example, people in Q1 will be in the bottom 25 percentile and candidates in Q4 are above 75% percentile (top 25%).
• % (N/Total N): This is the distribution of quartiles based on number of candidates.
• Probability of Selection: This was calculated based on 95% confidence interval (CI), that is, the statistics can ascertain with 95% surety that the probability is true.
• Odds: The Odds Ratio (or Odds) measures how much more (or less) likely selection is for one group compared to the reference group (Q1 here).
• p-value: The p-value quantifies how likely it is that the observed result occurred by chance alone. A small p-value (<0.05) will suggest that the result is statistically significant and unlikely due to chance. In contrast, a large p-value (>0.05) suggests the result is not statistically significant.

Note: Psychological traits such as grit and resilience were assessed through self-reported questionnaires, which might introduce biases.

4.1 TABE Score

Achieving the maximum score in  TABE (Test of Adult Basic Education) is a clear advantage, highlighting the importance of basic academic proficiency.

Candidates who achieved 12.9 had a 40% probability of selection, compared to 30% for those below this level. The odds of selection were 1.68 times higher for candidates with a score of 12.9, and this result is statistically significant (p = 0.002).

4.2 Grit Score

Candidates in grit’s highest quartile (Q4) had a 36% probability of selection, compared to 26% for the lowest quartile (Q1).

The selection odds were 1.63 times higher for Q4 than Q1, which is statistically significant (p = 0.025). However, the intermediate quartiles (Q2 and Q3) showed inconsistent results, with p-values suggesting these effects might be due to chance.

What do I think? While grit positively correlates with selection, its proportion of variation is modest. We know you need to be mentally resilient at SFAS, and this test calls into question how relevant a self-attested grit score is.

4.3 CD-RISC Score

As measured by the CD-RISC, resilience is a critical psychological factor for SFAS success.

The selection odds were 2.27 times higher for Q4 compared to Q1, which is highly significant (p < 0.001). Intermediate quartiles (Q2 and Q3) also strongly correlated with selection probability with p-values 0.001 and 0.003 respectively.

4.4 ASVAB GT Score

Strong cognitive abilities, as indicated by the ASVAB GT score, positively influence selection probability at SFAS.

Selection probability increases with higher ASVAB GT scores. Candidates in the highest quartile (Q4) had a 39% probability of selection, compared to 25% for the lowest quartile (Q1).

The selection odds were 2.19 times higher for Q4 compared to Q1, with the result being statistically significant (p = 0.002). However, the middle quartiles (Q2 and Q3) showed weaker and non-significant relationships.

4.5 VIQ Score (Verbal IQ)

Verbal intelligence plays a substantial role in SFAS success, particularly for candidates scoring in the top quartile.

The odds of selection for Q4 candidates were 2.50 times higher, and this result is highly significant (p < 0.001). Lower quartiles (Q2 and Q3) showed mixed effects, with some non-significant results.

4.6 PIQ Score (Performance IQ)

Strong performance-based cognitive abilities (e.g., spatial reasoning and problem-solving) are valuable predictors of success at SFAS.

Candidates in the highest quartile (Q4) of PIQ had a 44% probability of selection, compared to 25% for the lowest quartile (Q1).

The selection odds were 2.32 times higher for Q4 than Q1, with statistically significant results (p < 0.001).

4.7 FSIQ Score (Full-Scale IQ)

As measured by the FSIQ, General intelligence is one of the most important psychological predictors of success at SFAS.

Selection probability dramatically increases with higher FSIQ scores. Candidates in the highest quartile (Q4) had a 48% probability of selection, compared to only 23% for the lowest quartile (Q1).

The odds of selection for Q4 candidates were 3.18 times higher compared to Q1, and this result is highly significant (p < 0.001).

5. Physiological Predictors of Success in SFAS

In case you jumped directly into this section, let me quickly explain the metrics before we dive into the statistics:

• Quartile: To protect data privacy for candidates and USAJFKSWCS, scores were categorized into quartiles which helped generalize the results. For example, people in Q1 will be in the bottom 25 percentile and candidates in Q4 are above 75% percentile (top 25%).
• % (N/Total N): This is the distribution of quartiles based on number of candidates.
• Probability of Selection: This was calculated based on 95% confidence interval (CI), that is, the statistics can ascertain with 95% surety that the probability is true.
• Odds: The Odds Ratio (or Odds) measures how much more (or less) likely selection is for one group compared to the reference group (Q1 here).
• p-value: The p-value quantifies how likely it is that the observed result occurred by chance alone. A small p-value (<0.05) will suggest that the result is statistically significant and unlikely due to chance. In contrast, a large p-value (>0.05) suggests the result is not statistically significant.

5.1 C-Reactive Protein (CRP), nmol/L

CRP is a marker of inflammation in the body. Higher levels indicate stress or infection, while lower levels are linked to better recovery and overall health.

Candidates with CRP levels below 9.5 nmol/L had a 33% probability of selection, compared to only 24% for those with higher stats. The selection odds were 1.65 times higher for candidates with lower CRP, and this result is statistically significant (p = 0.012).

Takeaway: Keeping inflammation low through proper recovery and overall health management may improve selection odds.

5.2 Cortisol, nmol/L

Cortisol is a stress hormone that helps regulate energy and response to challenges. Moderate-to-high cortisol levels (not excessively high) may indicate better stress resilience, contributing to selection success.

Candidates in the highest quartiles of cortisol (Q3 and Q4) had a 35% probability of selection, compared to 25% for Q1. The odds of selection for Q3 and Q4 were 1.63 and 1.62 times higher, respectively, with statistically significant results (p = 0.027 and p = 0.030).

5.3 DHEA-s, μmol/L

DHEA-s is a hormone linked to energy, stress tolerance, and physical performance. Higher levels suggest better physiological readiness.

Interestingly, candidates in the lowest quartile (Q1) had the highest probability of selection (36%), while those in higher quartiles (Q2-Q4) had slightly lower probabilities (~28%-30%).

That said, the results in Q3 (p = 0.076) and Q4 (p = 0.158) suggest that they were not statistically significant enough to draw any conclusions. 

5.4 Testosterone, nmol/L

Testosterone is a key hormone for muscle mass, recovery, and physical performance. Moderate-to-high levels support endurance and strength.

Candidates in the highest quartile (Q4) of testosterone had a 35% probability of selection, compared to 30% in Q1. However, the selection odds for Q4 were only 1.28 times higher, and the result was not statistically significant either (p = 0.255).

5.5 Sex Hormone-Binding Globulin (SHBG), nmol/L

SHBG regulates the availability of hormones like testosterone. Higher levels often indicate hormonal balance and health.

Candidates in the highest quartile (Q4) of SHBG had a 36% probability of selection, compared to 28% in Q1. The selection odds for Q4 were 1.48 times higher, but this result was only marginally significant (p = 0.069), if at all.

5.6 Epinephrine, pmol/L

Epinephrine, or adrenaline, is a stress hormone that drives the fight-or-flight response. Balanced levels indicate effective stress response.

Candidates in the lowest quartile (Q1) of epinephrine had the highest probability of selection (35%) compared to 25%-34% in other quartiles. The selection odds were significantly lower in Q3 (odds = 0.60, p = 0.019).

My takeaway is that lower epinephrine levels may reflect better stress regulation, increasing selection probability.

5.7 Norepinephrine, pmol/L

Norepinephrine, another stress hormone, helps regulate focus, energy, and the body’s response to stress.

Candidates in the lowest quartile (Q1) of norepinephrine had the highest probability of selection (35%), while other quartiles had probabilities ranging from 28%-32%.

The selection odds slightly decreased in higher quartiles, though none of the results were statistically significant (p > 0.05).

6. Anthropomorphic and Body Composition Predictors of Success in SFAS

This is part of the July 2025 update of this article. The second research paper primarily focuses on Anthropomorphic factors and body composition.

Let me quickly explain the metrics before we dive into the statistics:

• Quartile: To protect data privacy for candidates and USAJFKSWCS, scores were categorized into quartiles which helped generalize the results. For example, people in Q1 will be in the bottom 25 percentile and candidates in Q4 are above 75% percentile (top 25%).
• Mean ± SD: The average and standard deviation for the variable across the sample.             |
• Predicted Probability of Selection: Modeled likelihood of being selected at each quartile.
• p-value: Statistical significance of the model or difference between quartiles.
• Δ (Delta): Change in predicted probability of selection relative to Q1 (the baseline group).
• OR (Odds Ratio): The odds of selection for each quartile compared to Q1.
• 95% CI: Confidence interval for the odds ratio – tells us the range of likely values.
• Model R ²: Indicates how well the variable predicts selection (higher = better).

6.1 Height (cm)

Candidates in the tallest quartile (Q4) had a 37.9% chance of selection, nearly doubling the 19.1% probability seen in the shortest group (Q1). Statistically, the odds of selection for Q4 candidates were 2.58 times higher than those in Q1.

The trend holds steady across quartiles, with each step up in height corresponding to a meaningful boost in selection odds. All p-values are below 0.01, confirming statistical significance.

However, the overall predictive power is modest since the R ² value is merely 0.03. It means that height alone accounts for approximately 3% of the variance in selection outcomes. So while taller candidates may have an edge, height is only one small piece of the puzzle.

This category had another table: Height (cm)-adjusted for body mass (kg)

After adjusting for body mass, taller candidates still showed a clear advantage in selection odds. Q2, Q3, and Q4 all had notably higher selection probabilities compared to Q1, with Q2 peaking at 33.6%.

However, only Q2’s improvement (p = 0.024) reached statistical significance. Q3 and Q4 showed promising trends, but their p-values (0.070 and 0.071) were just above the typical 0.05 threshold. The data suggests these differences could be due to happenstance.

The model R ² of 0.06 means height (adjusted for body mass) explains about 6% of the variance in selection outcomes.

6.2 Body mass (kg)

Selection probability rose sharply with an increase in body mass. While candidates in Q1 had a 16.5% chance of being selected, that number almost doubled for Q2 (30.3%) and peaked at 39.1% in Q4.

These differences were statistically significant across all higher quartiles. Q4 candidates had 3.25 times higher odds of selection than Q1, with a tight confidence interval (CI: 2.03-5.20), reinforcing the strength of this relationship.

With a model R ² of 0.06, body mass explained 6% of the variation in selection outcomes-making it one of the more meaningful anthropometric predictors in this study.

This category had another table: Body mass (kg)-adjusted for height (cm)

Even after adjusting for height, heavier candidates had significantly higher chances of being selected at SFAS. Candidates in Q3 and Q4 had a selection probability of around 36%-nearly double the 18.2% seen in Q1.

Q3 showed the highest odds of selection (2.57), followed closely by Q4 (2.52). All comparisons except Q1 vs Q3 were statistically significant, with p-values well below the 0.01 threshold.

With a model R ² of 0.06, body mass adjusted for height explained 6% of the variation in selection.

6.3 BMI (kg/m²)

Candidates with higher BMI generally had better odds of selection. Q2 showed the highest selection probability (37.6%) and the highest odds ratio (2.82), followed closely by Q3 and Q4.

That said, the relationship is not perfectly linear. Q4 (the heaviest group) saw a slight drop in predicted selection probability compared to Q2 and Q3, though still significantly above Q1.

All p-values indicate statistical significance (p ≤ 0.001), confirming that these findings are unlikely due to chance. The model R ² here is 0.04, indicating that BMI explains approximately 4% of the variation in selection. BMI is obviously not the full picture, but I call it a useful proxy.

6.4 Body fat (%)

Candidates in the leanest quartile (Q1) had the highest selection probability (51.6%), while those in Q4 (highest body fat) dropped to just 13.8%.

The odds of selection in Q4 were 85% lower than in Q1 (OR = 0.15), a statistically significant difference (p = 0.003). The steep drop from Q1 to Q3 and Q4 indicates that excess body fat is a strong negative predictor for SFAS success.

The model R ² value here is 0.13-indicating that body fat alone accounts for 13% of the variation in selection outcomes. That makes it one of the strongest single predictors in this study.

6.5 Fat mass (kg)

Candidates with the lowest fat mass (Q1) had a 43.3% chance of selection, which is significantly higher than the 17.2% observed in Q4. The odds of selection in Q4 were 73% lower than in Q1 (OR = 0.27), a statistically significant result (p = 0.034).

That said, Q2 and Q3 were not statistically significant (p > 0.05), indicating that only extreme increases in fat mass notably reduce selection odds.

The model R ² = 0.07, meaning fat mass explains 7% of the variation in SFAS selection outcomes-moderate predictive power, but not as strong as total body fat percentage.

6.6 Lean mass (kg)

Lean mass showed a positive correlation with selection probability. Candidates in Q4 (highest lean mass) had a 58.6% chance of selection-nearly triple the 20.0% observed in Q1. Their odds of selection were 5.67 times higher and were statistically significant (p = 0.003).

However, Q2 and Q3 results weren’t statistically significant (p > 0.05), meaning the substantial increase in selection odds was only clear at the uppermost quartile.

Model R ² was 0.12, suggesting lean mass explained 12% of the variation in selection outcomes. It is a surprisingly strong predictor compared to others in the body composition category despite the confusion in Q2 and Q3.

6.7 Bone Mineral Content (kg)

Higher BMC was associated with improved selection probability, especially in Q4 where candidates had a 51.7% chance of selection versus 26.7% in Q1. The odds of selection for Q4 were nearly 3x greater, though the p-value of 0.052 falls just short of the standard 0.05 threshold for statistical significance.

Lower quartiles (Q2 and Q3) did not demonstrate statistically significant improvements. In fact, Q2 showed slightly worse outcomes than Q1, though not meaningfully so.

With an R ² of 0.07, BMC explained about 7% of the variance in selection-a moderate predictor with potential relevance when used correctly.

6.8 Body Mineral Density (g/cm²)

Bone mineral density (BMD) emerged as one of the most powerful predictors of selection in this study. Candidates in the highest quartile (Q4) had a 57.1% chance of selection-nearly triple the probability seen in Q1. With an odds ratio of 5.33 and a statistically significant p-value (0.005), the relationship is robust.

The effect becomes clear in Q3 as well, where selection odds were 3.5 times higher than in Q1. While Q2 dipped unexpectedly, its wide confidence interval and non-significant p-value (0.527) suggest that this drop is likely due to random noise.

Most importantly, BMD had the highest explanatory power among all anthropometric variables studied, with an R ² of 0.19. That means it alone explained nearly 19% of the variance in selection outcomes.

7. Understanding Study Limitations

While this study provides valuable insights, it’s important to understand certain limitations before drawing real-life conclusions for yourself.

Physiological markers such as DHEA-S, epinephrine, norepinephrine, and cortisol were measured only once before the start of the course. Since researchers were not allowed to interfere with selection procedures, tracking these markers dynamically in response to stress was impossible. Consequently, the study could not capture how acute stress might alter these markers, which would have provided a more direct correlation to candidate performance.

A single measurement may not accurately reflect chronic testosterone levels. Multiple measurements over time would have provided a more robust assessment of how testosterone levels correlate with physical performance. Chronic exposure to higher testosterone levels likely has greater effects on muscle mass and strength, which are key predictors of SFAS success, especially during pre-course training.

As mentioned earlier, psychological assessments were all self-attested, which may have introduced biases. Self-assessment tools can be influenced by participants’ subjective perceptions, potentially limiting their accuracy compared to standardized intelligence or aptitude tests.

Source:

1. Physical performance, demographic, psychological, and physiological predictors of success in the U.S. Army Special Forces Assessment and Selection course by Emily K. Farina, Lauren A. Thompson, Joseph J. Knapik, Stefan M. Pasiakos, James P. McClung, and Harris R. Lieberman (link)
2. Anthropometrics and Body Composition Predict Physical Performance and Selection to Attend Special Forces Training in United States Army Soldiers by Emily K. Farina, Lauren A. Thompson, Joseph J. Knapik, Stefan M. Pasiakos, James P. McClung, Harris R. Lieberman (Military Medicine, Volume 187, Issue 11-12, Nov-Dec 2022, link)