Optimizing Resistance Training Outcomes: Comparing In-Person Supervision, Online Coaching, and Self-Guided Approaches: A Randomized Controlled Trial

Abstract

Gavanda, S, Held, S, Schrey, S, Oberwetter, K, Lazzaro, P-GM, Pergelt, M, and Geisler, S. Optimizing resistance training outcomes: comparing in-person supervision, online coaching, and self-guided approaches: a randomized controlled trial. J Strength Cond Res 39(11): 1129–1137, 2025—This randomized, parallel-group trial investigated the effects of supervised (SUP), app-guided (APP), and self-guided (PDF) 10-week, thrice-weekly full-body resistance training (RT) on strength, body composition, well-being, and supervision satisfaction (S-SRQ) in trained men and women (n = 79, 48% women; 30.7 ± 7.8 years, 1.75 ± 0.1 m, 77.5 ± 17.5 kg). Adherence was highest in SUP (88.2%), followed by APP (81.2%) and PDF (52.2%). At p ≤ 0.05, body mass (+1.8 ± 1.9 kg, p = 0.006) and fat-free mass (+1.4 ± 0.9 kg, p ≤ 0.001) increased significantly in SUP. Muscle mass gains were observed in SUP (+1.4 ± 0.9 kg, p = 0.009) and PDF (+0.9 ± 1.3 kg, p = 0.047). All groups improved squat 1-repetition maximum (1RM) (SUP: +26.6 ± 6.5 kg, p ≤ 0.001; APP: +19.2 ± 11.0 kg, p ≤ 0.001; PDF: +19.4 ± 11.7 kg, p ≤ 0.001) and bench press 1RM (SUP: +9.1 ± 3.5 kg, p ≤ 0.001; APP: +8.2 ± 4.0 kg, p ≤ 0.001; PDF: +7.7 ± 5.8 kg, p ≤ 0.001). Supervised showed significantly greater squat gains than APP and PDF (p ≤ 0.044). Well-being (WHO-5) improved in SUP (+15.7 ± 16.2 points, p ≤ 0.001) and PDF (+9.0 ± 20.4 points, p = 0.032). Satisfaction with supervision was significantly higher in SUP (96.7 ± 4.3%) than in APP (92.0 ± 7.1%, p = 0.005). In conclusion, supervised RT resulted in superior improvements in strength, body composition, well-being, and supervision satisfaction compared with app-guided or self-guided training. Although APP and PDF resulted in some positive effects, their magnitude was generally smaller. These findings underscore the value of in-person coaching in optimizing RT outcomes. However, app-based RT shows promise for maintaining adherence, offering a viable alternative when full supervision is not feasible.

Introduction

Resistance training (RT) is widely recognized for its benefits in enhancing muscular strength, hypertrophy, and overall physical performance, making it a vital component of fitness and health-oriented exercise programs (1). Although RT is frequently conducted under supervision in fitness or health-related facilities, a significant portion of individuals opt for unsupervised or minimally supervised training (7). Notably, supervised RT offers key advantages, such as enhanced exercise technique, increased adherence, higher training intensities, and better outcomes in muscle strength and hypertrophy (25,35).

Despite the clear advantages of supervised training, a gap remains in understanding its specific effects compared with unsupervised approaches, particularly as technological advancements introduce new hybrid models of RT guidance, such as app-based supervision. The importance of online training platforms, such as training apps, in facilitating RT is becoming increasingly relevant, with mobile exercise apps recently rising to rank 2 on the ACSM's Worldwide Fitness Trends list (28). Emerging research on app-supported RT indicates that these platforms can offer personalized programming, instructional videos, guidance, progress tracking, and educational resources, potentially bridging the gap between fully supervised and entirely self-directed training (23,26). However, the effectiveness of app-based RT relative to in-person supervision, particularly in trained individuals, remains insufficiently explored. App-based RT lacks the immediate, in-person feedback and motivational support provided by face-to-face supervision, which can enhance adherence and physical outcomes, especially for novice trainees (11). Thus, further research is needed to determine whether app-based guidance can serve as an effective alternative or complement to traditional supervision.

For example, studies by Mazzetti et al. (25) and Coleman et al. (6) demonstrate greater muscular adaptations with supervised compared with unsupervised RT. In contrast, a recent meta-analysis suggests that supervision provides only small-to-moderate benefits for physical performance and body composition in both novice and experienced athletes, implying that its incremental advantages may depend on individual factors and the training context (9). However, the integration of app-guided RT as a remote coaching hybrid between supervised and unsupervised training, distinct from self-guided training, remains underexplored (13,35,42).

Against this background, we aimed to address this research gap by comparing the effects of supervised RT, self-guided unsupervised RT, and app-guided RT, on maximal strength, body composition, and subjective well-being in resistance-trained adults. This approach allows us to investigate whether digital, app-based supervision can replicate some of the benefits of traditional in-person coaching, providing new insights into the role of emerging technologies in RT guidance. Furthermore, it clarifies the impact of different supervision levels on RT outcomes. We hypothesize that supervised RT will result in greater improvements in maximal strength, body composition, and well-being than app-guided and self-guided training. Our findings may contribute to the optimization of RT programs across various settings and inform the development of more effective and accessible unsupervised or app-guided RT interventions.

Methods

Experimental Approach to the Problem

The study was conducted using a randomized, parallel-group, repeated measures design to evaluate the effects of a 10-week, thrice-weekly full-body RT intervention. After a 1-week familiarization period with 3 sessions per week, during which subjects practiced the maximal strength tests and became accustomed to the required RT exercises, they were randomly assigned to 1 of 3 intervention groups: (a) supervised training (SUP), where a certified coach monitored adherence, provided encouragement, immediate guidance on load progression, and verbal feedback on exercise technique, with a coach–subject ratio of 1:1 to 1:4; (b) app-guided training (APP), which included instructional videos on exercise technique, adherence monitoring, progress tracking, and time-delayed feedback on execution technique through video; and (c) training provided in PDF format (PDF), with no control mechanisms except for adherence monitoring at the end of the intervention through completed training logs.

The primary outcome measure of this study was maximum strength, assessed by the 1-repetition maximum (1RM) in the squat and bench press, which represent lower and upper body strength, respectively. Secondary outcomes included body composition, subjective well-being, and the quality of the relationship between the coach (supervisor) and the subject (supervisee). All testing was done at the same time of the day. To minimize dietary changes as a confounding variable, all subjects were instructed to maintain their usual eating habits, including supplement consumption, throughout the study period.

Subjects

Based on the literature suggesting small-to-moderate effects of supervision on muscle strength (9), we conducted an a priori power analysis using G*Power (version 3.1.9.6) (7). Assuming a small-to-moderate effect size for the group × time interaction (partial eta-squared [pes] = 0.05, corresponding to f = 0.23), an α-level of 0.05, and a power of 0.95, a repeated-measures ANOVA with 3 groups and 2 measurement time points indicated that a total of n = 78 would be required. Assuming a relevant dropout rate, 112 subjects were enrolled in this randomized controlled interventional trial. Given an adherence rate of at least 85% (26 out of 30 sessions), 79 subjects were included in the final analysis (Table 1). The reasons for dropping out were nonstudy-related injury (n = 5), illness (n = 20), and personal reasons (n = 8). All subjects were informed of the benefits and risks of the investigation, at least 18 years old (age range: 18–58 years), had no health impairments, had at least 1 year of RT experience, including all exercises used in this study, and were familiarized with the test and training procedures before the study began. All subjects provided written informed consent and agreed to abstain from additional training during the intervention period. The study protocol adhered to the Declaration of Helsinki and received approval from the local ethics committee (Number 062021) (15,43). There were no significant differences (p ≥ 0.135; p ≤ 0.10) between groups in terms of height, body mass (BM), and training experience (Table 1). However, subjects in the SUP group were significantly younger than those in the APP and PDF groups. With respect to the >85% adherence threshold, 88.2% (15 of 17) of SUP, 81.2% (39 of 48) of APP, and 52.2% (25 of 47) of PDF subjects met the required training session completion criteria.

Table 1.

Anthropometric data of the personal training (PT), training guided through an online application (APP), and training plan provided in PDF format only (PDF) training group.^*†

Parameter	SUP	APP	PDF	ANOVA
N	15 (6 women)	39 (20 women)	25 (12 women)	—
Age (y)	25.9 ± 4.9	31.8 ± 8.1	31.9 ± 7.9	p = 0.026, pes = 0.10
Height (cm)	175.6 ± 10.4	174.9 ± 10.1	174.4 ± 10.9	p = 0.936, pes = 0.00
Body mass (kg)	81.3 ± 20.5	76.1 ± 18.2	76.7 ± 14.6	p = 0.603, pes = 0.00
RT experience (y)	2.8 ± 2.4	5.2 ± 4	4.5 ± 4.2	p = 0.135, pes = 0.10

^*N = sample size; RT = resistance training.

^†Data are given as mean ± SD. In addition, p value and effect size (partial eta squared [pes]) of the 1 × 3 ANOVA are given.

Procedures

Body Composition

Subjects' BM was measured using an electronic scale (Seca 803, Hamburg, Germany) while they wore only underwear, without socks or shoes. Fat mass (FM), fat-free mass (FFM), and muscle mass (MM) were assessed using bioelectrical impedance analysis (BIA) (Akern BIA 101, Florence, Italy) and the BodyGramPro software (Version 3.0, Akern, Florence, Italy). Before the BIA measurement, subjects were instructed to avoid eating, drinking, or engaging in intense physical activity for at least 2 hours before the assessment. They rested in a supine position for 10 minutes to stabilize body fluids and improve measurement accuracy. Bioelectrical impedance analysis 101 has demonstrated good reliability for body composition assessments, with intraclass correlation coefficients (ICC) of 0.81 (30).

Maximum Strength

Before strength testing, subjects completed a standardized warm-up protocol. This included 5 minutes of low-intensity jogging, followed by a 5-minute dynamic mobility routine targeting all major joints. Subsequently, warm-up sets of 10, 5, 3, and 1 repetitions with 50, 65, 80, and 90% of subjects' estimated 1RM, followed by 1, 2, 3, and 4 minutes of rest, were performed for the squat as recommended elsewhere (10). The weight was then increased or reduced until no more valid repetition could be performed. An attempt was considered valid if the depth of the squat was sufficient and could be completed without assistance from the spotter. The depth of the squat was standardized using a “touch and go” box squat, with box height selected to ensure that the subject's hip crease was at least below the knee. Barbell placement was standardized as high bar (on the upper traps), while stance and grip were left to the subject's preference. Strong verbal encouragement was given by the research team for all 1RM attempts. After each 1RM attempt, subjects were given a 4-minute rest period. The goal was to determine the 1RM within a maximum of 5 attempts. The 1RM bench press was then performed following the same procedure. A repetition was considered valid if the bar touched the chest and could be brought to full arm extension without assistance, while maintaining contact with the bench (head, shoulders, buttocks) and both feet on the ground. The median ICC for 1RM tests was reported to be 0.97, indicating excellent reliability (12).

Subjective Well-Being

Subjects' subjective well-being was assessed using the German version of the 5-item World Health Organization Well-Being Index (WHO-5) (41). It consists of 5 items rated on a 5-point Likert scale, reflecting well-being experienced for the past 2 weeks. Scores from the 5 items are summed to produce a raw score between 0 and 25, which is then multiplied by 4, yielding a total score between 0 and 100, where 0 represents the lowest possible well-being and 100 represents the highest. The WHO-5 is considered highly reliable (36).

Short Supervisory Relationship Questionnaire

At the end of the intervention phase, subject satisfaction with supervision was assessed using the Short Supervisory Relationship Questionnaire (S-SRQ), a validated and reliable tool for evaluating supervisory satisfaction (4). The S-SRQ comprises 3 subscales—safety, education, and structure—measured through 18 items rated on a 7-point Likert scale. In the SUP group, all 18 items were completed. However, in the APP group, only the 14 items related to safety and education were assessed because of the absence of in-person meetings. Higher scores on the S-SRQ indicate greater satisfaction with supervision. Previous research has demonstrated high reliability of the S-SRQ (4).

Resistance Training Intervention

Subjects in all 3 groups followed an identical exercise protocol, differing only in the level of supervision and support provided. The SUP group received individualized in-person coaching from certified trainers during each session, offering real-time feedback and personalized adjustments to exercise technique and load. The APP group used an online application (TrueCoach, Boulder, CO) for guidance, which included video demonstrations, progress tracking, and automated reminders, but lacked direct supervision. The PDF group was provided with a static, downloadable training plan in PDF format, containing written instructions for load progression rules, exercises, sets, repetitions, tempo, and rest intervals, facilitating entirely self-directed training.

The 10-week training program was structured into 2 distinct mesocycles. The first mesocycle, lasting 6 weeks, was designed to promote muscle hypertrophy. The second mesocycle, spanning 4 weeks, focused on maximal strength development. The program concluded with a deload week, during which training volume was reduced to mitigate fatigue, lower the risk of overtraining, and enhance recovery in preparation for final measurements.

The program followed a full-body workout design (see Table 2 for exercise selection and order) and was conducted 3 times per week on nonconsecutive days to ensure sufficient recovery. Each session began with a warm-up that included 5–10 minutes of low-intensity general activity, such as jogging or rowing, followed by 5–10 minutes of specific warm-up exercises including bodyweight squats, lunges, and leg swings. The main lifts adhered to a linear periodization model with progressive intensity increases throughout the program (Table 3) (14). All dynamic exercises were executed through a full range of motion until reaching momentary concentric muscle failure, defined as the point at which subjects, despite maximum effort, are unable to complete the concentric portion of their current repetition or maintain proper exercise technique (37). The cadence for dynamic exercises was set to 2 seconds for the eccentric phase and 1 second for the concentric phase, with no isometric pauses at the top or bottom of the movement. Subjects were instructed to adjust weights using the “2-for-2 Rule” (10), which recommends increasing the load if 2 or more repetitions beyond the target range can be completed in 2 consecutive sessions. For isometric core exercises, subjects were required to maintain good form and hold positions for a minimum of 20 seconds and a maximum of 45 seconds, depending on their training level.

Table 2.

Resistance training exercises and order during the intervention period.^*

Weeks 1–6				Weeks 7–10
	Session 1	Session 2	Session 3	Session 1	Session 2	Session 3
Main lifts	Back squat (BB) Bench press (BB)	Deadlift (BB) Standing shoulder press (BB)	Back squat (BB) Bench press (BB)	Back squat (BB) Bench press (BB)	Deadlift (BB) Standing shoulder press (BB)	Back squat (BB) Bench press (BB)
Assistance exercises	Good mornings (BB) Bent-over row (BB) 1-arm bench press (DB) French press (BB)	Split squat (DB) 1-arm row (DB) 1-leg standing calf raise (DB) Zottman curl (DB)	Romanian deadlift (BB) Incline bench press (BB) Bent-over rear delt fly (DB) French press (DB)	1-leg elevated glute bridge (BW) Seesaw row (DB) Bench press (DB) 1-arm overhead triceps extensions (DB)	Side lunge (DB) Chest-supported row (DB) 1-leg standing calf raise (DB) Seated biceps curl (DB)	1-leg Romanian deadlift (DB) Alternating incline bench press (DB) YTWs (DB) Triceps kickbacks (DB)
Core exercises	Side plank (BW)	Copenhagen planks (BW)	Plank (BW)	Side plank rotation (BW)	Reverse plank (BW)	Hollow body hold (BW)

^*BB = barbell; BW = body weight; DB = dumbbell.

Table 3.

Sets, repetitions, and interset rest periods for main lifts, assistance exercises, and core exercises through the intervention period.

	Weeks 1–3	Weeks 4–6	Weeks 7–9	Week 10
Main lifts	Sets × reps 3 × 12 Interset rest 1.5–2 min	Sets × reps 3 × 8 Interset rest 1.5–2 min	Sets × reps 4 × 5 Interset rest 3 min	Sets × reps 4 × 5 Interset rest 3 min
Assistance exercises	Sets × reps 3 × 10–15 Interset rest 1–1.5 min	Sets × reps 3 × 10–15 Interset rest 1–1.5 min	Sets × reps 3 × 8–15 Interset rest 1–1.5 min	Sets × reps 2 × 8–15 Interset rest 1–1.5 min
Core exercises	Sets × time 3 × 20–45 s* Interset rest 1 min	Sets × time 3 × 20–45 s* Interset rest 1 min	Sets × time 3 × 20–45 s* Interset rest 1 min	Sets × time 2 × 20–45 s* Interset rest 1 min

^*Depending on training status.

Statistical Analyses

All statistical analyses were conducted using R (version 4.0.5) and RStudio (version 1.4.1106) software. The significance level was set at α = 0.05. The Shapiro–Wilk test was used to verify normality, while variance homogeneity was assessed visually through residual plots (19). Baseline anthropometric and training experience characteristics (age, height, body mass, training experience, and S-SRQ) across the 3 groups (SUP, APP, PDF) were compared using 1-way ANOVAs. For S-SRQ data, a z-transformation was applied before the ANOVA procedure. To analyze each outcome variable (bench press 1RM, squat 1RM, fat mass, fat-free mass, muscle mass, body mass, and WHO-5), a 3 (group: SUP vs. APP vs. PDF) × 2 (time: pre vs. post) repeated measures ANCOVA (rANCOVA) was conducted, with preintervention values included as covariates to control for baseline differences (22). Repeated-measures ANCOVAs were performed using the aov() function in R, which treats repeated measurements as fixed effects within a general linear model, without explicitly modeling the within-subject covariance structure. Given the pre–post design with only 2 measurement occasions, this approach is considered robust and does not require assumptions of sphericity or more complex covariance. Significant main effects and interactions were followed by Bonferroni-corrected post hoc tests to identify between-group differences. Effect sizes for the ANCOVAs were calculated as pes, interpreted as small (0.01), medium (0.06), or large (0.14) according to Cohen's guidelines (5).

In addition, standardized mean differences (SMDs) were calculated to quantify within-group pre–post effects and interpreted based on Cohen's thresholds (5) 0.2 = small, 0.5 = moderate, and 0.8 = large. In this study, SMDs were computed as the mean change divided by the SD of the change scores within each group, reflecting the magnitude of change relative to within-subject variability. This method is widely used in exercise science to evaluate the practical significance of training adaptations. Results are presented as means ± SDs.

Results

All mean pre–post data, including SD and percentage changes for all groups, are provided as an overview in Table 4.

Table 4.

Pre- and postoutcome data (mean ± SD) with percentage changes for all 3 training groups: supervised training (SUP), training guided through an online application (APP), and training plan provided in PDF format only (PDF).^*

Parameter	SUP pre	SUP post	Pre–post change	APP pre	APP post	Pre–post change	PDF pre	PDF post	Pre–post change
Body mass (kg)	81.3 ± 20.5	83.1 ± 21.4	+2.1 ± 2.1%	76.1 ± 18.2	76.4 ± 17.8	+0.7 ± 2.7%	76.7 ± 14.6	76.8 ± 14.7	+0.2 ± 2.8%
Fat mass (kg)	20.7 ± 10	21 ± 10.1	+2.5 ± 1.6%	17.9 ± 7.4	18.7 ± 7.3	+5.5 ± 9.6%	18.2 ± 7.9	18 ± 7	+3.6 ± 20.9%
Fat-free mass (kg)	60.7 ± 14.3	62.1 ± 14.5	+2.4 ± 1.6%	58.2 ± 13.5	57.8 ± 13.1	−0.6 ± 3.0%	58.8 ± 12.7	58.9 ± 12.5	+0.2 ± 4.0%
Muscle mass (kg)	57.7 ± 13.6	59 ± 13.8	+2.5 ± 1.6%	37.2 ± 9.7	37.7 ± 9.4	+1.7 ± 4.8%	38.3 ± 9.4	39.2 ± 9.4	+2.5 ± 3.5%
Squat 1RM (kg)	86.7 ± 36.7	113.3 ± 38.3	+35.9 ± 16.7%	88.6 ± 29	107.7 ± 32.3	+24.0 ± 17.3%	95.5 ± 32.5	114.9 ± 34.9	+22.3 ± 14.7%
Bench press 1RM (kg)	62.2 ± 31.9	71.3 ± 34	+17.7 ± 8.9%	56.9 ± 25.5	65.1 ± 26.3	+17.8 ± 13.9%	64.5 ± 29.1	72.2 ± 30	+14.4 ± 11.3%
WHO-5 (%)	55.7 ± 17.2	71.5 ± 13	+39.2 ± 46.3%	68.3 ± 13	73.5 ± 13.5	+9.9 ± 22.6%	67.2 ± 15.6	76.2 ± 12.8	+23.5 ± 55.4%

^*1RM = 1-repetition maximum; WHO-5 = 5-item World Health Organization Well-Being Index.

Body Composition

The group × time rANCOVA revealed significant interaction and group effects for BM adaptations (Figure 1A). Post hoc tests indicated a significant increase in BM from pre- to postintervention for the SUP group (p = 0.006, SMD = 0.95). In addition, the SUP group demonstrated significantly greater BM increases compared with the PDF group (p = 0.004, SMD = 0.85). Regarding FM (Figure 1B), no significant interaction effects were observed; however, the ANCOVA revealed significant main effects for both time (p = 0.02) and group (p = 0.03). Post hoc analyses did not identify statistically significant differences between specific groups. Notably, a significant pre-to-post increase in FM was observed in the APP group (p = 0.004, SMD = 0.46). These findings indicate overall group-level differences in FM and suggest a modest FM increase specifically in the app-based training group. In contrast, the group × time rANCOVA revealed significant interaction and group effects for FFM adaptations (Figure 1C). Post hoc tests showed significant pre-to-post increases in FFM for the SUP group (p ≤ 0.001, SMD = 1.52). In addition, the SUP group demonstrated significantly greater FFM increases than the APP group (p ≤ 0.001, SMD = 1.38) and the PDF group (p = 0.021, SMD = 0.81). Regarding MM (Figure 1D), no significant interaction effects were observed, but significant time and group effects were detected. Post hoc tests revealed significant pre-to-post MM increases for the SUP group (p = 0.009, SMD = 1.52) and the PDF group (p = 0.047, SMD = 0.57). In addition, the SUP group demonstrated significantly greater MM increases compared with the APP group (p = 0.047, SMD = 0.64). Across all body composition models (BM, FM, FFM, and MM), baseline values were included as covariates and showed statistically significant effects (p < 0.001), highlighting the importance of initial body composition for postintervention changes.

Figure 1.

Body mass (A), fat mass (B), fat-free mass (C), and muscle mass (D) data (mean ± SD) for all 3 training groups—supervised training (SUP), training guided through an online application (APP), and training plan provided in PDF format only (PDF)—are represented in green, blue, and orange, respectively. Individual preintervention (circles) and postintervention (triangles) data points are shown. Interaction, time, and group effects from the group × time rANCOVA are included, with significance levels (p-values) and effect sizes (partial eta squared, pes) presented. Post hoc significance is indicated as ***, **, and * for p ≤ 0.001, p ≤ 0.01, and p ≤ 0.05, respectively. Mean differences (MD) and standardized mean differences (SMD) are provided for pairwise comparisons.

Maximum Strength

The group × time rANCOVA revealed significant interaction, time, and group effects for squat 1RM adaptations (Figure 2A). Post hoc tests showed significant pre-to-post changes for SUP (p ≤ 0.001, SMD = 4.08), APP (p ≤ 0.001, SMD = 1.74), and PDF (p ≤ 0.001, SMD = 1.66). Furthermore, the SUP group demonstrated significantly greater squat 1RM improvements than the APP group (p = 0.018, SMD = 0.82) and the PDF group (p = 0.044, SMD = 0.79). Regarding bench press 1RM, no significant interaction or group effects were observed, but significant time effects were detected (Figure 2B). Post hoc tests revealed significant pre-to-post changes for all groups. In all strength-related rANCOVA models (bench press 1RM and squat 1RM), the baseline values were included as covariates and were found to be highly significant (p < 0.001), indicating a strong influence of initial performance level on the postintervention outcomes.

Figure 2.

Squat (A) and bench press (B) 1RM data (mean ± SD) for all 3 training groups—supervised training (SUP), training guided through an online application (APP), and training plan provided in PDF format only (PDF)—are shown in green, blue, and orange, respectively. Individual pre- (circles) and post- (triangles) data points are presented. Furthermore, interaction, time, and group effects of the group × time rANCOVA are reported, along with significance values (p) and effect sizes (partial eta squared [pes]). Post hoc significance is indicated as ***, **, and * for p ≤ 0.001, p ≤ 0.01 and p ≤ 0.05, respectively. Mean differences (MD) and standardized mean differences (SMD) are provided for pairwise comparisons.

Subjective Well-Being

Regarding the WHO-5 data, the group × time rANCOVA revealed no significant interaction but significant time and group effects (Figure 3). Subsequent post hoc tests revealed significant pre-to-post improvement for the SUP group (p ≤ 0.001, SMD = 0.97) and the PDF group (p = 0.032, SMD = 0.44). In the analysis of psychological well-being (WHO-5), the preintervention score was also a highly significant covariate (p < 0.001), suggesting that baseline well-being substantially influenced the observed postintervention values.

Figure 3.

5-item World Health Organization well-being index (WHO-5) data (mean ± SD) for all 3 training groups—supervised training (SUP), training guided through an online application (APP), and training plan provided in PDF format only (PDF) are given in green, blue, and orange, respectively. Thereby, individual pre- (circles) and post- (triangles) data are presented. Furthermore, interaction, time, and group effects of the group × time rANCOVA are given. Thereby, significances (p) and effect sizes (partial eta squared [pes]) are presented. Post hoc is indicated as ***, **, and * for p ≤ 0.001, p ≤ 0.01 and p ≤ 0.05, respectively. Mean differences (MD) and standardized mean differences (SMD) are given for pairwise comparison.

Short Supervisory Relationship Questionnaire

The S-SRQ data revealed significantly higher values for the PT group (96.7 ± 4.3%) than for the APP group (92 ± 7.1%) (p = 0.005, SMD = 0.74).

Discussion

This study aimed to investigate the effects of 3 different approaches to resistance training—fully supervised (SUP), app-guided (APP), and self-directed (PDF)—on body composition, maximal strength, subjective well-being, adherence, and supervision satisfaction in trained adults. The findings support the hypothesis that SUP is the most effective training modality for most of the assessed parameters. Specifically, SUP resulted in significantly greater increases in FFM, maximal strength in complex exercises (e.g., squat), and subjective well-being than APP and PDF. In addition, adherence rates were notably higher in SUP than in APP and PDF, with subjects in the SUP group also reporting significantly greater satisfaction with their supervision experience. Although APP and PDF training also induced positive adaptations, their effects were generally smaller, highlighting the unique advantages of direct supervision. These results emphasize the critical role of personal coaching in optimizing RT outcomes and provide valuable insights for refining training strategies.

Consequently, our findings underscore the superiority of supervised RT in promoting FFM compared with app-guided and self-directed training, where no significant changes were observed. These results align with prior studies, such as those by Stefanov et al. (38), who investigated a combined aerobic and resistance exercise program in sedentary men and women, and Storer et al. (39), who examined RT in already training men aged 30–44 years. Both studies reported significant FFM and lean body mass gains in supervised interventions, while unsupervised training showed no measurable improvements. Moreover, we found that BM increased exclusively in the SUP group, likely because of the gains in FFM. This finding is consistent with previous research, including the above-mentioned study by Storer et al. (39), which observed similar BM increases with supervised training. The enhanced outcomes in the SUP group likely result from continuous feedback and technical guidance by coaches, facilitating optimal muscle activation and progressive overload. In addition, the motivational presence of supervisors encourages higher training intensities (6,25), which are critical for hypertrophy (8,33) and, consequently, for changes in BM and FFM. This is supported by the observation that MM significantly increased in the SUP group.

In contrast, MM also significantly increased in the PDF group, which seems contradictory to the previously discussed findings, where BM and FFM did not change in the PDF group. This discrepancy warrants further discussion. One possible explanation is that BIA may be less valid for measuring skeletal muscle mass accretion compared with other methods such as dual-energy X-ray absorptiometry (DEXA), magnetic resonance imaging (MRI), or ultrasound measurements (1D, 2D, or 3D) (16). Consequently, BIA-based MM measurements in the PDF group may have been influenced by factors such as hydration status (31) or meal timing (18). Owing to logistical constraints, most subjects were unable to complete the BIA measurement in the morning on an empty stomach, as recommended by Kyle et al. (20). Although subjects were provided with specific instructions regarding food and fluid intake before the measurement, variations in compliance could have affected the results. Future studies should account for these shortcomings, and the present findings concerning MM should be interpreted with caution.

No significant between-group differences were observed for FM, with only a time effect being noted. This may reflect the absence of a caloric-restricted diet, the lack of aerobic training as in Meng et al. in older adults (27), and the relatively short 10-week intervention—which may have limited FM reductions in trained individuals. Interestingly, despite regular RT, FM increased slightly across all groups, particularly in the APP and PDF groups. This finding underscores that without caloric restriction or nutritional control, RT alone may not lead FM reductions, because the energy expenditure from RT was likely insufficient to induce a caloric deficit given that subjects maintained their normal diet. These results align with previous evidence suggesting that exercise interventions targeting body composition are most effective when combined with dietary strategies (34). Moreover, the compensatory eating phenomenon after exercise may have unintentionally increased caloric intake, thereby contributing to fat mass gain (17,29).

Significant group differences were observed in squat 1RM, with the SUP group showing the greatest improvements compared with the APP and PDF groups. This aligns with prior research emphasizing the advantages of supervision in complex, multijoint exercises such as the squat (6,25). Effective squat performance requires precise coordination of multiple joints and muscle groups (32), making technical feedback essential to optimize execution, maximize force production, and minimize injury risk (14,25). The presence of coaches in the SUP group likely facilitated refined movement patterns and progressive load adjustments, enabling greater neuromuscular adaptations.

In contrast, no significant group differences were found for bench press 1RM, because all groups showed similar time-dependent improvements. This is consistent with previous studies in trained individuals (6,39), suggesting that simple exercises such as the bench press require less external supervision. For example, Coleman et al. (6) reported no significant differences in bench press strength gains between supervised and unsupervised groups in young resistance-trained males and females, as both groups benefited equally from consistent training stimuli. In this study, trained subjects in the APP and PDF groups likely relied on their prior training experience and familiarity with the movement to achieve comparable strength gains even in the absence of direct feedback.

The differences between squat and bench press outcomes highlight the benefits of supervision in exercises that require greater coordination, stabilization, and technical precision. Supervision likely enhances neuromuscular coordination and performance, thereby supporting long-term adaptations. These findings emphasize the importance of tailoring supervision levels to the specific demands of exercises and the experience levels of trainees to optimize strength adaptations. For example, although the APP group showed strength improvements to a lesser extent, app-supported resistance training (RT) could serve as a valuable tool in training regimens for trained athletes or for programs involving less complex exercises. In such cases, app-based training may help bridge the gap between fully supervised and entirely self-directed training. This approach could be particularly useful in contexts where financial or time constraints are present, or when athletes need to be coached remotely.

Regarding subjective well-being, our results indicate significant improvements in both the SUP and PDF groups, while no significant changes were observed in the APP group. The lack of improvement in the APP group may be attributed to reduced social interaction and delayed feedback, factors known to limit the psychological benefits of exercise (11,35). Brocki et al. (2) similarly found that the absence of direct interpersonal interaction diminished the mental health impact of exercise in postcancer rehabilitation. In contrast, the SUP group likely benefited from continuous positive reinforcement, accountability, and motivation provided by direct supervision—factors critical for mental health improvements (42). This aligns with Langeard et al. (22), who found similar well-being benefits in both face-to-face and telepresence-based supervision, emphasizing the importance of perceived connection and support. Surprisingly, the PDF group also showed significant well-being improvements, possibly because of the autonomy and self-determination associated with self-directed training, which may enhance intrinsic motivation (40). By contrast, APP training may lack both the emotional engagement of direct supervision and the autonomy of self-directed formats, diminishing its impact on well-being.

Our findings underscore the critical role of supervised RT in enhancing adherence, supervision satisfaction, and overall training outcomes. Adherence rates were highest in the SUP group, consistent with evidence suggesting that direct supervision supports accountability and consistency (11,35) In contrast, illness-related dropouts were more frequent in the PDF (n = 12) and APP (n = 7) groups than in the SUP group (n = 1). Similarly, nonstudy-related injuries were most prevalent in the PDF group (n = 4), followed by the SUP group (n = 1), with none reported in the APP group. Although these high injury and illness rates in the PDF group may partly be a matter of chance, it raises speculation as to whether they were used as socially acceptable excuses for discontinuing participation, particularly in unsupervised settings where accountability is lower (11,35). These findings align with the “honest” quitting phenomenon, where the SUP group had no dropouts because of personal reasons (n = 0), while the APP (n = 2) and PDF (n = 6) groups reported more discontinuations attributed to factors such as lack of time or family obligations. This suggests that the higher accountability inherent in the SUP group may provide stronger motivation to persist. Future RT supervision studies should consider implementing anonymous follow-up surveys to better understand subjects' true reasons for dropout.

Subjects also reported significantly greater supervision satisfaction in the SUP group, likely because of personalized guidance and immediate feedback—factors consistently linked to better adherence and performance (6,25). Meta-analyses, such as that by Lacroix et al. (21), confirm that supervised training outperforms unsupervised approaches in strength and hypertrophy, particularly when frequent and individualized feedback is provided. Mazzetti et al. (25) demonstrated that supervised subjects used heavier loads and achieved better technical execution, resulting in superior strength gains. Similarly, Mann et al. (24) found that significant improvements in body composition and strength occurred only in supervised groups. These benefits are most pronounced in longer interventions, because shorter programs often show similar efficacy between supervised and unsupervised training (3). In addition, lower coach-to-subject ratios yield stronger effects, as noted by Gentil and Bottaro (11), while telepresence supervision can offer similar benefits to face-to-face models when perceived connection and real-time feedback are maintained (22). However, the higher satisfaction reported in the SUP group emphasizes the added value of in-person interactions for motivation and psychological support.

This study has several limitations that should be considered for future research. First, there was an age difference between the SUP group and the other groups. Although this difference did not seem to have an impact on the results, it should be considered when interpreting the findings. Second, we did not perform test–retest reliability measures for our outcomes. Therefore, we are unable to report measurement error or calculate ICCs for the assessments used. However, the published literature demonstrates good-to-excellent reliability for the instruments used. The absence of direct test–retest reliability data in this study should be considered when interpreting the results. Third, the lack of dietary control likely influenced FM results, and a longer intervention period may be necessary to observe significant FM and strength adaptations (25,38). Another limitation of this study is the lack of volume load data, as the training app used did not have an export function to externally analyze training volume between groups. These data could have a better understanding of the relationship between supervision, training intensity, and RT effectiveness. Future studies should include monitoring of volume load for more comprehensive analysis. In addition, future studies should consider other strength outcomes, such as reactive strength, power, and strength endurance. Using BIA instead of more advanced methods, such as MRI or DEXA, limited measurement precision. Future studies should consider using more accurate tools for body composition assessment and hypertrophy. The exclusion of subjects with adherence below 85% may have biased the results toward more motivated individuals, highlighting the need for further exploration of adherence strategies in less motivated populations. Furthermore, the APP and PDF groups lacked real-time feedback, suggesting the potential value of hybrid models that combine app-based training with periodic direct supervision (22).

Practical Applications

This study highlights the clear advantages of SUP RT in enhancing FFM, maximal strength, subjective well-being, adherence, and supervision satisfaction compared with app-guided (APP) and self-directed (PDF) training. Supervised showed superior outcomes in complex, multijoint exercises such as the squat, which require precise technical guidance, whereas APP and PDF were effective but produced less significant overall results. These findings underscore the importance of direct supervision, particularly for complex exercises and individuals who need higher accountability. They also highlight the potential of app-based training as a scalable, hybrid solution when combined with periodic in-person or telepresence support (22,25). For practitioners, adjusting the level of supervision based on exercise complexity and individual needs can optimize training outcomes. Coaches should focus on providing feedback and fostering motivation in supervised settings while using app-based tools to enhance adherence and scalability. These results provide valuable insights into refining training strategies to meet both individual and contextual demands.