A Study on the Effectiveness of Tool Box Meeting Educational Materials Based on Information Quantity

Abstract

This study analyzed the effects of various educational materials used in Tool Box Meetings conducted prior to work at construction sites on educational effectiveness. Specifically, the study examined the impact of changes in information quantity, linguistic explanation, and the number of educational materials on the cognitive load of construction workers. The study involved 345 construction workers. Group A utilized visual materials with higher information quantity compared to Group B. Group B, in turn, used visual materials that simplified the information to match linguistic explanations provided for Group A’s materials. Group C conducted the education meeting by reducing the number of educational materials from 13 to 8 after using Group B’s materials. Cognitive load, based on recall counts and recall rates, was then analyzed. In Group A, the use of visual educational materials with high information quantity was associated with reduced learning effectiveness, likely due to increased cognitive load. Meanwhile, in Group B, using educational materials that simplified information to match linguistic explanations resulted in an increase in recall counts and recall rates. In Group C, reducing the number of educational materials resulted in no difference in recall counts compared to Group B; however, there was an increase in the overall recall rate. Based on these research findings, it was concluded that utilizing visually simplified materials aligned with linguistic explanations and considering the cognitive load of workers to establish an appropriate number of educational materials are effective approaches in Tool Box Meeting education.

1. Introduction

Figure 1 presents a graphical representation of industrial accident statistics in South Korea from 2012 to 2021. During this period, the fatality rate in the construction industry ranged from 2.01‱ to 2.32‱ per 10,000 workers, consistently higher than the overall industry average of 1.07‱ to 1.47‱ [1]. As of 2021, the fatality rate in construction was 2.17 times higher than the national industrial average and 1.8 times higher than that of the manufacturing sector. That year, a total of 122,713 workers were injured in industrial accidents, with 2080 fatalities, of which 551 cases (26.5%) occurred in the construction industry (

Table 1. Accident occurrence by industry in 2021. (Unit: persons).

Industry	Number of Employees	Number of Work Injury Cases	Number of Fatalities
Construction industry	2,378,751	29,943	551
Manufacturing industry	3,959,780	31,709	512
Mining industry	10,257	3336	349
Other industries	12,957,777	57,725	668
Total	19,306,565	122,713	2080

).

Safety training has become more important in the construction industry to prevent accidents. However, challenges, like short duration trainings for temporary workers, one-time sessions, and difficulty in assessing work environments, persist. To address these, South Korean construction companies have implemented on-site Tool Box Meetings (TBM) to provide training.

At construction sites, TBM typically lasts 10 to 15 min, with about 5 min allocated for safety and health training. Despite the short duration, the training often includes a wide range of information, which can lead to cognitive overload. Cognitive load refers to the total amount of mental effort required to perform a cognitive task. It interacts with the limited capacity of working memory, such that excessive cognitive load can overwhelm working memory and reduce learning effectiveness in processing and storing information simultaneously [2]. When the amount of information exceeds this capacity, it can either fail to be processed effectively or cause overload [3]. Cognitive overload occurs when the cognitive resources required to perform a task exceed the capacity of the cognitive structure [4]. According to cognitive load theory, working memory consists of multiple processing channels; information from sensory registers, such as visual and auditory inputs, is temporarily stored in working memory before being transferred to long-term memory [5].

This theory has been expanded through research in multimedia learning. According to Mayer (2001), learners understand and retain information more effectively when verbal and visual materials are presented together in a coherent and well-aligned format [6]. Sweller et al. (2011) emphasized that reducing extraneous cognitive load—by simplifying visuals and removing unnecessary information—enhances working memory performance and learning efficiency [7]. Moreno and Mayer (2007) further suggested that segmenting complex information and aligning visuals with corresponding verbal explanations help to reduce overload and promote effective learning [8]. These principles are particularly relevant in TBM training, which requires the efficient delivery of safety content in a limited time.

Cognitive load can be divided into three types: germane, extraneous, and intrinsic. Germane cognitive load is beneficial for learning, promoting content organization, schema formation, and automatic processing. It can enhance motivation and engagement. Extraneous cognitive load arises from irrelevant information or distractions that do not contribute to learning objectives, negatively affecting learning outcomes. Intrinsic cognitive load is related to the complexity of the task or content, which depends on the connections between elements and the learner’s prior knowledge or expertise [9,10].

Memory is the cognitive process of encoding, storing, and retrieving information, which is fundamental to learning and behavior [11]. According to the multi-store memory model, memory processing occurs in three stages: sensory memory, short-term memory, and long-term memory. Information briefly resides in sensory memory, is transferred to short-term memory, and, through rehearsal, moves into long-term storage [12]. Ebbinghaus’ forgetting curve highlights that memory decay is most rapid shortly after learning: approximately 58% of information is retained after 19 min; only 21% remains after a month [13].

Based on these insights, this study aimed to explore effective TBM training methods for construction workers, focusing on cognitive load and the effects of varying visual complexity and the number of training items. To do so, three versions of training materials were developed: A-type included 13 visually dense items, B-type presented the same items with simplified visuals and aligned verbal cues, and C-type featured 8 essential items derived from the B-type set (

Table 2. Summary of instructional styles and material characteristics for Groups A, B, and C.

Group	Number of Items	Visual Complexity	Explanation Style
A	13	High	Visual-focused materials with dense information
B	13	Medium (Simplified from A)	Simplified visuals matched with verbal guidance
C	8	Medium (Same as B)	Reduced number of items for focused delivery

).

Therefore, the aim of this study is to examine how the complexity of visual information and the number of educational items used in Tool Box Meeting (TBM) materials affect the cognitive load and memory retention of construction workers. Based on this analysis, the study seeks to suggest strategies to enhance the effectiveness of on-site safety training.

To address this aim, recall experiments were conducted to investigate the effects of visual complexity and the number of training items on learning performance and cognitive load.

Accordingly, the study is guided by the following research questions:

RQ1. 

Does the visual complexity of TBM training materials affect the recall performance of construction workers?

RQ2. 

Does reducing the number of TBM training items improve workers’ memory retention?

The following hypotheses were developed:

H1. 

TBM materials with high visual complexity will result in lower recall performance than simplified materials;

H2. 

A reduced number of TBM training items will increase the recall rate of construction workers.

2. Worker Safety Training

2.1. Legal Safety Training

Safety training is a proactive approach aimed at reducing the risks of accidents and maintaining a healthy lifestyle by addressing safety threats. The goal is to implement measures to minimize the impact of potential accidents and prevent injuries, fatalities, or property damage [14]. Effective safety training is essential for establishing a safe construction site and managing safety programs. It is a core component of construction safety strategies [15], with trained workers being more effective at preventing accidents [16].

In South Korea, legal safety training includes regular training, hiring training, training for job changes, special training, and basic safety and health training. For temporary workers, the regulations require at least 1 h of training at hiring or when tasks change, and 2 h for special training. Full-time workers must complete at least 8 h of training at hiring, 2 h for task changes, and 16 h for special training (

Table 3. Training duration and frequency based on the safety education curriculum.

Curriculum	Target Group		Duration	Frequency
Regular training	Office workers		More than 3 h	Per quarter
	Workers other than office workers	Workers directly engaged in sales	More than 3 h	Per quarter
		Workers other than those directly engaged in sales	More than 6 h	Per quarter
	Manager		More than 16 h	Per year
Recruitment training	Temporary workers		More than 1 h	Per occurrence
	Permanent worker		More than 8 h	Per occurrence
Training for changes in work tasks	Temporary workers		More than 1 h	Per occurrence
	Permanent worker		More than 2 h	Per occurrence
Special training	Temporary workers engaged in specified tasks		More than 2 h	Per occurrence
	Temporary workers engaged in tower crane signaling work		More than 8 h	Per occurrence
	Workers, excluding temporary workers, engaged in specified tasks		More than 16 h For short-term or intermittent tasks, at least 2 h.	Per occurrence
Basic safety and health training for the construction industry	Temporary workers		More than 4 h	Only once

) [17,18,19].

Most construction workers are employed on a wage or salary basis; however, daily workers or those with short-term contracts receive significantly shorter training durations. This lack of awareness about work-related risks increases the likelihood of accidents among temporary workers, who make up the majority of construction site staff. As shown in Table 3, temporary workers have fewer and shorter training sessions.

2.2. TBM

As construction progresses, reinforcing safety education for daily workers becomes challenging due to emerging hazards. To address this, many construction companies conduct TBM before work each day. TBM is a brief meeting, often with a tool box, held before work or during shift changes, and is known by various names such as Take 5 Talk, Pre-Start Safety, and OSHA Tool Box Talk [20].

Floyd [21] described TBM as an informal training session led by skilled individuals, lasting 10 to 15 min. Kaskutas [22] stated that TBM improves communication, reduces injuries, and enhances safety. Harrington [23] noted that TBM is vital for raising workers’ risk awareness and promoting safe habits. TBM improves safety by increasing awareness, participation, and preparedness, while ensuring compliance with safety regulations, ultimately enhancing construction site safety.

In South Korea, TBM are not legally mandated by the Industrial Safety and Health Act; there are no standardized procedures or methods explicitly outlined for their implementation. However, construction companies have established their own procedures for conducting TBM in order to reduce accidents.

Table 4. Example of TBM procedures.

Step		Content
1	Gathering	Meeting at the work site by work team (up to 15 people)
2	Greetings and stretching	Mutual salutation, checking physical condition, warming up, etc
3	Personal protective equipment (PPE) check	Check the condition of personal protective equipment
4	Emergency Evacuation procedures	Check emergency evacuation routes, emergency contact network, and AED location
5	Hazard prevention Training	Communicate work details, hazards, and safety measures
6	Hazardous work demonstration	Explain hazards or demonstrate safe work methods
7	Hazard identification	Clearly specify and identify hazardous factors
8	Notification delivery	All notices, conflicts with other tasks, and other necessary items

presents an example of the procedures for TBM implementation in a domestic construction company.

Of the 8 steps, steps 5–7 correspond to safety education. TBM is conducted for approximately 10 to 15 min, with around 5 min dedicated to safety education, using paper and tablets. The safety education is composed of work stoppage, accident cases, and safety standards, and includes visual items that convey a large amount of information. Examples of TBM training items are shown in Figure 2, Figure 3 and Figure 4.

According to cognitive load theory, the effectiveness of training is heavily influenced by the way information is structured and delivered. In the context of TBM safety education, where time is limited and workers are exposed to a high volume of visual materials in a short period, managing cognitive load becomes especially critical. When educational content exceeds the working memory capacity of construction workers, it can result in cognitive overload, which may impair learning and recall.

Therefore, applying cognitive load theory to safety education means structuring materials in a way that minimizes extraneous load, manages the intrinsic load according to worker experience, and enhances germane load to support meaningful learning. By simplifying visual content, aligning it with verbal explanations, and reducing the number of presented items, educators can improve comprehension and memory retention among workers, particularly those with limited experience or lower technical expertise. This approach is essential for increasing the effectiveness of brief, high-impact sessions like TBM.

3. Materials and Methods

3.1. Experimental Design

In the current domestic regulatory environment, there is no systematic safety and health education system for daily workers. To address this, TBM sessions have become an important tool; however, they are brief, typically lasting 10 to 15 min, with only 5 min dedicated to safety education, which may reduce effectiveness. This study aims to identify effective TBM components by analyzing cognitive load in daily workers. Recall experiments will examine the impact of visual information complexity and the number of educational items.

Three recall tests were conducted with different educational items: A-type, which included items with high visual information complexity; B-type, which contained items with less visual information complexity than A-type; and C-type, which consisted of 8 items, reducing the 13 items used in B-type (Figure 5, Figure 6 and Figure 7).

The instructional materials used in each group differed in both the amount of visual information and the alignment between images and verbal explanations.

A-type materials displayed 13 safety items on a single slide, with densely packed icons and small text. The visuals were complex; the text was not always aligned clearly with the images.

B-type materials also used 13 items; however, the layout was simplified. Icons were spaced out more clearly and the verbal explanations were positioned next to each corresponding image.

C-type materials included only 8 items, reducing visual overload. Each item had a larger icon and matching verbal explanation, which made it easier to focus on and understand.

The study, conducted from 3 November to 4 November 2022, involved 345 workers at a construction site in Pyeongtaek, analyzing the correlation between recall rates and the number of items presented.

3.2. Participants

As of 2021, there were 1.38 million workers in the construction industry, with 513,280 daily workers [1]. A sample of 345 daily workers from this total, with a 95% confidence level and a 5% margin of error, was used for the experiment. Each of the three groups had 115 workers; they were assigned based on the educational items received. Recall counts and rates were tested; the results are shown in

Table 5. Differences between groups A, B, and C.

			Group of People			Total	χ2
			A	B	C		(p)
Nationality	Local	Persons	115	112	113	340	0.000 (1.000)
		%	100	100	100	100
	Foreigner	Persons	0	0	0	0
		%	0	0	0	0
Gender	Male	Persons	107	104	102	313	0.302 (0.860)
		%	93.0	92.0	91.1	92.1
	Female	Persons	8	9	10	27
		%	7.0	8.0	8.9	7.9
Age	Twenties	Persons	21	25	22	68	4.724 (0.803)
		%	18.4	22.1	19.5	20.0
	Thirties	Persons	33	25	23	81
		%	28.9	22.1	20.4	23.8
	Forties	Persons	25	25	24	74
		%	21.9	22.1	21.2	21.8
	Fifties	Persons	34	38	43	115
		%	29.8	33.6	38.1	33.8
	60 and above	Persons	1	0	1	2
		%	0.9	0.0	0.9	0.6
Construction experience	Less than one year	Persons	28	35	43	106	15.534 (0.114)
		%	24.6	31.0	38.1	31.2
	Less than 1 to 2 years	Persons	18	16	9	43
		%	15.8	14.2	8.0	12.6
	Less than 2 to 3 years	Persons	8	9	11	28
		%	7.0	8.0	9.7	8.2
	Less than 3 to 5 years	Persons	12	18	20	50
		%	10.5	15.9	17.7	14.7
	Less than 5 to 10 years	Persons	26	24	16	66
		%	22.8	21.2	14.2	19.4
	More than 10 years	Persons	22	11	14	47
		%	19.3	9.7	12.4	13.8
Accident experience	Indirect experience	Persons	43	51	45	139	5.008 (0.286)
		%	37.7	45.1	39.8	40.9
	Direct experience	Persons	9	4	3	16
		%	7.9	3.5	2.7	4.7
	No experience	Persons	62	58	65	185
		%	54.4	51.3	57.5	54.4

.

To analyze differences in recall performance across groups, a one-way ANOVA was conducted, followed by Scheffé’s post hoc test to identify specific group differences. Significance was determined at p < 0.05. For key outcome variables (recall count and recall rate), 95% confidence intervals were calculated to quantify the precision of group means.

Chi-squared analysis, along with Fisher’s exact test, were used to examine demographic differences. The tests showed no significant differences in nationality (χ² = 0.000, p = 1.000), gender (χ² = 0.302, p = 0.860), age (χ² = 4.724, p = 0.803), construction experience (χ² = 15.534, p = 0.114), or accident experience (χ² = 5.008, p = 0.286).

A priori power analysis was conducted using G*Power 3.1 to determine the required sample size for ANOVA with the three groups. Assuming a medium effect size (f = 0.25), alpha = 0.05, and power = 0.80, the minimum required sample size was calculated to be 159 participants (53 per group). The actual sample size of 345 participants (115 per group) exceeded this requirement, ensuring adequate statistical power for the analyses.

3.3. Materials and Procedure

The educational items used in this experiment were categorized into three types based on the visual complexity of the information and quantity commonly used in current TBM. The items were as follows: A-type educational items, which had a high level of visual complexity, consisted of 2 work stoppage cases, 4 accident cases, and 7 safety standards; B-type items, which had a lower visual complexity compared to A-type, included the same number of 2 work stoppage cases, 4 accident cases, and 7 safety standards; and C-type items, which were a reduced version of the B-type, consisting of 1 work stoppage case, 2 accident cases, and 5 safety standards.

The educational items for A-type used in this experiment were sourced from items provided by the construction company in the Pyeongtaek area, to which the participants belonged. These items excluded notices and other items, and instead focused on work stoppage cases, accident cases, and safety standards. Detailed examples of these items are shown in Figure 5, Figure 6 and Figure 7. After the training, which used each of the three types of educational items, a recall test was conducted to measure the recall frequency and recall rate.

Each type of educational item was presented for 5 min and 35 s (A and B-type) or 3 min and 34 s (C-type), with consistent content across all types, delivered in a multimedia format with the same recorded audio. TBM was conducted at 7:00 AM.; the recall test was given 40 to 50 min later. Workers were informed that the study aimed to assess recall effectiveness and were instructed to mark “no recollection” if they couldn’t remember.

For recall evaluation, workers were considered to have successfully recalled items if they explained both the work stoppage case and accident cause. They were asked to write down all details they could recall for the work stoppage and accident cases, and provide correct answers for safety standards, with questions prompting answers not covered in the training.

4. Recall Test Results

Prior to conducting the ANOVA, assumptions of normality and homogeneity of variance were tested. Normality was assessed using the Shapiro–Wilk test, which showed no significant deviation from normality across groups (p > 0.05). Homogeneity of the variances was verified using Levene’s test, which confirmed equal variances across groups (p > 0.05), validating the use of one-way ANOVA.

The recall count and rate for different educational item types are summarized in

Table 6. Post-verification and variance analysis.

Variable	Type	Average	Standard Deviation	F	p	Post-Verification
Recall count	A	3.02	2.517	10.441	0.000 **	B,C > A
	B	4.46	2.623
	C	4.14	2.064
	Total	3.87	2.485
Recall rate	A	0.23	0.191	61.884	0.000 **	C > B > A
	B	0.34	0.202
	C	0.52	0.258
	Total	0.35	0.257

** p < 0.01.

. ANOVA was used to examine differences; the Scheffe test was applied for post-verification. The recall count was 3.02 for A-type, 4.46 for B-type, and 4.14 for C-type, with a statistically significant difference (F = 10.441, p < 0.001, η² = 0.057), indicating a medium effect size. For recall rates, the recall rate was 0.23 for A-type, 0.34 for B-type, and 0.52 for C-type, with a statistically significant difference (F = 61.884, p < 0.001, η² = 0.27), indicating a large effect size. The F-value for the recall rate was 61.884, indicating a significant difference.

4.1. Impact of Visual Information Complexity

As shown in Figure 8 and Figure 9, B-type had a higher recall count than A-type, except for item 6, which had a 1.5 times higher recall count and rate. These results suggest that highly complex visual information increases cognitive load, reducing educational effectiveness. Simplifying visual information rather than overwhelming workers with excessive details would improve TBM effectiveness.

4.2. Impact of Number of Educational Items

As shown in Table 5, the average recall count for B-type (4.46) and C-type (4.16) was higher for B-type; however, there was no statistically significant difference. However, the recall rate for B-type was 34%, while C-type had a recall rate of 52%, showing a 1.5 times improvement. C-type, which consisted of eight randomly selected items from B-type, demonstrated higher recall counts for all items (Figure 10 and Figure 11). This suggests that reducing the number of educational items to eight or fewer improves the effectiveness of TBM safety training.

4.3. Effect of Individual Characteristics

4.3.1. Age

ANOVA analysis was conducted to examine the differences in variables by age across the three types, followed by Scheffe post-verification. The results indicated that there were no statistically significant differences in the recall count and rate (

Table 7. The mean and standard deviation of the educational items by age.

Variable	Type	Age	Average	Standard Deviation	F	p	Post-Verification
Recall count	A	Twenties (a)	2.86	2.435	4.295	0.003 **	d > b
		Thirties (b)	1.79	1.916
		Forties (c)	3.36	2.871
		Fifties (d)	4.15	2.376
		60 and over (e)	2.00	0.000
		Total	3.04	2.520
	B	Twenties	3.56	2.022	1.569	0.201	n/a
		Thirties	4.44	2.709
		Forties	4.48	2.903
		Fifties	5.03	2.726
		Total	4.45	2.646
	C	Twenties	3.95	1.588	0.959	0.433	n/a
		Thirties	4.35	2.347
		Forties	4.42	2.185
		Fifties	3.88	1.966
		60 and over	1.00	0.000
		Total	4.08	2.032
Recall rate	A	Twenties (a)	0.18	0.173	4.758	0.001 **	d > b,e
		Thirties (b)	0.10	0.145
		Forties (c)	0.20	0.213
		Fifties (d)	0.29	0.186
		60 and over (e)	0.10	0.000
		Total	0.19	0.191
	B	Twenties	0.27	0.155	1.572	0.200	n/a
		Thirties	0.34	0.210
		Forties	0.35	0.224
		Fifties	0.39	0.211
		Total	0.34	0.204
	C	Twenties	0.50	0.199	0.955	0.436	n/a
		Thirties	0.55	0.293
		Forties	0.55	0.273
		Fifties	0.49	0.246
		60 and over	0.13	0.000
		Total	0.51	0.254

** p < 0.01. Means sharing the same letter are not significantly different at p < 0.05 (post hoc test).

).

4.3.2. Work Experience

ANOVA analysis and the Scheffe test revealed differences in recall based on work experience. As shown in

Table 8. Mean and standard deviation of educational item types based on work experience.

Variable	Type	Work Experience	Average	Standard Deviation	F	p	Post-Verification
Recall count	A	Less than one year (a)	1.86	1.900	3.936	0.003 **	d > a
		Less than 1 to 2 years (b)	3.22	2.669
		Less than 2 to 3 years (c)	2.13	2.295
		Less than 3 to 5 years (d)	4.75	2.800
		Less than 5 to 10 years (e)	2.69	2.223
		More than 10 years (f)	4.14	2.624
		Total	3.03	2.526
	B	Less than one year (a)	3.17	2.216	3.214	0.010 *	e > a
		Less than 1 to 2 years (b)	4.25	1.732
		Less than 2 to 3 years (c)	5.11	2.205
		Less than 3 to 5 years (d)	5.06	2.999
		Less than 5 to 10 years (e)	5.63	2.975
		More than 10 years (f)	4.73	2.724
		Total	4.45	2.646
	C	Less than one year	3.56	1.803	1.497	0.197	n/a
		Less than 1 to 2 years	4.67	1.658
		Less than 2 to 3 years	3.64	1.963
		Less than 3 to 5 years	4.80	2.016
		Less than 5 to 10 years	4.50	2.394
		More than 10 years	4.14	2.349
		Total	4.08	2.032
Recall rate	A	Less than one year	0.10	0.132	3.931	0.003 **	d > a
		Less than 1 to 2 years	0.21	0.204
		Less than 2 to 3 years	0.11	0.189
		Less than 3 to 5 years	0.32	0.217
		Less than 5 to 10 years	0.17	0.165
		More than 10 years	0.28	0.205
		Total	0.19	0.191
	B	Less than one year (a)	0.24	0.171	3.180	0.010 *	e > a
		Less than 1 to 2 years (b)	0.33	0.133
		Less than 2 to 3 years (c)	0.39	0.170
		Less than 3 to 5 years (d)	0.39	0.232
		Less than 5 to 10 years (e)	0.43	0.231
		More than 10 years (f)	0.36	0.210
		Total	0.34	0.204
	C	Less than one year	0.45	0.225	1.492	0.198	n/a
		Less than 1 to 2 years	0.58	0.208
		Less than 2 to 3 years	0.46	0.244
		Less than 3 to 5 years	0.60	0.251
		Less than 5 to 10 years	0.56	0.300
		More than 10 years	0.52	0.294
		Total	0.51	0.254

* p < 0.05, ** p < 0.01. Means sharing the same letter are not significantly different at p < 0.05 (post hoc test).

, in A-type, workers with 3–5 years of experience recalled more content than those with less than 1 year. In B-type, workers with 5–10 years of experience recalled more than those with less than 1 year. However, in C-type, no significant differences were found. Figure 12 and Figure 13 show that workers with less than one year of experience had lower recall counts and rates across all types, likely due to the impact of experience on learning and memory. Overall, B-type had a higher average recall rate than A-type; C-type had a higher recall rate than B-type.

4.3.3. Job Position

ANOVA analysis and post-verification using the Scheffe method showed no statistically significant differences in the recall count and overall recall rate (

Table 9. Mean and standard deviation of educational items by job position.

Variable	Type	Position	Average	Standard Deviation	F	p	Post-Verification
Recall count	A	Team leader	3.46	2.883	1.897	0.155	n/a
		Technical position	3.12	2.286
		Non-technical position	2.23	2.338
		Total	3.03	2.526
	B	Team leader	4.45	2.296	0.779	0.461	n/a
		Technical position	4.66	2.805
		Non-technical position	3.93	2.344
		Total	4.45	2.646
	C	Team leader	3.89	2.804	0.544	0.582	n/a
		Technical position	4.25	2.071
		Non-technical position	3.83	1.807
		Total	4.06	2.033
Recall rate	A	Team leader	0.22	0.216	2.022	0.137	n/a
		Technical position	0.20	0.178
		Non-technical position	0.13	0.171
		Total	0.19	0.191
	B	Team leader	0.34	0.176	0.774	0.464	n/a
		Technical position	0.36	0.217
		Non-technical position	0.30	0.181
		Total	0.34	0.204
	C	Team leader	0.49	0.350	0.548	0.580	n/a
		Technical position	0.53	0.259
		Non-technical position	0.48	0.226
		Total	0.51	0.254

), based on job title (team leaders, skilled workers, and unskilled workers). However, Figure 14 and Figure 15 reveal that non-skilled workers had lower recall counts and rates compared to skilled workers across all three types. The recall rates for B-type were higher than A-type, and C-type had higher recall rates than B-type, with the gap between skilled and non-skilled workers narrowing. This suggests that educational items with a balanced number of items and low visual complexity are particularly important for non-skilled workers.

4.3.4. Accident Experience

The ANOVA analysis and Scheffe’s method showed no statistically significant differences in recall count and rate (

Table 10. The mean and standard deviation of the educational items by accident experience.

Variable	Type	Accident Experience	Average	Standard Deviation	F	p	Post-Verification
Recall count	A	Indirect	3.37	2.582	0.964	0.384	n/a
		Direct	3.44	3.046
		No	2.73	2.410
		Total	3.03	2.526
	B	Indirect	4.92	2.741	2.394	0.096	n/a
		Direct	5.75	3.202
		No	3.95	2.460
		Total	4.45	2.646
	C	Indirect	4.31	2.076	0.615	0.543	n/a
		Direct	3.33	3.215
		No	3.95	1.964
		Total	4.08	2.032
Recall rate	A	Indirect	0.22	0.195	0.951	0.389	n/a
		Direct	0.22	0.228
		No	0.17	0.184
		Total	0.19	0.191
	B	Indirect	0.38	0.212	2.413	0.094	n/a
		Direct	0.45	0.246
		No	0.30	0.190
		Total	0.34	0.204
	C	Indirect	0.54	0.259	0.603	0.549	n/a
		Direct	0.42	0.403
		No	0.50	0.246
		Total	0.51	0.254

) based on accident experience. Workers with direct or indirect accident experience, and those with no accident-related experience, all had similar recall performance. This indicates that direct or indirect experience with accidents did not impact the effectiveness of the education. Direct experience refers to personally encountering an accident, while indirect experience refers to witnessing or hearing about an accident.

5. Discussion

This study examined how the complexity and quantity of TBM training materials affect cognitive load and recall performance among construction workers. The findings indicated that simplifying visual materials and reducing the number of items led to improved recall, particularly for less experienced and non-technical workers. These results align with cognitive load theory, which suggests that instructional design should minimize extraneous cognitive load and optimize intrinsic and germane load to support learning.

The superior recall performance in Group B and C supports previous research by Mayer (2001) and Sweller et al. (2011), emphasizing that well-aligned, simplified multimedia content can significantly enhance memory retention in high-risk, time-constrained settings [6,7]. This suggests that even short safety briefings, like TBMs, can be made more effective when designed according to evidence-based learning principles.

Moreover, the study has practical implications beyond the construction industry. Fields, such as manufacturing, logistics, and healthcare, where fast-paced environments require quick and effective safety communication, may benefit from the same principles. Adopting simplified, segmented training materials may enhance training outcomes across diverse sectors.

However, the study has limitations. Variables, such as literacy level, language comprehension, and prior exposure to training content, were not fully controlled. Future studies should explore how cognitive capacity and educational background interact with material design. Additionally, while recall performance was tested shortly after training, the long-term retention effects were not measured.

These findings provide a foundation for improving the effectiveness of TBM and similar training formats by integrating cognitive science principles into content design.

This study contributes to the theoretical development of cognitive load theory by applying it to the field of occupational safety training, specifically within the time-constrained environment of construction TBM sessions. By demonstrating how simplified, visually aligned, and limited-scope materials improve learning outcomes, the study shows how CLT can be operationalized in real-world safety education settings. This expands the applicability of CLT beyond classroom or multimedia environments into short-format, high-risk, and field-based training programs. Moreover, it offers a structured framework for developing safety training materials that reduce extraneous load, adapt to intrinsic task complexity, and enhance germane load—making safety communication more effective and cognitively efficient for diverse worker populations.

6. Conclusions

This study investigated the effectiveness of safety education conducted in during TBM in consideration of cognitive load, The results are as follows:

First, when using educational items with high visual complexity (A-type) in TBM, the effectiveness of the training was found to decrease in terms of cognitive load. The recall count for the A-type items was 3.02; the recall rate was 23.2%. In contrast, the B-type educational items, which have low visual complexity, showed a recall count of 4.46 and a recall rate of 34.3%, which is an increase of about 1.5 times compared to A-type.

Second, as a result of analyzing the cognitive load according to the number of TBM training items, there was no significant difference in the recall count between B-type items (4.46) and C-type items (4.15). However, in terms of the overall recall rate, C-type items had a significantly higher rate at 52%, compared to 34% for B-type items. This finding indicates that focusing training on 8 key items, which are considered most important for accident prevention out of the 13 items, can enhance the effectiveness of the training content for accident prevention. It also indicates that simply spreading a large number of training items does not necessarily result in better training outcomes.

Third, the analysis of individual characteristics revealed no significant differences based on age, job position, or accident experience. However, when comparing new workers with less than one year of experience to other worker groups, it was found that reducing visual complexity and limiting the number of educational items allows for more effective delivery of training content to new workers. Additionally, when reducing visual complexity and providing the appropriate number of educational items, non-technical workers benefitted more from the training than technical workers.

Fourth, the findings of this study align with the principles of cognitive load theory and multimedia learning. According to Mayer’s cognitive theory of multimedia learning [6], people learn better from words and pictures than from words alone, particularly when information is presented in a coherent, integrated format. This supports the study’s finding that simplified visual materials aligned with verbal explanations led to better recall performance.

Furthermore, previous studies have shown that reducing extraneous load and segmenting information in multimedia learning environments enhances working memory efficiency [7,8]. The current results, demonstrating the benefits of reducing visual complexity and limiting the number of educational items, are consistent with these multimedia learning principles and further validate their application in high-risk, time-limited training settings like TBM.

Finally, this study has certain limitations that should be considered. While the demographic variables, such as age, work experience, and job type, were statistically analyzed, the study did not fully account for how differences in literacy levels, prior training exposure, or language comprehension may have affected participants’ cognitive load or memory performance. Particularly in construction settings, workers may have diverse educational backgrounds, which could influence how they process visual or textual training materials. Future research should incorporate assessments of literacy and cognitive capacity to better tailor safety education to worker characteristics.

These findings may also have broader applicability beyond construction settings. Industries, such as manufacturing, logistics, and healthcare, where workers often receive brief, routine safety training under time constraints, could benefit from the same principles. Simplifying training materials, reducing cognitive load, and aligning visuals with verbal instructions can improve training effectiveness in various high-risk or fast-paced work environments. Adapting the TBM framework using cognitive load principles may support better learning and retention across diverse sectors.

In conclusion, TBM training items should utilize resources that simplify the complexity of visual information; the appropriate number of items should be selected in order of importance, considering the limits of cognitive load. When a large amount of information is delivered in a short time, the effectiveness of the training is diminished due to cognitive overload. Therefore, rather than overwhelming trainees with excessive information, it would be more effective to prioritize and deliver the most critical training items for accident prevention, reflecting the characteristics of the field. This approach is believed to contribute towards conducting effective training for accident prevention.

Furthermore, it is necessary to develop effective TBM training items that take into account the characteristics of cognitive load by varying the complexity of visual information and the number of items. Such an approach can contribute to implementing effective TBM for accident prevention.