Exploring Factors Influencing Regular Exercise Intention in High School Females: An Empirical Study Based on the Theory of Planned Behavior

Abstract

Promoting regular exercise among adolescent females is a critical public health objective. This study advances the Theory of Planned Behavior (TPB) by integrating “past behavior” as a key external variable to explore the determinants of exercise intention among high school females in Taiwan. Employing a convenience sampling method, we collected 392 valid questionnaires from senior and vocational high school students. Structural equation modeling results revealed that past behavior significantly predicted both attitude (β = .339, p < .001) and perceived behavioral control (β = .537, p < .001). In turn, attitude (β = .385, p < .001) and perceived behavioral control (β = .326, p < .001) were strong predictors of exercise intention. Notably, subjective norm did not directly influence intention (β = .066, p > .05). This study confirms the enhanced predictive power of the extended TPB model and provides a key theoretical contribution by demonstrating the pivotal mediating roles of attitude and perceived control. Practically, the findings suggest that interventions should focus on fostering positive initial exercise experiences and enhancing self-efficacy to effectively promote sustained physical activity in this demographic.

Keywords

regular exercise theory of planned behavior past behavior high school females structural equation modeling

Introduction

Regular physical activity plays a crucial role in disease prevention, supporting a healthy lifestyle, reducing medical costs, and contributing to broader economic benefits (Rodrigues & Teixeira, 2023). Higher levels of physical activity, whether in occupational or leisure contexts, along with better cardiorespiratory fitness, are associated with a reduced risk of cardiovascular disease (Verma et al., 2022). In addition to aiding weight management, regular exercise is linked to a lower incidence of chronic diseases and enhanced quality of life (Phillips & Mullan, 2023).

Adolescence is widely recognized as a critical developmental stage for establishing long-term exercise habits. Experiences during this period are strong predictors of adult physical activity behaviors (Myers et al., 2019). X. Chen et al. (2005) noted that many lifestyle patterns and chronic disease risk factors emerge during adolescence. Promoting regular exercise during this life stage can yield significant health benefits and support the development of sustainable, healthy lifestyles into adulthood (Flynn et al., 2018). In particular, female adolescents have been shown to engage in physical activity less frequently than their male counterparts (Joseph et al., 2014). Chiu (2020) reported that female students accumulate significantly fewer minutes of exercise than males across all grade levels. Studies have also observed a notable decline in physical activity participation among females as they age (Pate et al., 1995; Verma et al., 2022). For instance, only 38.9% of upper secondary students reported regular exercise participation, with females comprising an even smaller percentage (Chiu, 2020). Furthermore, Y. C. Chang et al. (2016) found that only 23.7% of female students in upper secondary education regularly engaged in extracurricular exercise. These patterns underscore the urgent need to understand and address the barriers preventing adolescent girls from maintaining regular physical activity, making this demographic not only a population at high risk but also a critical point of leverage for long-term public health interventions.

To explore the psychological drivers behind these trends, the current study adopts the Theory of Planned Behavior (TPB) as the theoretical framework (Pai & Yeh, 2017; Rodrigues et al., 2021). However, for habitual behaviors like exercise, the original TPB model can be limited. Several studies have extended TPB by including past behavior as an external variable, which has been shown to enhance the model’s explanatory and predictive power (Gholamnia-Shirvani et al., 2018; Y. C. Hsu et al., 2009). Past experiences can influence current attitudes, perceptions of control, and normative beliefs, thereby affecting both intention and actual behavior (Ajzen, 2002; Kim & Kim, 2021).

Despite this, a clear research gap persists. While the extended TPB model has been applied in various contexts, its application to the specific demographic of female adolescents within an Asian cultural setting remains largely underexplored. It is unclear whether the powerful influence of past behavior operates similarly in a population facing unique academic and social pressures.

Therefore, this study addresses the gap by integrating past behavior into the TPB framework to elucidate the psychological mechanisms underlying exercise intentions among Taiwanese senior and vocational high school female students. By extending the model, it evaluates whether habit meaningfully augments the TPB’s explanatory power and clarifies how past behavior shapes intention formation. Focusing on this specific student population, the study provides contextually grounded evidence and identifies the most influential determinants of intention, thereby informing the design of targeted, school-based health promotion strategies.

Literature Review

Regular Exercise

With the advancement of high technology and automated lifestyles, sedentary behavior has become a global concern (Lee & Yu, 2014), leading to declines in physical and mental well-being (P. Y. Yang et al., 2012). This issue is particularly pressing for adolescents, as sedentary patterns established in youth often persist into adulthood. A 2023 systematic review confirmed a significant association between increased screen time and poorer mental health outcomes, including depressive symptoms and anxiety, especially among female adolescents (Santos et al., 2023). Conversely, research by Stamatakis et al. (2009), and Yasunobe et al. (2023) has consistently shown that regular exercise is a powerful preventative measure against a wide range of chronic diseases.

Beyond disease prevention, regular exercise offers critical psychological and social benefits during the formative adolescent years. It is known to slow the aging process and reduce cancer risks (Herbert et al., 2020). For female adolescents specifically, recent studies highlight that physical activity can significantly enhance psychological resilience and improve mental health outcomes (Qiu et al., 2025). This makes early intervention essential for fostering long-term health (Gaia et al., 2021). Following the recommendations of Pate et al. (1995) for daily activity, this study adopts the operational definition of “regular exercise” provided by Lee and Yu (2014): engaging in physical activity at least three times a week for a minimum of 20 min per session at a moderate intensity. This definition provides a clear and measurable benchmark for our research.

To understand the persistence of such behavior, it is crucial to consider the role of habit. Past behavior has been identified as a powerful predictor of future actions, largely because consistent repetition in stable contexts fosters automaticity (Hagger, 2019). This is not merely an alternative to conscious decision-making; rather, it is a foundational element. A major 2024 meta-analysis by Hagger and Hamilton (2024) demonstrated that past behavior consistently adds predictive power to the Theory of Planned Behavior (TPB) across various longitudinal studies. Similarly, another recent meta-analysis focusing on young people found that habit strength is a key factor that helps bridge the gap between intending to be active and actually performing the behavior (Zhu et al., 2025). These insights suggest that direct experiences (past behavior) are critical because they shape the attitudes and perceptions of control that underpin behavioral intentions (Honkanen et al., 2005; Y. C. Hsu et al., 2009), providing a direct theoretical bridge to the TPB framework.

The Theory of Planned Behavior and the Role of Past Behavior

To understand the psychological drivers of behavior, social psychology offers robust frameworks, among which the Theory of Planned Behavior (TPB) is preeminent, particularly in health contexts. The TPB evolved from the Theory of Reasoned Action (TRA) (Fishbein & Ajzen, 1980), which posited that behavioral intentions were the direct antecedents of behavior. Recognizing that individuals do not always have complete volitional control over their actions, Ajzen (1985, 2020) incorporated perceived behavioral control (PBC), creating the TPB. This addition significantly enhanced the model’s ability to predict behaviors, like regular exercise, that require navigating personal and environmental barriers (Gholamnia-Shirvani et al., 2018; Ma et al., 2023).

The TPB proposes that one’s intention is determined by three core constructs: Attitude, Subjective Norm, and PBC (Ajzen, 2002; Rodrigues et al., 2021). While numerous studies have confirmed the predictive power of these constructs, a key area of ongoing research synthesis involves understanding the model’s limitations and potential extensions (Hagger & Hamilton, 2025). One of the most discussed issues is the “intention-behavior gap,” where positive intentions do not always translate into action. This has led scholars to explore additional variables to augment the TPB. Among these, past behavior has emerged as one of the most powerful predictors, a point reinforced in a major 2023 meta-analysis which confirmed its consistent contribution to predicting future behavior alongside TPB variables (Hagger & Hamilton, 2024).

Furthermore, the components of the TPB itself are subject to deeper analysis. For instance, PBC is often conceptualized as being closely related to Bandura’s concept of self-efficacy (Ajzen, 2002), and recent research continues to affirm that higher self-efficacy is a cornerstone for forming strong exercise intentions in adolescents (Balla et al., 2024). The influence of Subjective Norm can also be culturally contingent, and its specific impact on adolescent exercise in an Asian context still requires more nuanced investigation (Harinurdin et al., 2024). In fact, a new line of inquiry explores how other psychological factors, such as “anticipated regret,” can add significant predictive power to the TPB for health behaviors like physical activity, suggesting that emotional forecasting also plays a role in intention formation (Tinnes-Vigne et al., 2025).

Therefore, this study is situated within this critical extension of the TPB. By integrating past behavior, we aim to test the sufficiency of the core TPB model and clarify whether habit and prior experience are foundational elements for building the attitudes and control perceptions that drive exercise intentions in the specific cultural and developmental context of Taiwanese female adolescents.

Hypothesis Derivation

The Theory of Planned Behavior (TPB) provides a robust framework for predicting behavioral intentions. Given the extensive body of existing research supporting the theory’s core relationships, particularly in exercise contexts, this study employs directional hypotheses. This approach is chosen over null hypotheses as it offers a more powerful and precise test of a well-established theory, where the positive direction of the relationships between constructs is theoretically expected (Ajzen, 2020). Based on the literature reviewed, the following hypotheses are proposed:

Past behavior is a critical antecedent in the extended TPB model. Direct, positive experiences with exercise build skills and enjoyment, which logically shape a more favorable personal evaluation of the activity (Hagger, 2019; Y. C. Hsu et al., 2009). Successfully engaging in past exercise also enhances one’s confidence and perceived capability to perform it in the future, directly informing their sense of control (Ajzen, 2020). Recent studies continue to confirm that habit strength, derived from past behavior, is a foundational element for building stable intentions (Zhang & Huang, 2024). Therefore:

H1: Past behavior positively influences attitude toward regular exercise.

H2: Past behavior positively influences perceived behavioral control over regular exercise.

Attitude itself is also shaped by other cognitions. The link from subjective norm to attitude (H3) can be explained through the psychological process of social alignment and internalization. From a conceptual standpoint, when an individual perceives that significant others endorse a behavior, adopting a congruent personal attitude can serve as a mechanism to strengthen social bonds and affirm one’s identity within that group (Ajzen, 2002). Recent research in contexts with high social influence confirms that subjective norms, representing external social pressure, can significantly shape an individual’s internal, personal evaluation (attitude) of a behavior (L. Wang et al., 2023).

Similarly, the pathway from perceived behavioral control to attitude (H4) is grounded in principles of self-efficacy and cognitive appraisal. A high level of PBC, which reflects strong self-efficacy, means an individual feels competent and capable of performing the behavior. This feeling of competence is intrinsically rewarding. Consequently, the behavior itself is appraised not as a daunting challenge, but as an achievable opportunity for demonstrating competence, which in turn fosters a more positive attitude toward it (Hagger & Hamilton, 2024). The conceptual link between educational/social support, the subsequent increase in self-efficacy (a core component of PBC), and the formation of a more positive attitude has been supported in recent models exploring behavioral intentions among students (Ng et al., 2024). Thus:

H3: Subjective norm positively influences attitude toward regular exercise.

H4: Perceived behavioral control positively influences attitude toward regular exercise.

Finally, the core TPB model delineates the primary predictors of behavioral intention. A positive attitude and a high sense of perceived behavioral control are consistently found to be the strongest and most direct drivers of intention to engage in health behaviors (Rodrigues et al., 2021; Zong et al., 2023). While classic TPB also posits a direct, positive influence of subjective norm on intention (H6), its effect can be highly context-dependent and is sometimes the weakest predictor in the model (La Barbera & Ajzen, 2021).

From our observations within the Taiwanese high school context, students operate under intense academic pressure, where decisions about personal time, such as for exercise, are frequently framed as individual responsibilities rather than social activities. In such an environment, it is plausible that internal drivers like personal enjoyment (attitude) might substantially outweigh perceived social pressure from peers or family (Xiao et al., 2025). Recent integrated models also suggest that while social factors are present, their direct link to intention is often mediated by more personal factors like motivation (Mkrtichian, 2025). Therefore, while we hypothesize a positive relationship for H6 based on the original theory, it represents a critical link to test. The potential for its rejection could provide valuable insight into the unique motivational landscape of this specific student population, making it an important hypothesis to investigate rather than assume.

H5: Attitude toward regular exercise positively influences behavioral intention.

H6: Subjective norm positively influences behavioral intention.

H7: Perceived behavioral control positively influences behavioral intention.

Based on the above literature, the conceptual framework for this study is presented as follows (Figure 1).

Figure 1.

Research structure.

Methods

Questionnaire Survey Design

The questionnaire used in this study consists of two main sections. The first section focuses on the design of measurement items, which were developed based on a comprehensive review of relevant domestic and international literature. The items measuring past behavior were adapted from the scales proposed by Verplanken and Orbell (2003) and Hagger (2019). Items assessing attitude toward regular exercise were derived from the dimensions outlined by Y. C. Chang et al. (2016) and Gaia et al. (2021). The subjective norm scale was based on the frameworks provided by Schiffman and Kanuk (2000) and Y. C. Chang et al. (2016). The scale for perceived behavioral control was informed by the work of Taylor and Todd (1995), Y. C. Chang et al. (2016), and Y. T. Chang et al. (2019). Finally, items measuring behavioral intention were drawn from the dimensions proposed by Parasuraman et al. (1996), Lee and Yu (2014), and Gholamnia-Shirvani et al. (2018).

The second section of the questionnaire collects demographic information, including school type, grade level, experience participating in sports clubs, preferred type of sport, and the number of times the respondent engaged in regular exercise during the past 6 months. A 5-point Likert scale was employed to assess agreement with each statement, ranging from 1 (“strongly disagree”) to 5 (“strongly agree”). Table 1 provides a detailed overview of the questionnaire items and their respective sources.

Table 1.

Questionnaire Item and Reference.

Variables	Items		Source
Past behavior		1. Regular exercise becomes the habit in my daily life.	Verplanken and Orbell (2003), Hagger (2019)
		2. I would automatically engage in regular exercise.
		3. I often engage in regular exercise.
Attitude towards regular exercise	Cognition	1. I consider that engaging in regular exercise could benefit individual physical and mental health.	Y. C. Chang et al. (2016), Gaia et al. (2021)
		2. I consider that engaging in regular exercise could enhance learning efficiency.
		3. I consider the need for regular exercise.
	Affection	4. Engaging in regular exercise allows me acquiring pleasant experience.
		5. I feel that engaging in regular exercise is advantageous to me.
		6. For me, it is not a waste of time to engage in regular exercise.
	Behavior	7. I would devote to the learning in PE course to enhance the ability and knowledge for regular exercise.
		8. I would invest in time learning and preparing for engaging in regular exercise.
		9. In comparison with other activities, regular exercise is my favorite choice.
Subjective norm for regular exercise		1. Family support is the key factor in my participation in regular exercise.	Schiffman and Kanuk (2000), Y. C. Chang et al. (2016)
		2. Friends’ support is the key factor in my participation in regular exercise.
		3. Teachers’ support is the key factor in my participation in regular exercise.
		4. Sports stars or groups are the key factors in my participation in regular exercise.
Perceived behavior control of regular exercise		1. I am confident in continuously participating in regular exercise.	Taylor and Todd (1995), Y. C. Chang et al. (2016), G. Z. Chen et al. (2018)
		2. I have sufficient information sources as the reference to decide the engaged regular exercise.
		3. I have sufficient time to engage in regular exercise.
		4. I have sufficient sports knowledge and skills to engage in regular exercise.
Behavioral intention of regular exercise		1. I would introduce the advantage of regular exercise to friends and relatives.	Parasuraman et al. (1996), Lee and Yu (2014), Gholamnia-Shirvani et al. (2018)
		2. I would recommend good sports items to friends and relatives.
		3. I would encourage friends and relatives to engage in regular exercise.
		4. I plan to engage in regular exercise in the following 6 months.
		5. I would try to engage in regular exercise in the following 6 months.

Participants and Procedure

The target participants for this study were senior and vocational high school female students in Taiwan. Data collection was conducted during March and April 2023. A convenience sampling method was employed for participant recruitment. Five high schools in central Taiwan were selected based on their accessibility and willingness to participate in the research. With the cooperation of school administrators, questionnaires were distributed to students during non-instructional periods, such as self-study sessions.

The inclusion criterion for participation was being a currently enrolled female student at one of the selected schools. Questionnaires that were substantially incomplete were excluded from the final analysis. In total, 400 questionnaires were distributed, and 392 valid responses were collected, resulting in a response rate of 98.3%. Due to the non-probability nature of the sampling method, the findings should be interpreted as being specific to the participant group. All collected data were analyzed using SPSS Windows 19.0 and Structural Equation Modeling (SEM) to assess reliability and validity and to test the proposed hypotheses.

To minimize potential risks, this study employed a non-invasive questionnaire survey and did not include any sensitive or personally identifiable questions. Participation was entirely voluntary, and participants were informed that they could decline participation or withdraw from the study at any time without penalty. The potential benefits of the research—including enhancing understanding of the factors influencing regular exercise intentions among female adolescents and informing future school-based health promotion strategies—were considered to outweigh any minimal risk involved. Prior to questionnaire administration, participants were clearly informed of the study’s purpose and procedures, and informed consent was implied through the voluntary completion of the survey.

Result

Basic Attribute Analysis

The basic data analysis of samples, Table 2, shows 40.3% and 59.7% of senior and vocational high schools, respectively, 19.4%, 31.6%, and 49.0% of G10, G11, and G12, respectively; engaging in regular exercise 1 to 2 times in past 6 months appears the most, about 24.5%; without participation experience in sports clubs appears the most, about 65.3%; and, with preferred sports item appears the most, about 91.8%.

Table 2.

Basic Data Analysis of Samples.

Variable	Type	No. sample	Percentage (%)	Variable	Type	No. sample	Percentage (%)
School type	Senior high school	158	40.3	Have a favorite sport	Yes	360	91.8
School type	Vocational high school	234	59.7	Have a favorite sport	No	32	8.2
Grade	Grade 1	76	19.4	Number of times engaging in regular exercise in past 6 months	Never	37	9.4
	Grade 2	124	31.6		1–2 times	96	24.5
	Grade 3	192	49.0		3–4 times	55	14.0
Experience in sports club	Yes	136	34.7		5–10 times	64	16.3
	No	256	65.3		11–20 times	48	12.2
					More than 20 times	92	23.5

Measurement Model Analysis

The overall theoretical model analysis results, Table 3, reveal the standardized factor loading of the theoretical model higher than 0.5 achieves the significance (Taylor & Todd, 1995). CR above 0.60 is acceptable, and the average variance extracted of dimensions higher than 0.40 is acceptable. Accordingly, the measurement of dimensions in this study achieve acceptable standards.

Table 3.

Measurement Model Analysis Result.

Indicator		Factor loading	CR	AVE
Past behavior	A1	0.921	0.934	0.826
	A2	0.913
	A3	0.892
Attitude towards regular exercise	Cognition	0.724	0.750	0.504
	Affection	0.707	0.798	0.571
	Behavior	0.730	0.814	0.594
Subjective norm for regular exercise	C1	0.864	0.880	0.649
	C2	0.794
	C3	0.848
	C4	0.708
Perceived behavior control of regular exercise	D1	0.834	0.908	0.711
	D2	0.807
	D3	0.854
	D4	0.877
Behavioral intention of regular exercise	F1	0.825	0.924	0.708
	F2	0.846
	F3	0.888
	F4	0.837
	F5	0.808

Structural Equation Modeling Analysis

Evaluation of Theoretical Model

The adequacy of the proposed theoretical model was evaluated using Structural Equation Modeling (SEM), assessing a range of goodness-of-fit indices as recommended by Kline (2005) and Jeng et al. (2017). The model produced the following fit statistics: χ² = 1,067.424, χ²/df = 4.864, GFI = 0.811, AGFI = 0.843, RMR = 0.031, RMSEA = 0.078, NFI = 0.851, and CFI = 0.877.

While some scholars recommend a stringent threshold of 0.90 for indices like the GFI and CFI, this may not always be appropriate for more complex models. As Doll et al. (1994) argued, achieving a 0.90 standard can be difficult as the number of estimated parameters in a model increases, suggesting that a value above 0.80 can be considered an acceptable fit. Similarly, Ullman (2001) noted that the NFI can be underestimated in smaller samples, also warranting a more flexible threshold of 0.80. Given the complexity of our extended TPB model, our GFI of 0.811, NFI of 0.851, and CFI of 0.877 are therefore considered to be within an acceptable range, providing support for the model’s structure.

Interpreting the other indices further supports the model’s robustness. The chi-square to degrees of freedom ratio (χ²/df = 4.864) falls within the commonly accepted range of less than 5.0. Furthermore, the low Standardized Root Mean Square Residual (RMR = 0.031) indicates that there is very little unexplained variance between the model and the observed data. The Root Mean Square Error of Approximation (RMSEA = 0.078) is also below the 0.08 threshold, suggesting an adequate approximation of the model to the population.

In this study, the primary goal was to test the theoretically-driven extended TPB model as a whole, and therefore, comparisons with alternative nested models were not conducted (Table 4). In summary, while no single index is perfect, the collective evidence from the absolute, incremental, and parsimonious fit indices suggests that the proposed theoretical model provides a reasonably robust and meaningful explanation of the psychological factors influencing exercise intention in our sample. The overall fit is deemed acceptable for proceeding with hypothesis testing.

Table 4.

Overall Model Fit.

Statistical test		Standard value	Model value	Model fit result
Absolute fit indices	χ²	p > .05	1,067.427
	χ²/df	<3 or <5	4.864	Match
	GFI	>0.9 or 0.8	0.811	Match
	AGFI	>0.9 or 0.8	0.843	Match
	RMR	<0.05	0.031	Match
	RMSEA	<0.08	0.078	Match
Incremental fit indices	NFI	>0.9 or 0.8	0.851	Match
Incremental fit indices	CFI	>0.90 or 0.8	0.877	Match
Model parsimony fit	PNFI	>0.5	0.733	Match
Model parsimony fit	PGFI	>0.5	0.630	Match

Research Hypothesis Testing

The effects among variables in this study are tested, Figure 2. Except H5 and H7 with t not higher than the standard value 1.96, the rest eight hypotheses are supported, with p reaching the significance .05. The path coefficient of the structural model and hypothesis test are shown in Table 5.

Figure 2.

Path diagram of research relation model.

Table 5.

Path Coefficient and Hypothesis Test of Theoretical Structural Model.

Hypothesis	Relations between dimensions	Hypothesis relation	Path value	t	Test result
H1	Past behavior → attitude towards regular exercise	Positive	0.339	6.359***	Supported
H2	Past behavior → perceived behavior control of regular exercise	Positive	0.537	5.365***	Supported
H3	Subjective norm for regular exercise → attitude towards regular exercise	Positive	0.707	7.109***	Supported
H4	Perceived behavior control of regular exercise → subjective norm for regular exercise	Positive	1.768	5.408***	Supported
H5	Attitude towards regular exercise → behavioral intention of regular exercise	Positive	0.385	5.041***	Supported
H6	Subjective norm for regular exercise → behavioral intention of regular exercise	Positive	0.066	1.240	Unsupported
H7	Perceived behavior control of regular exercise → behavioral intention of regular exercise	Positive	0.326	4.892***	Supported

p < .01. ***p < .001.

The path coefficient of past behavior to attitude towards regular exercise shows 0.339 and t = 6.359 that H1 is supported, revealing the higher regular exercise times in the past, the higher attitude towards regular exercise. Such a result conforms to the research results of Mowatt et al. (1988), L. M. Yang and Ku (2004), and G. Z. Chen et al. (2018). The path coefficient of past behavior to perceived behavior control of regular exercise appears 0.537 and t = 5.365 that H2 is supported, revealing the higher regular exercise times in the past, the higher perceived behavior control of regular exercise. The result conforms to the research results of Taylor and Todd (1995) and Rodrigues et al. (2021).

The path coefficient of subjective norm for regular exercise to attitude towards regular exercise shows 0.707 and t = 7.109 that H3 is supported, revealing the higher subjective norm for regular exercise, the higher attitude towards regular exercise. The result conforms to the research results of Wu et al. (2003) and Y. T. Chang et al. (2019). The path coefficient of perceived behavior control of regular exercise to subjective norm for regular exercise appears 1.768 and t = 5.408 that H4 is supported, revealing the higher perceived behavior control of regular exercise, the higher subjective norm for regular exercise. Such a result conforms to the research results of Wu et al. (2003) and C. M. Hsu and Wu (2016).

The path coefficient of attitude towards regular exercise to behavioral intention of regular exercise appears 0.385 and t = 5.041 that H5 is supported, revealing the higher attitude towards regular exercise, the higher behavioral intention of regular exercise. Such a result conforms to the research results of Kim and Kim (2021), Gatch and Kendzierski (1990), and Chi and Hsu (2005). The path coefficient of subjective norm for regular exercise to behavioral intention of regular exercise is 0.066 and t = 1.240, smaller than the standard value 1.96, that H6 is not supported, revealing that subjective norm for regular exercise does not remarkably affect behavioral intention of regular exercise. The result does not conform to the hypothesis in this study, but matches the research results of C. T. Wang et al. (2010) and Atsalakis and Sleap (1996).

The path coefficient of perceived behavior control of regular exercise to behavioral intention of regular exercise appears 0.326 and t = 4.892 that H7 is supported, revealing the higher perceived behavior control of regular exercise, the higher behavioral intention of regular exercise. The result conforms to the research result of Kim and Kim (2021).

Variance Analysis and Group Comparison

To explore potential differences in the core theoretical constructs based on participants’ exercise history, an Analysis of Variance (ANOVA) was conducted. The results, shown in Table 6, indicated that the variable “times engaging in regular exercise in past 6 months” had a statistically significant effect on the model’s variables.

Table 6.

Analysis of Variance of Behavioral Intention of Regular Exercise in Various Background Variable.

Individual background variable	Significance
School type	.159
Grade	.106
Times engaging in regular exercise in past 6 months	.000***
Participation experience in sports clubs	.690
Have a favorite sport	.294

***

p < .001.

To further investigate the nature of this effect and identify which specific groups differed from one another, a post hoc multiple comparison analysis was necessary. This required categorizing the continuous frequency data into distinct groups. The justification for this grouping strategy was twofold. First, the cut-off points were established to create four qualitatively distinct and meaningful categories reflecting different levels of exercise engagement: a “low frequency group” (0–2 times), a “medium-low frequency group” (3–10 times), a “medium frequency group” (11–20 times), and a “high frequency group” (more than 20 times). This categorization allows for a more nuanced interpretation of how varying degrees of past exercise experience relate to the TPB constructs.

Second, this categorization was conducted while ensuring statistical robustness. As can be seen in the sample demographics in Table 1, the sample size for each of the four groups after grouping was confirmed to be above 30 participants, which is considered sufficient for reliable statistical comparison and meets the assumptions for this type of analysis.

For a supplementary analysis comparing broader engagement levels, a dichotomous grouping was also utilized, classifying participants into a “low sports participation group” (10 times or fewer in the past 6 months) and a “high sports participation group” (11 times or more). The results of the multiple comparisons based on these groupings are detailed in the subsequent section.

Structural Model Analysis of Times Engaging in Regular Exercise in Past 6 Months

To discuss the effects of senior and vocational high school female students with different times of engagement in exercise in past 6 months on the regular exercise intention, the regular exercise times of “low frequency group”, “medium low frequency group”, “medium frequency group”, “high frequency group”, “low sports participation group”, and “high sports participation group” are preceded structural model reliability and validity analyses. The analysis results show the Cronbach’s α of the structural models higher than the standard value .7, revealing the internal stability and consistency. CR of the dimensions is higher than the acceptable standard of 0.60∼0.70, and the average variance extracted (AVE) is higher than the acceptable standard of 0.40∼0.50 that the models present good convergent validity. The structural model analyses are shown as below.

Structural Model Analysis of Engagement in Low Frequency Regular Exercise in Past 6 Months

From Figure 3, “attitude towards regular exercise” appears the highest positive effect on “behavioral intention of regular exercise” (path coefficient = 0.701), followed by “perceived behavior control of regular exercise” (path coefficient = 0.258). It reveals the more positive attitude towards regular exercise, the higher intention of regular exercise. An individual perceiving the higher control would have high behavioral intention of regular exercise.

Figure 3.

Structural model analysis result of engagement in low frequency regular exercise in past 6 months.

Structural Model Analysis of Engagement in Medium Low Frequency Regular Exercise in Past 6 Months

Figure 4 shows “Perceived behavior control of regular exercise” appears the highest positive effect on “behavioral intention of regular exercise” (path coefficient = 0.398), revealing the higher perceived behavior control of regular exercise, the higher behavioral intention of regular exercise of senior and vocational high school female students merely engaging in medium low frequency regular exercise in past 6 months. The more positive regular exercise habit in the past also shows higher behavioral intention of regular exercise. The path coefficients of “subjective norm for regular exercise”, and “attitude towards regular exercise” to “behavioral intention of regular exercise” appear 0.003, and 0.191, respectively, revealing insignificant effects of above two variables on senior and vocational high school female students with medium low frequency of regular exercise in past 6 months.

Figure 4.

Structural model analysis results of engagement in medium low frequency regular exercise in past 6 months.

Structural Model Analysis of Engagement in Medium Frequency Regular Exercise in Past 6 Months

Figure 5 shows positive effects of “attitude towards regular exercise” (path coefficient = 0.562) on “behavioral intention of regular exercise”, revealing the more positive attitude towards regular exercise, the higher behavioral intention of regular exercise, the higher behavioral intention of regular exercise.

Figure 5.

Structural model analysis results of engagement in medium frequency regular exercise in past 6 months.

Structural Model Analysis of Engagement in High Frequency Regular Exercise in Past 6 Months

Figure 6 shows “Attitude towards regular exercise” (path coefficient = 0.467) and “perceived behavior control of regular exercise” (path coefficient = 0.364) present positive effects on “behavioral intention of regular exercise”, revealing the higher attitude towards regular exercise, the higher behavioral intention of regular exercise; the higher individual perceived control, the higher behavioral intention of regular exercise; and, the more positive of past regular exercise habit, the higher behavioral intention of regular exercise.

Figure 6.

Structural model analysis results of engagement in high frequency regular exercise in past 6 months.

Structural Model Analysis of Engagement in Regular Exercise of Low Sports Participation Group in Past 6 Months

Figure 7 shows “Attitude towards regular exercise” (path coefficient = 0.545), “perceived behavior control of regular exercise” (path coefficient = 0.330) present positive effects on “behavioral intention of regular exercise”, revealing the higher attitude towards regular exercise, the higher behavioral intention; the higher perceived behavior control, the higher behavioral intention; and the more positive past regular exercise habit, the higher behavioral intention.

Figure 7.

Structural model analysis of engagement in exercise of low sports participation group in past 6 months.

Structural Model Analysis of Engagement in Regular Exercise of High Sports Participation Group in Past 6 Months

Figure 8 shows the highest path coefficient (0.587) of “past behavior” to “perceived behavior control of regular exercise”, revealing the more positive of past regular exercise habit, the higher perceived behavior control. “Attitude towards regular exercise” (path coefficient = 0.547), “perceived behavior control of regular exercise” (path coefficient = 0.282), appear positive effects on “behavioral intention of regular exercise”, revealing the higher attitude towards regular exercise, the higher behavioral intention of regular exercise; the higher perceived behavior control, the higher behavioral intention of regular exercise.

Figure 8.

Structural model analysis of engagement in exercise of high sports participation group in past 6 months.

Research Hypothesis Support of Times Engaging in Regular Exercise in Past 6 Months

Figures 3 to 8 show that past behavior of regular exercise does not appear significant effects on behavioral intention of regular exercise of senior & vocational high school female students with low frequency and medium frequency, but presents positive effects on those with medium low frequency and high frequency regular exercise, low sports participation group and high sports participation group. Attitude towards regular exercise does not show remarkable effects on behavioral intention of regular exercise of senior & vocational high school female students with medium low frequency regular exercise in past 6 months, but presents notable effects on the rest. Perceived behavior control of regular exercise does not significantly affect behavioral intention of regular exercise of medium frequency group, but shows positive and remarkable effects on senior & vocational high school female students with low frequency, medium low frequency, and high frequency regular exercise in past 6 months as well as low sports participation group and high sports participation group.

According to different exercise times in past 6 months, the support of various dimensions to research hypotheses analyzed in this study are organized in Table 7.

Table 7.

Research Hypothesis Support of Times Engaging in Regular Exercise in Past 6 Months.

Path relation	Times for regular exercise
Path relation	Low frequency	Medium low frequency	Medium frequency	High frequency	Low sports participation group	High sports participation group
H1. Past behavior → attitude towards regular exercise	✓	×	×	✓	✓	✓
H2. Past behavior → perceived behavior control of regular exercise	×	×	✓	×	✓	✓
H3. Subjective norm for regular exercise → attitude towards regular exercise	×	✓	×	×	✓	✓
H4. Perceived behavior control of regular exercise → subjective norm for regular exercise	×	✓	×	×	✓	×
H5. Attitude towards regular exercise → behavioral intention of regular exercise	✓	×	✓	✓	✓	✓
H6. Subjective norm for regular exercise → behavioral intention of regular exercise	×	×	×	×	×	×
H7. Perceived behavior control of regular exercise → behavioral intention of regular exercise	✓	✓	×	✓	✓	✓

Note. ✓ = supported; × = not supported.

Discussion and Conclusion

This study aimed to explore the factors influencing the intention of regular exercise among female upper secondary students in Taiwan by integrating “past behavior” into the Theory of Planned Behavior (TPB) framework. The findings suggest that this extended model holds considerable explanatory power for this specific demographic. Our results indicate that students’ attitudes toward exercise and their perceived behavioral control are primary drivers of their intention to exercise regularly. Notably, positive past experiences with exercise appear to be a foundational element, significantly shaping both more favorable attitudes and a stronger sense of personal control.

From a theoretical perspective, these findings contribute to the ongoing discussion about the sufficiency of the original TPB model, particularly for habitual health behaviors. The strong influence of past behavior underscores its value as an external variable, suggesting that for exercise, habit formation may be as crucial as conscious deliberation. An interesting and perhaps culturally specific finding is the non-significant effect of subjective norm on behavioral intention. This outcome aligns with the possibility raised during our hypothesis development, suggesting that for Taiwanese female adolescents, the decision to engage in regular exercise is driven more by internal factors, such as personal enjoyment (attitude) and self-confidence (perceived behavioral control), rather than by perceived social pressure from peers or family. This observation warrants further investigation and offers a nuanced perspective on the application of TPB in non-Western contexts.

The practical implications of these findings are pertinent for educators and school health administrators. Instead of solely focusing on promoting the benefits of exercise, interventions should perhaps prioritize creating positive and successful early experiences with physical activity. This could involve designing a wider variety of non-competitive sports and activities that cater to diverse interests and skill levels, thereby building students’ confidence and positive associations with exercise (enhancing past behavior). Furthermore, educational efforts could focus on developing students’ self-efficacy, for instance, by teaching them how to set achievable goals and overcome common barriers, which in turn would strengthen their intention to exercise. The results suggest that fostering an internal motivation may be more effective than relying on social encouragement alone.

It is important to acknowledge the limitations of this study, which may guide future research. First, the use of a convenience sampling method means the findings may not be fully generalizable to all female high school students in Taiwan. Future studies could employ probability sampling techniques to enhance sample representativeness. Second, our measurement of past behavior focused on the psychological construct of habit strength, such as automaticity and frequency. While this aligns with the theoretical basis of the extended TPB model, it did not capture other physical dimensions of exercise patterns, such as the specific duration, consistency, or intensity of past activities. Future studies could employ more objective or varied measures, such as exercise logs or accelerometer data, to explore whether these dimensions provide additional explanatory power. Third, this study utilized a cross-sectional design, which captures a single point in time. A longitudinal approach would be beneficial to track how exercise intentions translate into actual behavior over time and to better understand the process of habit formation.

In conclusion, this study provides insights into the psychological determinants of exercise intention in a key demographic. By highlighting the pivotal roles of past behavior, attitude, and perceived behavioral control, it offers a more comprehensive understanding that can hopefully inform the development of more effective, evidence-based health promotion strategies for young women.

Footnotes

ORCID iDs

Tsu-Ming Yeh

Fan-Yun Pai

Ethical Considerations

This study was conducted in accordance with ethical standards for research involving human participants. The research employed a non-invasive questionnaire survey and did not involve any sensitive or personally identifiable information. Participation was entirely voluntary, and anonymity and confidentiality were strictly maintained throughout the research process. The study design involved minimal risk to participants, and the potential benefits of the research were considered to outweigh any minimal risks involved.

Consent to Participate

Prior to data collection, participants were clearly informed of the purpose and procedures of the study. Participation was voluntary, and participants were free to decline participation or withdraw at any time without penalty. Informed consent was implied through the voluntary completion of the questionnaire.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.*

References

Ajzen

(1985). From intentions to actions: A theory of planned behavior. In Kuhl

Beckman

(Eds.), Action-control: From cognition to behavior (pp. 11–39). Springer Verlag.

Ajzen

(2002). Residual effects of past on later behavior: Habituation and reasoned action perspective. Personality and Social Psychological Review, 6, 107–122.

Ajzen

(2020). The theory of planned behavior: Frequently asked questions. Human Behavior and Emerging Technologies, 2(4), 314–324.

Atsalakis

Sleap

(1996). Registration of children in a physical activity program: An application of the theory of planned behavior. Pediatric Exercise Science, 8(2), 166–176.

Balla

Polet

Kokko

Hirvensalo

Vasankari

Lintunen

Hagger

M. S.

(2024). Predicting adolescents’ physical activity intentions: Testing an integrated social cognition model. International Journal of Behavioral Medicine, 31(1), 41–54.

Chang

Y. C.

Yeh

T. M.

Chen

S. Y.

(2016). The relationships between leisure sports participation motivation, and leisure benefits for senior and senior & vocational high school female students. Journal of Sport, Leisure and Hospitality Research, 11(4), 1–28.

Chang

Y. T.

Chen

H. C.

Yang

C. K.

Chang

C. M.

(2019). A study on relay race performance of junior high school students by means of planned behavior theory. Journal of Sport and Recreation Research, 13(4), 1–19.

Chen

G. Z.

Chan

C. C.

Chen

M. Y.

(2018). Running on the clouds: A study of the user behavioral intention model on Nike+ Run club. Taiwan Journal of Sports Scholarly Research, 64, 37–54.

Chen

Sekine

Hamanishi

Wang

Gaina

Yamagami

Kagamimori

(2005). Lifestyles and health-related quality of life in Japanese school children: Across-sectional study. Preventive Medicine, 40, 668–678.

10.

Chi

L. K.

Hsu

C. C.

(2005). The prediction of regular exercise intention and behavior among elementary school teachers—A test the theory of planned behavior. Bulletin of Sport and Exercise Psychology of Taiwan, 7, 115–129

11.

Chiu

Y. P.

(2020). A study on exercise participative motivation and exercise self-efficacy—A case of elementary school teachers in Dounan area. Journal of Sport and Recreation Research, 15(1), 1–12.

12.

Doll

W. J.

Xia

Torkzadeh

(1994). A confirmatory factor analysis of the end-user computing satisfaction instrument. MIS Quarterly, 12(2), 259–274.

13.

Fishbein

Ajzen

(1980). Understanding attitudes and predicting social behavior (pp. 4–129). Prentice-Hall.

14.

Flynn

R. M.

Staiano

A. E.

Beyl

Richert

R. A.

Wartella

Calvert

S. L.

(2018). The influence of active gaming on cardiorespiratory fitness in Black and Hispanic youth. Journal of School Health, 88(10), 768–775.

15.

Gaia

J. W. P.

Ferreira

R. W.

Pires

D. A.

(2021). Effects of physical activity on the mood states of young students. Journal of Physical Education, 32(1), 1–9

16.

Gatch

C. L.

Kendzierski

(1990). Predicting exercise intention the theory of planned behavior. Research Quarterly For Exercise and Sport, 61(1), 100–102.

17.

Gholamnia-Shirvani

Ghofranipour

Gharakhanlou

Kazemnejad

(2018). “Women and active life”: An extended TPB-based multimedia software to boost and sustain physical activity and fitness of Iranian women. Women & Health, 58(7), 834–850.

18.

Hagger

M. S.

(2019). Habit and physical activity: Theoretical advances, practical implications, and agenda for future research. Psychology of Sport and Exercise, 42, 118–129.

19.

Hagger

M. S.

Hamilton

(2024). Longitudinal tests of the theory of planned behaviour: A meta-analysis. European Review of Social Psychology, 35(1), 198–254.

20.

Hagger

M. S.

Hamilton

(2025). Progress on theory of planned behavior research: Advances in research synthesis and agenda for future research. Journal of Behavioral Medicine, 48(1), 43–56.

21.

Harinurdin

Safitri

K. A.

Kountur

(2024). Applying the theory of planned behavior in examining the intention to improve corporate tax compliance through compensation and counterproductive work behavior. SAGE Open, 14(4), 1–15.

22.

Herbert

Meixner

Wiebking

Gilg

(2020). Regular physical activity, short-term exercise, mental health, and well-being among university students: The results of an online and a laboratory study. Frontiers in Psychology, 11, 509.

23.

Honkanen

Olsen

S. O.

Verplanken

(2005). Intention to consume seafood: The importance of habit. Appetite, 45(2), 161–168.

24.

Hsu

C. M.

T. J.

(2016). A study of regular exercise intention for female elementary school teacher in Taoyuan. Journal of Recreation Sport and Health Promotion, 11, 67–77.

25.

Hsu

Y. C.

Chuang

L. J.

Yeh

C. K.

T. K.

(2009). Inclusion of past behavior in the theory of planned behavior to predict leisure and recreation intention and behavior of high school students. Tourism Management Research, 8(2), 145–162.

26.

Jeng

M. Y.

Pai

F. Y.

Yeh

T. M.

(2017). The virtual reality leisure activities experience on elderly people. Applied Research Quality Life, 12, 49–65.

27.

Joseph

R. P.

Royse

K. E.

Benitez

T. J.

Pekmezi

D. W.

(2014). Physical activity and quality of life among university students: Exploring self-efficacy, self-esteem, and affect as potential mediators. Quality of Life Research, 23(2), 659–667.

28.

Kim

Y. J.

Kim

E. S.

(2021). Analysis of Korean fencing club members’ participation intention using the TPB model. International Journal of Environmental Research and Public Health, 18(6), 2813.

29.

Kline

R. B.

(2005). Principles and practice of structural equation modeling (2nd ed.). Guilford.

30.

La Barbera

Ajzen

(2021). Moderating role of perceived behavioral control in the theory of planned behavior: A preregistered study. Journal of Theoretical Social Psychology, 5(1), 35–45.

31.

Lee

Y. H.

C. L.

(2014). Determinants of theory of planned behavior on regular exercise behavior among university students. Sports & Exercise Research, 16(1), 45–58

32.

Yin

Hipel

K. W.

(2023). Exploring factors influencing the application accuracy of the theory of planned behavior in explaining recycling behavior. Journal of Environmental Planning and Management, 66(3), 445–470.

33.

Mkrtichian

(2025). The role of motivation theories in physical activity engagement: A theoretical synthesis. Educological Discourse, 50(3), 49–62.

34.

Mowatt

Depauw

K. P.

Hulac

G. M.

(1988). Attitudes toward physical activity among college students. The Physical Educator, 45(2), 103–108.

35.

Myers

Kokkinos

Nyelin

(2019). Physical activity, cardiorespiratory fitness, and the metabolic syndrome. Nutrients, 11(7), 1–18.

36.

H. S.

Ramayah

Kee

D. M. H.

(2024). Examining mediating role of theory of planned behaviour in the links between personality, education, opportunities and entrepreneurial intentions. Journal of Infrastructure, Policy and Development, 8(10), 4974.

37.

Pai

F. Y.

Yeh

T. M.

(2017). Investigating the antecedents of customer behavioral intentions by service acceptance model. Human Factors and Ergonomics in Manufacturing & Service Industries, 25(4), 428–438.

38.

Parasuraman

Zeithaml

V. A.

Berry

L. L.

(1996). The behavioral consequences of service quality. Journal of Marketing, 60, 31–46.

39.

Pate

Pratt

Blair

Haskell

Macera

Bouchard

(1995). Physical activity and public health: A recommendation from the centers for disease control and prevention and the American college of sports medicine. Journal of the American Medical Association, 273, 402–407.

40.

Phillips

L. A.

Mullan

B. A.

(2023). Practical parsimony and complexity in conceptualising habit. Health Psychology Review, 17(3), 497–504.

41.

Qiu

Huang

Xiao

Nie

Zhou

Liu

(2025). The correlation between physical activity and psychological resilience in young students: A systematic review and meta-analysis. Frontiers in Psychology, 16, 1557347.

42.

Rodrigues

Teixeira

(2023). The relationships between passions, intentions, habit and exercise frequency. Journal of Sports Sciences, 41(8), 803–811.

43.

Rodrigues

Teixeira

D. S.

Cid

Monteiro

(2021). Have you been exercising lately? Testing the role of past behavior on exercise adherence. Journal of Health Psychology, 26(10), 1482–1493.

44.

Santos

R. M. S.

Mendes

C. G.

Sen Bressani

G. Y.

de Alcantara Ventura

de Almeida Nogueira

Y. J.

de Miranda

D. M.

Romano-Silva

M. A.

(2023). The associations between screen time and mental health in adolescents: A systematic review. BMC Psychology, 11(1), 1–21.

45.

Schiffman

L. G.

Kanuk

L. L.

(2000). Consumer behavior (7th ed.). Prentice Hall.

46.

Stamatakis

Hamer

Lawlor

D. A.

(2009). Physical activity, mortality, and cardiovascular disease: Is domestic physical activity beneficial?: The Scottish Health Survey-1995, 1998, and 2003. American Journal of Epidemiology, 169(10), 1191–1200.

47.

Taylor

Todd

P. A.

(1995). Assessing IT usage: The role of prior experience. MIS Quarterly, 19(4), 561–570.

48.

Tinnes-Vigne

Houssemand

Guay

Poncelet

Dierendonck

(2025). Understanding the use of social and emotional learning in elementary schools: A theory of planned behaviour perspective. European Journal of Investigation in Health, Psychology and Education, 15(4), 48.

49.

Ullman

J. B.

(2001). Structural equation modeling. In Tabachnick

B. G.

Fidell

L. S.

(Eds.), Using multivariate statistics (4th ed.; pp. 653–771). Allyn and Bacon.

50.

Verma

A. K.

Singh

Patwardhan

(2022). Patterns of physical activity among university students and their perceptions about the curricular content concerned with health: Cross-sectional study. JMIRx Med, 3(2), e31521.

51.

Verplanken

Orbell

(2003). Reflections on past behavior: A self-report index of habit strength. Journal of Applied Social Psychology, 33, 1313–1330.

52.

Wang

C. T.

Chen

C. T.

Wang

S. L.

(2010). A study on behavioral intentions of college students participating in leisure activities. Vanung Journal, 32, 279–291.

53.

Wang

Zhang

Ding

Y. Y.

Wong

P. P. W.

(2023). The effect of social and personal norm on intention to patronize green hotels: Extension of theory of planned behavior. Journal of China Tourism Research, 19(2), 311–334.

54.

Fan

L. W.

P. L.

(2003). The research of behavioral intention of college students’ participation in ecotourism. Journal of Environment and Management, 4(1), 1–28.

55.

Xiao

Huang

(2025). The impact of fitness social media use on exercise behavior: The chained mediating role of intrinsic motivation and exercise intention. Frontiers in Psychology, 16, 1635912.

56.

Yang

L. M.

Y. C.

(2004). The development and application of exercise attitude scale for university student. Physical Education Journal, 37, 149–161.

57.

Yang

P. Y.

K. H.

Chen

H. C.

Chien

M. Y.

(2012). Exercise training improves sleep quality in middle-aged and older adults with sleep problems: A systematic review. Journal of Physiotherapy, 58(3), 157–163.

58.

Yasunobe

Akasaka

Yamamoto

Onishi

Minami

Yoshida

Rakugi

(2023). The relationship between changes in exercise habits and psychosomatic activities in older hypertensive patients during the COVID-19 pandemic. Hypertension Research, 46(1), 208–213.

59.

Zhang

C. Q.

Huang

(2024). Can daily affect impact intentions and next day behavior of physical activity? A daily diary study. Mental Health and Physical Activity, 27, 100627.

60.

Zhu

Tao

Guo

Zhang

Wang

Zhou

Zhang

(2025). The relationship between habit and identity in health behaviors: A systematic review and three-level meta-analysis. Applied Psychology: Health and Well-Being, 17(2), e70017.

61.

Zong

Liu

Gao

(2023). Exploring the mechanism of consumer purchase intention in a traditional culture based on the theory of planned behavior. Frontiers in Psychology, 14, 1110191.