Adoption of Islamic microfinance in Indonesia an empirical investigation: an extension of the theory of planned behaviour

Abstract The theoretical framework of the Theory of Planned Behaviour (TPB) has been widely used in analysing behavioural intentions to adopt Islamic finance. However, there is still little research investigating the adoption of Islamic microfinance. This study aims to determine the factors influencing behavioural intentions to adopt Islamic microfinance by developing the Theory of Planned Behaviour (TPB). The development is done by adding religiosity and knowledge variables as antecedents of attitude. A random sample of 262 members of Islamic microfinance. Variance-based partial least-squares structural equation modelling (PLS-SEM) was employed for data analysis. The results show that religiosity and have a significant influence on attitudes. Meanwhile, attitudes, subjective norms, and perceived behavioural control also significantly positively affect behavioural intentions to adopt Islamic microfinance. Then subjective norms greatly determine behavioural intentions to adopt Islamic microfinance. The findings of this study are important for Islamic microfinance to change marketing strategies. Islamic microfinance managers must develop a positive attitude through creating awareness and benefits of Islamic microfinance services by involving community leaders, especially religious leaders.


PUBLIC INTEREST STATEMENT
Choosing products and services following religious values is an essential consideration because it is a form of commitment to carry out religious orders. Interest is prohibited in the religion of Islam. Therefore conventional financial institutions question based on religion. Islamic financial services, including Islamic microfinance, have been available for decades and are still not accepted by the public. This study provides insight into Islamic microfinance to assist managers in implementing marketing strategies effectively and efficiently by involving religious leaders in introducing Islamic microfinance. This study aims to determine the factors influencing behavioural intentions to adopt Islamic microfinance. This study provides information that religiosity and knowledge have an essential role in shaping attitudes towards Islamic microfinance.

Introduction
Islamic microfinance, or in Indonesia known as Baitul Maal wat Tanwil (BMT) is in demand by the public after the Covid-19 pandemic. This condition is evidenced by the number of requests for Islamic microfinance services (Interview with Islamic Microfinance Manager, 2021). The increasing number of financing applications, one of which is caused by the implementation of a Islamic microfinance system that is following sharia Islamic in its operations (Mawardi et al., 2018). Besides being caused by the application of Islamic sharia, Islamic microfinance is an alternative to obtain financing for people who have difficulty accessing banking (Maulana et al., 2018). Increased public knowledge of sharia microfinance is also the cause of demand for sharia microfinance services. this is shown by the increase in Islamic financial literacy, which was previously shown by an increase in financial literacy which was previously 8.1% in 2016 8.93% in 2019 (OJK, 2021). This increase in demand for Islamic microfinance services shows that the public's desire to adopt Islamic microfinance is relatively high.
The adoption of Islamic microfinance in this study was analysed using the Theory of Planned Behaviour (TPB) framework. TPB offers that the direct antecedent of a behaviour is the intention to behave. This intention is assumed to be determined by three considerations or beliefs: attitude, subjective norm, and perceived behavioural control (Ajzen, 2020). From previous research results, TPB is still little used in the context of behavioural intentions to adopt Islamic microfinance. Several studies related to the behavioural intention to adopt Islamic microfinance have been carried out by Maulana et al. (2018), Kachkar and Djafri (2021), and Umar et al. (2021). However, the three studies yielded different conclusions. Research conducted by Maulana et al. (2018) concludes that the adoption of Islamic microfinance is influenced by perceived behavioural control, while subjective attitudes and norms are not proven. Meanwhile, research by Kachkar and Djafri (2021) concludes that subjective norms do not affect the adoption of Islamic microfinance. As for Umar et al. (2021) all TPB variables have a significant influence on the adoption of Islamic microfinance. This condition illustrates that research on the application of Islamic microfinance with the TPB concept is still relevant to be carried out because the results of previous studies have not produced inconsistent conclusions.
In addition to using the original concept of TPB, this study also adds variables of religiosity and knowledge as antecedents of attitude. Religiosity is basically a belief in religious values, the power of motivation, and a commitment to carry out religious orders. As a value, religion is a force that is internally able to influence its adherents in carrying out a behaviour (Felix & Braunsberger, 2016). Someone who has a high commitment to his religion will use religious values to determine attitudes as an abstract concept that is consciously or unconsciously constructed by the interpretation of information obtained through experience and experience about itself. It can provide the owner's mental strength and physical ability (Ibrahim et al., 2017). Meanwhile, attitude is the evaluative effect of an individual's positive or negative feelings in performing certain behaviours (Kaakeh et al., 2018). Thus, the higher the level of religiosity and knowledge of Islamic microfinance, it will form a positive attitude towards the services offered by Islamic microfinance. This is evidenced in a study that the number of applications for financing to Islamic microfinance institutions is one of them due to a system that is in accordance with Islamic sharia in its operations (Mawardi et al., 2018). Likewise, research conducted by Latif (2021) proves that knowledge can affect attitudes.
This study aims to prove the three elements of TPB-attitude, subjective norm, and perceived behavioural control in the context of Islamic microfinance adoption. In addition, this study also seeks to develop TPB by adding religiosity and knowledge variables as anticipation of attitude.
Overall, this study is expected to contribute to the literature by focusing on the relationship between religiosity and knowledge in the context of Islamic microfinance adoption.

Theory of planned behaviour (TPB)
TPB was first developed in 1985, until now it is the most widely used theory in understanding and predicting behaviour. TPB is widely applied to explain relevant factors that influence certain problems (Kumar et al., 2017). TPB is a development of the Theory of Reasoned Action (TRA).. The main difference between TRA and TPB is that there is one additional variable that influences intention, namely perceived behavioural control (Ajzen, 1991). TPB can be used to predict the intentions and behaviours associated with the purchase of one brand or product as well as in the choice between different brands or products (Ajzen, 2015). The TPB model explains that a person's actions or certain behaviours are determined by behavioural intentions (Maulana et al., 2018). This shows that the behavioural intention in the TPB concept is a direct antecedent in determining actual behaviour.
In short, the TPB concept offers that the direct antecedent of a behaviour is the intention to behave. Behavioural intentions are determined by three considerations or beliefs: attitudes, subjective norms, and perceived behavioural control (Ajzen, 2020). Attitude is a perception of the positive or negative consequences of a behaviour (Purwanto, 2021). The belief that performing a behaviour gets benefits leads to forming a positive attitude (Ajzen, 2015). A second consideration has to do with the perceived expectations and behaviour of the reference individual or group that is considered important, combined with that person's motivation to comply with the intended reference and determining perceived social pressures or norms regarding behaviour. The third consideration, Perceived Behavioural Control, is the perceived perception of the existence of factors that can affect a person's ability to perform a behaviour (Ajzen, 2015). This consideration is related to the individual's perception of how easy or difficult it is to perform a behaviour (Albashir et al., 2018).

Islamic microfinance in Indonesia
In Indonesia, there are three types of Islamic Microfinance Institutions, namely: Baitul Maal wat Tamwil (BMT), Sharia Rural Bank (BPRS) and Islamic micro banking (Maulana et al., 2018). In the context of this research, the Islamic microfinance in question is Baitul Maal wat Tamwil (BMT). Baitul Maal wat Tamwil (BMT) is a non-bank financial institution based on sharia principles (Prasada et al., 2020), namely carrying out its business on the principle of profit-sharing. There are two types of financial management in BMT operations, namely baitul maal and baitut tamwil. Baitul maal is a treasure house, while baitut tamwil is property development. Baitul maal financial management includes the collection and distribution of Zakat, Infaq, and Wakaf funds from the community (Maulana et al., 2018). Meanwhile, Baitul Tamwil is an institution that carries out productive business development and investment activities to improve micro-entrepreneurs welfare through financing and savings activities (Prasada et al., 2020) in accordance with Islamic sharia principles.

Religiosity and attitude
Religiosity is considered a quality of religion and devotion to its adherents' beliefs about a religion (Mansori et al., 2020) and is one of the main constructs used in research on religion (Abou-Youssef et al., 2015). In the context of this research, the religion in question is Islam, so religiosity in question is how committed a person is to the teachings of Islam. Religiosity is the belief in God with a commitment to follow predetermined instructions (Nurhayati & Hendar, 2019). Meanwhile, Johnson et al. (2001) define religiosity as how a person is committed to religion and its teachings reflected in their attitudes and behaviour. From this understanding, a person's attitude reflects his religious beliefs. Souiden and Rani (2015) explain that religion and attitudes have a strong correlation and a contribution to controlling attitudes. The attitude is defined as how a person has a favourable or unfavourable evaluation or judgment from someone (Ajzen, 1991). The belief influences this attitude in the consequences of the behaviour (Ajzen, 2005). Attitude is also defined as the evaluative effect of the individual's positive or negative feelings in performing certain behaviours (Kaakeh et al., 2018). Islamic microfinance does not apply an interesting system in its transactions but uses a profit-sharing scheme. The interest system is prohibited in Islamic teachings because interest is considered the same as usury. Thus, someone with high religiosity will positively affect Islamic microfinance. It is through the compatibility of the Islamic microfinance system with the teachings of Islam that, in the end, the attitude towards Islamic microfinance becomes positive.
Research conducted by Wijaya et al. (2020) proves that people use sharia microfinance products because they are in accordance with sharia principles in their operations. The study conducted by Jaffar and Musa (2013) proved that as many as 58% of users of Islamic microfinance financing stated that the driving factor for using Islamic microfinance financing was the belief that Islamic finance would bring justice to society and was more profitable than conventional financing. Thus, a high level of religiosity can affect attitudes towards Islamic microfinance. Research conducted by Souiden and Rani (2015), Purwanto (2021), and Abou-Youssef et al. (2015) proves that the level of religiosity has a significant influence on attitudes. Thus the hypothesis proposed in this study is: H1: Religiosity has a positive influence on the attitude toward Islamic microfinance.

Knowledge and attitude
Knowledge is an important variable in determining attitudes. Knowledge is information combined with experience, context, interpretation, and reflection. It is a highly feasible form of information readily applicable to decisions and actions. cognitive science theories can be defined as an abstract concept that is consciously or unconsciously constructed by the interpretation of a collection of information obtained through experience and experience based on knowledge itself (Ibrahim et al., 2017). Thus knowledge of the system of Islamic microfinance institutions can eliminate consumer doubts (Kaabachi & Obeid, 2016). Several studies on the relevance of knowledge in determining attitudes in the context of Islamic microfinance. As research conducted by Wulandari (2019) which proves the importance of knowledge in determining positive attitudes towards Islamic microfinance. Likewise with the research conducted by Awn and Azam (2020) which proves that knowledge can influence attitudes. Thus the hypothesis proposed in this study is: H2: Knowledge has a positive influence on the attitude toward Islamic microfinance

Attitude and behavioural intention to adoption Islamic microfinance
Attitude is the extent to which a person has a favourable or unfavourable evaluation or judgment of a person (Ajzen, 1991). The belief influences this attitude in the consequences of the behaviour (Ajzen, 2005). Thus, if a person believes that the Islamic microfinance products and services offered are affordable, good services, depend on the product, are by religious beliefs and are useful for improving the quality of life, then his attitude towards Islamic microfinance will positively affect. Empirically several studies have proven that attitudes have a significant effect on intentions. As research conducted by Ngadiman et al. (2014) proves that attitude has a significant effect on the intention to use Islamic microfinance products. Similarly, research conducted by Umar et al. (2021), Kachkar & Djafri, 2021) & Abdullahi et al. (2021 also proves that attitude has a significant effect on intention. Thus the hypothesis proposed in this study is: H3: Attitude has a positive influence on the intention to adopt Islamic microfinance

Subjective norms and behavioural intention to adoption Islamic microfinance
Subjective norms are defined as individual perceptions of the possibility of a reference group in the form of groups or individuals agreeing or disapproving of a behaviour (Baber, 2018). Subjective norms are part of the belief that certain people are unlikely to approve or approve of certain behaviours (Hudi et al., 2019). Subjective norms refer to the individual's perception of the surrounding social pressure in doing or not doing a certain behaviour. In TPB norms, it is determined by a belief in normative, which assesses certain social pressures of individuals (Ibrahim et al., 2017). In the context of this research, norms are related to support from others to use Islamic microfinance products. The higher the support, the higher the intention to use Islamic microfinance products in Magelang Regency. Several studies have proven that subjective norms affect behavioural intentions to use Islamic microfinance products. Mukarromah and Widana (2021) research proves that subjective norms affect behavioural intentions. Research conducted by Umar et al. (2021), Maulana et al. (2018), and Abdullahi et al. (2021) conclude that norms can influence the goals of using Islamic micro-institution products. Thus the hypothesis proposed in this study is: H4: Subjective norm has a positive influence on the intention to adopt Islamic microfinance.

Perceived behavioural control and behavioural intention to adoption Islamic microfinance
Perceived behavioural control refers to the perceived ease or difficulty of performing the behaviour. In other words, individuals' self-confidence in performing certain tasks significantly affects their intentions and behaviour (Ajzen, 1991). When a person does not have sufficient resources or information about Islamic microfinance, their intention to adopt it is hampered. The results of previous studies prove that Perceived behavioural control has a significant influence on behavioural intentions to adopt Islamic microfinance (Abdullahi et al., 2021;Kachkar & Djafri, 2021;Maulana et al., 2018;Umar et al., 2021). Thus, the hypothesis proposed in this study is as follows: H5: Perceived behavioural control has a positive influence on the intention to adopt Islamic microfinance.

Mediating effect of attitude to behavioural intention to adoption Islamic microfinance
Attitude is the most influential factor in explaining a person's intention to use Islamic microfinance products. Thus this research has been developed by involving the variables of religiosity and knowledge in TPB as antecedents of attitude. Attitude in the TPB is one of the variables that influence behavioural intentions. Thus, there are variables of religiosity and knowledge that can affect intentions through attitudes. Bananuka et al. (2020) research proves that attitudes can mediate the relationship between religiosity and behavioural intentions to adopt Islamic banking. Similarly, a study conducted by Oladapo et al. (2019) can mediate the relationship between knowledge and behavioural intentions to adopt Islamic banking. From the previous research, the hypothesis proposed in this study is H1-a: Attitude will mediate the relationship between religiosity and behavioural intention to adoption Islamic microfinance H2-a: Attitude will mediate the relationship between knowledge and behavioural intention to adoption Islamic microfinance

Conceptual framework
This study uses the TPB framework developed by Ajzen (1991). TPB is a theory that explains how people want to perform a behaviour (Suleman et al., 2021). Our study incorporates three elements of TPB: subjective norms, attitudes, and perceived behavioural control, and adds two new variables to the TPB model: religiosity and knowledge. Figure 1 shows the proposed structure model, consisting of three types of variables. The independent variables consist of religiosity, knowledge, subjective norms, perceived behavioural control, one mediating variable (attitude), and one dependent variable (behavioural intention to adopt Islamic microfinance). Each component model is based on a literature review. Overall, this study contributes to the literature by focusing on the relationship between religiosity and knowledge in TPB and identifying mediating attitudes that impact behavioural intentions to adopt Islamic microfinance.
Similarly, attitudes, subjective norms and perceived behavioural control have the same role in the TPB model, namely the variables that determine behavioural intentions. However, this study does not make subjective norms and perceived behavioural control mediating variables between religiosity and knowledge to behavioural intentions. As explained earlier, subjective norms refer to the individual's perception of the surrounding social pressure to do or not do certain behaviours (Ajzen, 1991). Thus subjective norms are encouraged by others; therefore, it will be difficult to ascertain the relationship between religiosity and subjective norms. Abou-Youssef et al. (2015) explained that each individual has a different religious level. Perceived behavioural control refers to the perceived ease or difficulty in performing the behaviour. Perceptions of problem or difficulty performing the behaviour in adopting Islamic microfinance related to the facilities and resources owned. In other words, the perception of behavioural control is not determined by the level of religiosity and knowledge.

Research methodology
The population in this study is active members of 14 Islamic microfinance institutions in the Magelang Regency. The research sample was determined by statistical power criteria using G*Power with effect size = 0.15, = 0.05, power = 0.95 (Memon et al., 2020). The result is that a minimum of 138 research data are needed, but 262 samples were collected in this study. This number exceeds the minimum sample criteria in the SEM-PLS analysis. The sampling technique used in this research is a probability sampling technique. This study applies the Partial Least Squares Structural Equation Modelling (PLS-SEM) method to analysed the data, providing reliability and valid evidence. According to Henseler et al. (2016), Smart PLS is a statistical tool with good capabilities, even in studies with small samples.

Measurement (scaling and questionnaire research)
All variables in this study were measured using a Likert scale with five alternative answers, namely "Strongly disagree" (1) to "Strongly agree" (5). All statement items used in this study use items that have been used by previous researchers-survey questionnaire in Indonesian language. The first part of the questionnaire briefly describes the rationale for the study and instructions for filling out the survey questionnaire and socio-demographic information. Questions include respondents, marital status, gender, region, occupation, and education. The second part contains statements about research variables.
The religiosity variable was measured by adopting the research of Yeniaras and Akarsu (2017). There are five dimensions of religiosity, namely ideological dimensions, ritualistic dimensions, intellectual dimensions, consequential dimensions, and experiential dimensions. Two statement items represent each dimension, and then the religiosity variable is measured by ten statements. The ideological dimension is measured by saying, "I believe that there is no God but Allah (SWT)". An example of a statement from the ritual dimension is "I pray five times a day regularly", a statement on the intellectual dimension "I stay away from earning in a haram way". The consequential dimension of the statement used "I try to avoid activities that hurt others" and the experiential dimension used such as "I feel sad and dissatisfied when I do something that goes against my beliefs".
Based on previous studies, the knowledge was measured through the statement "I know that the Baitul Maal wat Tamwil (BMT) service does not charge interest" (Ezeh & Nkamnebe, 2021). The four items were used to measure the knowledge variable. Based on previous studies, the attitude toward Islamic microfinance was measured through the statement, "Using Baitul Maal wat Tamwil (BMT) services is profitable" (Maulana et al., 2018). The six items were used to measure the attitude toward the Islamic microfinance variable. Based on previous studies, the subjective norm was measured through the statement, "People important to me will think that I should use Baitul Maal wat Tamwil (BMT) services" (Maulana et al., 2018). The three items were used to measure the subjective norm. Based on previous studies, the perceived behavioural control was measured through the statement, "I am sure that I can easily use Baitul Maal wat Tamwil (BMT) services" (Maulana et al., 2018). The three items were used to measure perceived behavioural control. Meanwhile, based on previous studies, the behavioural intention to adopt Islamic microfinance was measured through the statement, "I intend to use Baitul Maal wat Tamwil (BMT) services in the next three years" (Aziz & Afaq, 2018). The six items were used to measure the intention to adopt Islamic microfinance.

Data analysis technique
The analysis was carried out using the partial least-squares structural equation modelling (PLS-SEM) technique using SmartPLS software version 3.2.7. SEM-PLS data analysis was carried out in two steps: the measurement model analysis and the structural model analysis. The analysis of the measurement model is basically a test of validity and reliability. The validity test in the SEM-PLS analysis was determined by convergent validity and discriminant validity. An indicator is said to meet the criteria of convergent validity if it has a factor loading value of more than 0.7, has a composite reliability (CR) value that exceeds 0.7 and has an average variance extract (AVE) value greater than 0.5 (Hair et al., 2011). Meanwhile, the validity test used discriminant validity consisting of cross-loading and Fornell-Larcker criteria. It the cross-loading criteria if each indicator has a higher value when compared to indicators on other variables (Chawla & Joshi, 2018). Standards the Fornell-Larcker Criteria if the correlation between variables with AVE squared is higher than the correlation of other variables in the research model (Chawla & Joshi, 2018). The reliability test in this study was measured by the Cronbach value. It is said to be reliable, if an indicator has a value greater than 0.6 (Gottems et al., 2018). Thus, before testing the hypothesis, a model must meet the criteria for a measurement model or test for validity and reliability. After the research model meets the model measurement criteria, the next step is to conduct structural analysis. Structural analysis in SEM-PLS must meet the criteria, namely having a statistical significance weighted P-value < 0.05 (the basis for drawing conclusions). Has an R2 value of 0.75 (substantial), 0.50 ( Table 1 demographic information for the people who participated in the study. Male and female respondents respectively made up 59.5% and 40.5% of the total respondents. The sample also includes respondents from different age groups and different sectors. The largest age group is respondents aged 35+ to 40 years, with the number of respondents reaching 46.18%. In terms of employment, the majority, 46.18%, work as entrepreneurs. Based on the level of education, the highest respondent has a high school education with a total of 68.70% and from the income aspect, the respondents involved in this study are of the opinion between IDR. 5,000,001 to IDR. 10,000,000 with the number of respondents reaching 46.18%.

Demographic profiles of respondents
This study also provides an overview of religiosity levels. Religiosity levels follow the categories used in the research of Wijaya et al. (2020) and Mahdzan et al. (2017), which consist of Devout (High religion practice), Moderate (Modest religion practice) and Casual (Low religion practice). Given that the grand mean score is 4.35, we categorized the respondents as Casual if their religiosity mean score was 0.5 standard deviations below the mean (μ ≤ 3.85). Respondents who had religiosity mean scores that fell between 3.85 and 4.85 were categorized as Moderate (3.85 < μ > 4.85), while those who had religiosity mean scores of above 3.85 were categorized as Devout (μ ≥ 4.85). Descriptive statistics of the religiosity levels are presented later in Table 2.
The measurement of the model in SEM-PLS can be assessed by looking at the loading factor, Cronbach alpha, Composite reliability, and average variance extract (AVE), and discriminant validity. The model results can be seen in Table 3. Table 3 shows the composite reliability (CR), and the loading factor exceeds the recommended value of 0.7 (Hair et al., 2011). The results showed that the average variance extract (AVE) was higher than the definite value of 0.5 by Joseph F Hair et al. (2019). These results show that the research model has an adequate measurement model. Table 4 shows that the AVE value for each latent construct is more than 0.50, resulting in good convergent validity (Joseph F Hair et al., 2019). According to the Fornell and Larcker Criteria, the square root of AVE is greater than the correlation between the respective constructs. As shown in the items in bold diagonally in Table 4, the square root of the AVE for all variables exceeds the intercorrelation, indicating sufficient discriminant validity (Joseph F Hair et al., 2019).

Structural model assessment
Structural model analysis and measurement in this study were carried out with the help of Smart PLS 3. The direct and indirect effects were analysed according to the research objectives, as shown  in Figure 1. This study has a direct hypothesis, as shown in Figure 1 and Table 5. All hypotheses (H1, H2, H3, H4, and H5) are accepted because they have a p-value smaller than 0.05. In addition, the PLS-SEM bootstrap was applied to examine the indirect effect or mediating effect (Hair et al., 2014). Table 6 shows the results of the indirect effect test. Two mediating effects (H1-a and H2-a) were accepted, because their significance value was less than 0.05. Therefore, attitudes towards Islamic microfinance are able to mediate the relationship between religiosity and knowledge on behavioural intentions to adopt Islamic microfinance.  contributed 0.097 or 9.7 percent to the attitude variable which was included in the weak category (Joseph F Hair et al., 2019). Meanwhile, the attitude variables, subjective norms and perceived behavioural control contributed to the behavioural intention to adopt Islamic microfinance by 0.320 or 32.0 percent, which was included in the moderate category (Joseph F Hair et al., 2019). In addition, Table 6 also shows the magnitude of the value of Q Square, which is the value used to assess the predictive model. The result is that the attitude variables, subjective norms and Perceived behavioural control contribute weakly to behavioural intentions to adopt Islamic microfinance, because they have a Q Square number less than 0.20 (Joseph F Hair et al., 2019).
Moreover, Table 8 reveals that the output of f is Square. The effect is strong when the f squared value is 0.35, 0.15 is moderate, and 0.02 reflects a small impact. In this study, the value of f square for the variables of religiosity and knowledge gave a small effect on attitudes. Meanwhile, the perceived attitude and behavioural control variables have a small effect on behavioural intentions to adopt Islamic microfinance. The subjective norm variable describes a moderate effect on behavioural intentions to adopt Islamic microfinance.

Research findings and discussion
The influence of religiosity and knowledge on attitudes towards Islamic microfinance shows a value of 3.140 and 3.602, respectively, with a P value less than 0.05. The resulting values for these two variables are 0.179 and 0.225, respectively. These results support the hypotheses H1 and H2, namely that there is a significant influence between religiosity and knowledge on attitudes towards Islamic microfinance. Therefore, increasing the level of religiosity and knowledge of Islamic microfinance will lead to a significant increase in attitudes towards Islamic microfinance.
The results of this study support the results of research that have been carried out by Souiden and Rani (2015), Purwanto (2021), andAbou-Youssef et al. (2015), which conclude that religiosity has a significant influence on attitudes. In addition, this research also supports research conducted by Awn and Azam (2020), which proves that knowledge can influence attitudes.
This study shows that religiosity and knowledge can determine attitudes towards Islamic microfinance. Someone with high religiosity has a positive attitude towards sharia microfinance because sharia microfinance institutions do not run their business with an interest system but profit sharing. The interest system is prohibited from being implemented, so religious people have a positive attitude towards sharia microfinance. In addition to the level of religiosity, attitudes in this study are influenced by knowledge. A person with good knowledge of Islamic microfinance can increase a positive attitude. A person who knows that Islamic microfinance is not profitable, can operate with Islamic sharia principles and so on will naturally have a positive attitude towards Islamic microfinance.
The attitude variable in this study proved to have a significant positive effect on behavioural intentions to adopt Islamic microfinance. Data analysis shows that the t values are 4.615 and 0.265. The results of this study support the attitude research conducted by Ngadiman et al. (2014), Umar et al. (2021), Kachkar & Djafri, 2021) & Abdullahi et al. (2021 which also prove that it has a significant effect on intention. Thus, if someone believes that the sharia microfinance products and services offered are affordable, good service, dependent on the product, religious beliefs and useful for improving the quality of life, then his attitude towards sharia microfinance will be positive. The results of this study reject the results of research conducted by Maulana et al. (2018), which did not find that attitudes can affect behavioral intentions to adopt Islamic microfinance.
Furthermore, subjective norms have a significant and positive effect on behavioural intentions to adopt Islamic microfinance. The results of the data analysis show the t-value of 6.548 and the tvalue of 0.436. These results indicate that subjective norms are the most important predictor of behavioural intention to adopt Islamic microfinance. This means that people attach great importance to social pressure to perform certain types of behaviour. This implies that friends, family members, and colleagues can influence behavioural intentions to adopt Islamic microfinance. Results such as research conducted by Mukarromah and Widana (2021), Umar et al. (2021), Maulana et al. (2018), and Abdullahi et al. (2021) also conclude that subjective norms can influence behavioural intentions to complete the adoption of Islamic microfinance. Thus, this  Maulana et al. (2018) and Kachkar and Djafri (2021) which previously proved that subjective norms do not affect behavioural intentions to adopt Islamic microfinance. The results of this study are by conditions in the field that Islamic microfinance members tend to use Islamic microfinance service products if they get information from trusted people, such as religious leaders or relatives who have previously had experience in using Islamic microfinance service products.
Furthermore, this study finds that perceived behavioural control has a significant and positive effect on behavioural intentions to adopt Islamic microfinance. The data analysis results show that the t value is 2.536 and 0.146. These results indicate that perceived behavioural control is also the most important predictor in identifying behavioural intentions to adopt Islamic microfinance. Perceived behavioural control refers to the perceived ease or difficulty of performing the behaviour. In other words, an individual's belief in performing a particular task significantly affects intention (Ajzen, 1991). When people are not equipped with sufficient resources or information regarding Islamic microfinance, their intention to adopt Islamic microfinance. These results support previous studies such as Maulana et al. (2018), Kachkar and Djafri (2021), Umar et al. (2021), and Abdullahi et al. (2021) which also prove that perceived behavioural control has a significant influence on behavioural intentions to adopt Islamic microfinance.
The mediating effect of religiosity and knowledge on behavioural intentions to adopt Islamic microfinance through attitudes. The result shows that the t value of each variable is 2.495 and 2.572, with the value of each variable being 0.047 and 0.060. These results indicate that the analysis proves that the hypotheses H1-a and H2-a are acceptable. This means that attitudes are able to mediate the relationship between religiosity and knowledge on behavioural intentions of Islamic microfinance adoption. Attitude is the most influential factor in explaining a person's intention to use Islamic microfinance products. Thus, this research was developed by involving the variables of religiosity and knowledge in TPB as variables that influence attitudes. Attitude in TPB is one of the variables that affect intention. Thus, the variables of religiosity and knowledge can influence intentions through attitudes. This study is in line with the research of Bananuka et al. (2020) which proves that attitudes are able to mediate the relationship between religiosity and behavioural intentions. This study is also in line with research conducted by Oladapo et al. (2019) where attitudes are able to mediate the relationship between knowledge and behavioural intentions.

Conclusion and managerial implications
The main objective of this study is to contribute to the literature by focusing on behavioural intention analysis to adopt Islamic microfinance by adding religiosity and knowledge variables in the TPB concept. This study found that attitudes could mediate the relationship between religiosity and knowledge on behavioural intentions to adopt Islamic microfinance. This study also confirms that religiosity and knowledge have an important role in forming positive attitudes towards behavioural intentions to adopt Islamic microfinance.
This research confirms that TPB concepts, namely attitudes, subjective norms, and perceived behavioural control, significantly influence behavioural intentions to adopt Islamic microfinance. The subjective norm is the variable with the highest effect size of the three variables. In other words, this study confirms that subjective norms play an important role in determining behavioural intentions to adopt Islamic microfinance. This study also shows that attitudes can be influenced by religion and knowledge. This shows that religiosity and knowledge have a significant influence in shaping attitudes towards Islamic microfinance. In addition, attitudes can ultimately mediate the relationship between religiosity and knowledge on behavioural intentions to adopt Islamic microfinance.
This study also offers convenience to Islamic microfinance managers in Magelang Regency and Indonesia in general. First, Islamic microfinance managers should create a specific community that can introduce the services offered by Islamic microfinance. Second, managers, especially the marketing department, must design policies in introducing Islamic microfinance service products to increase adequate knowledge about Islamic microfinance because knowledge can form a positive and ultimately impact behavioural attitudes to adopt Islamic microfinance. Third, this research can be useful for Islamic microfinance managers in formulating marketing policies to attract new prospective members through religious activities and involving religious leaders in introducing Islamic microfinance.

Limitations and future research directions
This study has limitations, such as the number of samples that are not representative of the number of Islamic microfinance members in the Magelang Regency but only meet the minimum requirements in the SEM-PLS analysis. Furthermore, the respondents in this study were all members of Islamic microfinance without restrictions on how long they had been members of Islamic microfinance. Future studies can also add more demographic characteristics such as age, length of membership, and income to see how behavioural intentions to adopt Islamic microfinance are.