Recently, researchers in mathematics education have shown interest in mathematical creativity (Leikin & Pitta-Pantazi, 2013)(Ngiamsunthorn, 2020)(, 2020)(Sriraman et al., 2013). This is because of different factors such as the call from researchers in mathematics education (e.g., (Haylock, 1987). Another factor is the role of creativity in the 21st century particularly through its interconnectivity with other contemporary skills (, 2015). Specifically, problemsolving and posing require creative ideas, as expressed in the emergence of new and intriguing solutions and problems that extend known ones and lead to new mathematical phenomena.

Researchers have been paying attention to creativity in the classroom as it impacts the different aspects of students' learning in general, and mathematics education in particular (Daher, 2011). Despite various researchers' attempts to define mathematical creativity, no proposed definition has been universally accepted (Mann, 2006). Newell et al. (1962) state that problem-solving becomes creative when one or more of the conditions are satisfied. These conditions include the following: (1) the product has novelty and value; (2) the thinking requires modification or rejection of previous ideas; (3) the thinking is done under the conditions of high motivation and persistence, and it takes place over either a considerable time or at high intensity; (4) the problem as initially posed is vague or ill-defined. The definitions of creativity can be divided into two categories: definitions related to the final product and those related to the process (Daher et al., 2018)James, Lederman, Gerard, & Vagt-Traore, 2004). Sternberg and Lubart (1999), who adopted the end-product perspective, describe creativity as the ability to produce an unexpected work. Researchers who define creativity as a process describe it as the ability to think in a conceptual way. In the present study, we adopt the definition of creativity as involving three components: fluency, flexibility, and originality (Guilford, 1975)(Torrance, 1966). Our adoption of these components for studying the creativity of pre-service teachers in an educational context is supported by Silver (1997), who describes the abovementioned three components with regard to studying creativity. This description is emphasized again by (Sriraman et al., 2013) who state that varying combinations of the three components are utilized to study mathematical creativity. In the present study, fluency refers to the number of correct responses that a student provides for a problem.

Researchers are interested in factors that could influence students’ creativity. Three such factors are emotions in creative thinking (Soroa et al., 2015), self-efficacy in creative thinking (, 2011), and curiosity (Rinkevich, 2014). The present study is interested in predicting creative emotions through self-efficacy in creativity and curiosity. Next, we describe these three constructs.

Researchers claim that human activity is not emotionally neutral. They point to the relationship between emotions and different aspects of student learning-the cognitive aspect (Daher, 2011)(Understanding emotions in mathematical thinking and learning, 2017), the social aspect (Daher, 2015)(May & Fray, 2010), and the metacognitive aspect (Daher et al., 2018)(, 2014). (Radford, 2004) discusses the role of emotions in information processing in general: "Information is not emotionally neutral. In the intelligent selection and processing of information we are guided by subconsciously assimilated emotional markers that are a result of acculturation." We argue that these emotions could also be conscious, affecting the processing. This highlights the important role of emotions-in particular, creative emotions-in knowledge construction and problem-solving.

(, 2008) propose a model in which positive emotions facilitate creativity by leading people to feel less constrained and experience the situation as unproblematic, which would enable their cognitive flexibility. However, negative emotions facilitate perseverance and effort toward generating effective solutions to a problematic situation. (Soroa et al., 2015) build on this model to stress that positive emotions encourage more expansive divergent thinking and novelty. In the present study, we examine what could predict students' emotions in creativity, rather than examining emotions as a predictors of other constructs.

Self-efficacy is described as being concerned with the beliefs in what one can do with whatever resources one can has or can obtain (Bandura, 2007). Regarding selfefficacy in creativity, (Tierney & Farmer, 2011) state that it is the confidence of individuals in their own creative capabilities to produce unusual outcomes or solutions. This confidence positively affects their expression of creativity (Yu, 2013). Individuals who feel creative self-efficacy tend to engage in broader searches for information (Tierney & Farmer, 2002) and are more likely to access relevant knowledge and expertise from other individuals or from the group, which supports them in developing new and useful ideas, that is, creative ones. (Abbott, 2010) suggests looking at two dimensions of creative self-efficacy: creative thinking self-efficacy and creative performance self-efficacy. The present research focuses on the creative thinking component, which is an individual's belief in their own ability to express creative thinking.

(Abbott, 2010) suggests looking at two dimensions of creative self-efficacy: creative thinking self-efficacy and creative performance self-efficacy. The present research focuses on the creative thinking component, which is an individual's belief in their own ability to express creative thinking.

(Kashdan et al., 2009) describe curiosity as having overlapping attributes with intrinsic motivation, flow, and other variables. They emphasize that it possesses unique characteristics, such as being interested in new things and having an open and receptive attitude toward whatever is the object of attention. Thus, when people feel curious, they dedicate more attention to the object of their curiosity, which encourages them to persist with this object until goals are accomplished (Silvia, 2006). The educational function of curiosity lies in its encouragement of the curious student to learn and explore the object of their curiosity (Loewenstein, 1994).

(Kashdan et al., 2009) argue that curiosity is related to two aspects. First, it could be expressed as the willingness to embrace the unpredictable nature of everyday life. This is related to the embracing component of curiosity. Second, it could be expressed as the willingness to continue to accumulate new abilities and experiences, which is related to people’s inclination to stretch their capabilities. This is related to the stretching component of curiosity.

In the present research, we aim to examine students' curiosity and self-efficacy in mathematical creativity as predictors of student's creative emotions in mathematical problem-solving. Furthermore, we aim to examine which set of the five independent variables best predicts creative emotions in mathematical problem-solving. Few studies have focused on the antecedents of creative emotions in mathematical problem-solving, which is the aim of the present study. We use theories of curiosity and creative selfefficacy that are not specific to mathematics education, creating an opportunity to use them in mathematics education.

Existing research suggests that emotions play a considerable role in the creative problem-solving process (Sas & Zhang, 2010). This indicates the need to cultivate positive emotions for problem-solving so that creative problem-solving occurs. To cultivate these emotions, we need to know the antecedents of emotions. The present study attempts to do so considering two antecedents of creativity: self-efficacy in creativity and curiosity.

Although much research has examined emotions and their impact on creativity (Sas & Zhang, 2010), few studies have focused on self-efficacy and curiosity as factors influencing creative emotions in mathematical problem-solving, which is the focus of this study. This will provide educators with means to encourage creative emotions, which in turn will foster students' creativity in mathematical problem-solving and help prepare them for 21st-century skills (Daher et al., 2017).

Previous research has shown that self-efficacy positively affects students' creativity. For example, (, 2016) found that self-efficacy had a significantly positive effect on the creativity of students. In addition, curiosity was also related to creativity. (Starko, 2013) argues that a strong sense of curiosity is important in the early steps of the creative process. The present research was concerned with constructs related to self-efficacy and creativity: creative self-efficacy and creative emotions. We intend to verify whether the relations previously described are also valid for the constructs in the present research.

The sampling method was stratified convenience sampling. We distributed the questionnaire among Grades 8-10 students from schools in Nablus Governorate in Palestine. The schools were chosen based on their consent to distribute the questionnaire among their students. The sample comprised 500 Grades 8-10 Palestinian students (251 female students and 249 male students) aged 13-16 years. The participants were 156 Grade 8 students (31.2%), 182 Grade 9 students (36.4%), and 162 Grade 10 students (32.4%). The sampling design employed was nonprobability convenience sampling.

This study used three self-report questionnaires that served to measure the research constructs: creative emotions, self-efficacy in creativity, and curiosity. The emotion scale was adopted from (Soroa et al., 2015)-that is, the "divergent pleasant style" scale. This scale includes five items, such as "While suggesting innovative solutions for a mathematical problem, I tend to feel joy." The self-efficacy in creativity scale was adopted from (Abbott, 2010). It comprises 12 items that examine self-efficacy constructs: 4 self-efficacy in fluency items (e.g., "I can get a large number of different ideas or responses for a mathematical problem"), 4 self-efficacy in flexibility items (e.g., "I can think of many types of ideas while considering the solution of a mathematical problem"), and 4 self-efficacy in originality items (e.g., "I can think of ideas for the solution of a mathematical problem that no one else has"). The curiosity scale was adopted from (Kashdan et al., 2009). It comprises 10 items that examine 2 curiosity constructs: 5 stretching curiosity items (e.g., "I view challenging situations as an opportunity to grow and learn") and 5 embracing curiosity items (e.g., "I am the type of person who really enjoys the uncertainty of everyday life"). In the three scales, each item has 5 choices: 1: I strongly disagree, 2: I disagree, 3: I am not sure, 4: I agree, and 5: I strongly agree.

Validity and reliability analyses were performed for the six scales. To ensure face validity, the Arabic version of the questionnaire was presented to mathematics educators and teachers to give their opinion regarding the accuracy of items formulations as related to Grades 8-10.

To answer the research questions, we first computed the Pearson correlation coefficient between each of the two predictors in the present study (self-efficacy and curiosity) and the outcome variable (emotions). To determine the strength of the correlation between two variables, we relied on Cohen (1988): 0.1 < | r | < 0.3 indicates small correlation, 0.3 < | r | < 0.5 indicates medium/moderate correlation and | r | > 0.5 indicates large/strong correlation, where r represents Pearson correlation coefficient.

To answer the research questions, we first computed the Pearson correlation coefficient between each of the two predictors in the present study (self-efficacy and curiosity) and the outcome variable (emotions). To determine the strength of the correlation between two variables, we relied on (Cohen, 1988) To determine the percentage of variance explained by the predictor in the outcome variable, as well as the regression weights of the predictors in determining the outcome variable, multiple regression analysis was performed. In addition, to examine the set of variables from the five independent variables that could best predict creative emotions, we conducted stepwise regression analysis. This type of regression fits our second goal of examining the set of variables that could best predict creative emotions. Our utilization of stepwise regressions follows the claim of researchers that these methods are used in educational and psychological research to assess the order of importance of predicting variables and to select useful subsets of these variables (Huberty, 1989; Thompson, 1995).

The collected data satisfied the assumptions of regression analysis (Grande, 2015). First, the dependent variable—creative emotions—is measured on a continuous scale. Second, there are two or more independent variables, which can be either continuous or categorical. Here, we have five independent ratio variables (i.e., self-efficacy in fluency, self-efficacy in flexibility, self-efficacy in originality, stretching curiosity, and embracing curiosity). Third, there should be independence of observations (i.e., independence of

residuals). To examine this independence, we used the Durbin-Watson statistic and obtained a value of 1.945; this statistical value is acceptable as it lies between 1.5 and 2.5 (Lester, Inman, & Bishop, 2014). In addition, the standardized residuals lie in a rectangular shape, emphasizing the independence of observations, as shown in Figure 1 (Grande, 2015).

Fourth, there needs to be a linear relationship between (a) the dependent variable and each of the independent variables and (b) the dependent variable and independent variables collectively. To examine this assumption, we drew the normal probability plot of the regression standardized residual (Figure 2).

Fifth, the plot in Figure 2 is more-or-less around the line, indicating that the relationship between the predictors and the outcome variable is linear. Being around the line, the plot also indicates that the residuals follow a normal distribution, which is the sixth assumption of regression.

Seventh, the data must not show multicollinearity, which occurs when two or more independent variables are highly correlated (i.e., greater than 0.9) (Dohoo et al., 1997). The correlations between the independent variables were all below 0.9, as shown in Table 1, confirming the absence of multicollinearity.

Eighth, the predictor variables should correlate with the outcome variable at a value greater than 0.3, which is satisfied in our case, except in the case of embracing curiosity, as shown in Table 1. We retained the embracing curiosity construct but did not expect it to contribute meaningfully to explaining the variance in creative emotions.

To answer the research questions, regarding the prediction of creative emotions from self-efficacy in creativity and curiosity constructs, we first computed the Pearson correlations between each pair of research variables. Table 1 shows these correlations, along with the variables’ means and standard deviations.

EmoC=Emotions toward creativity, SE-Fl= Self-efficacy in Fluency, SE-Flx= Self-efficacy in Flexibility, Self-efficacy in originality, St-C=stretching construct of curiosity, Emb-C=Embracing construct of curiosity

Self-efficacy in fluency, flexibility, and originality as predictors of creative emotions

The first question asks whether students’ self-efficacy in fluency, flexibility, and originality significantly predict their creative emotions.Table 1 shows that creative emotions had a moderate correlation with each self-efficacy construct. Additionally, the correlations between self-efficacy constructs (the predictors) are approximately r = 0.5, suggesting that multiple regression is appropriate. A multiple regression analysis was conducted to predict students’ creative emotions from self-efficacy in fluency, flexibility, and originality. The three self-efficacy constructs were entered in the first step of the regression. The regression result indicated that all three components significantly predicted creative emotions (p<0.01). Moreover, the three self-efficacy constructs explained 29.6% of the variance (R2=0.30, F(3,496)=70.99, p<0.001). The resulting regression equation was: creative emotions = 1.40 + 0.14*self-efficacy in fluency + 0.16*self-efficacy in flexibility + 0.30*self-efficacy in originality. The contribution of construct of self-efficacy in creativity was significant as follows: t(496)=3.03, p<0.01 for self-efficacy in fluency, t(496)=7.72, p<0.01 for self-efficacy in flexibility, and t(496)=3.87, p<0.01 for self-efficacy in originality.

Stretching and embracing curiosity as predictors of creative emotions

The second research question asks whether students’ curiosity related to its two constructs predicts their creative emotions. We performed the same computations as in the previous analysis. Table 1 shows that creative emotions had a moderate correlation with the stretching component and a weak correlation with the embracing component. Additionally, the correlations between predictors are approximately r = 0.5, suggesting that multiple regression is appropriate with the curiosity components, except for the embracing component due to its weak correlation with creative emotions. To examine the prediction of creative emotions by the two curiosity constructs, multiple regression analysis was performed with both constructs entered in the first step. The results showed that the two curiosity constructs significantly predicted emotions toward creativity (p<0.01). In addition, the two curiosity components, as predictors, explained 17.8% of the variance (R2=0.18, F(2,497)=54.90, p<0.001). The appropriate regression equation turned out as follows: Emotions toward creativity = 1.74 + 0.44*Stretching curiosity + 0.04*Embracing curiosity. The contribution of the stretching component was significant (t(497)=9.91, p<0.01), whereas the embracing component did not contribute significantly (t(497)=0.86, p=0.39). Embracing curiosity was retained in the equation because the overall model was significant.

The third research question asked which combination of the three self-efficacy components and two curiosity components ( five independent variables) best predicted students’ creative emotions. We ran a stepwise analysis, entering all five independent variables as predictors of creative emotions. The final prediction model included three of the five predictors and was reached in three steps, with self-efficacy in fluency and embracing removed (see Table 2). The model was statistically significant, F(3, 496) = 82.55, p < 0.001, and explained approximately 32.9% of the variance of creative emotions (R2 = 0.33, Adjusted R2 = 0.33). Creative emotions were primarily predicted by higher levels of self-efficacy in originality and stretching curiosity, and to a lesser extent, by self-efficacy in flexibility.

The model results are shown in Table 3. Self-efficacy in originality received the strongest weight in the model, followed by stretching curiosity and, subsequently, through self-efficacy in flexibility.