Serviços customizados
Inglês (pdf)
Artigo em XML
Referências do artigo
Tradução automática
Enviar este artigo por email
Citado por SciELO 
Citado por Google 
Similares em
SciELO 
Similares em Google 
Permalink
Commonly defined as “tasks assigned to students by school teachers that are meant to be carried out during non-school hours” (Cooper, 1989, p. 7 ), homework is a widespread instructional activity across countries (Fan et al., 2017; Fernández-Alonso et al., 2015, 2022; Warton, 2001). It is an important aspect of daily life for many school-age children (Cooper et al., 2006; Corno & Xu, 2004; Dettmers et al., 2011; Fan et al., 2017), as they are often assigned homework for nearly every class they take (Corno, 2011). Hence, it seems obvious and logical that the value of homework perceived by children has important implications for homework practice and research (Rodríguez et al., 2020; Warton, 2001; Xu, 2020; Xu & Corno, 2022), as the value individuals attach to the outcomes of an activity exerts powerful influences on their effort, persistence, and achievement (Wigfield et al., 2015). According to expectancy-value theory (Eccles & Wigfield, 2002), value is defined as the degree to which an individual perceives a task is useful and worthwhile; it centers on the question “Do I Want to Do This Task?” (Wigfield et al. 2015, p. 659). An individual who is convinced that a task is useful and worthwhile is more likely to exert effort and to be successful in the task (Rodríguez et al., 2019). Thus, according to expectancy-value theory, the value children attach to homework is “critical for . . . the effort they will contribute to the endeavor and to the persistence they will display” (Warton, 2001, p. 157).
Extant literature often focuses on the value of homework from the perspectives of adults (e.g., the public, educators, and parents; Bempechat, 2004; Rosário et al., 2019a,b; Suárez et al., 2022; Van Voorhis, 2004; Sun et al., 2020a). Over the last 100 years, homework is a perennial topic of the public debate; its value, for example, has been linked to concerns about the U.S.'s ability to compete in a global economy (Gill & Schlossman, 2004). The value of homework has been linked to homework purposes perceived to be important by parents, teachers, administrators, and researchers (Cooper, 1989; Van Voorhis, 2004; Sun et al., 2020b). Yet, many homework purposes advocated by adults (e.g., public relations and parent-teacher communications) matter little to children (Warton, 2001; Xu, 2005, 2023).
Two notable exceptions to the lack of attention to the student viewpoint consist of one study with children in grade 3 (Xu & Corno, 1998) and another study with children in grades 6-8 (Xu & Yuan, 2003). In both studies, parents and teachers mentioned two purposes for homework: academic (to help children better understand the materials covered in class), self-regulatory (to help children develop self-regulatory capacities such as study skills and personal responsibility). While children in the above two studies agreed one purpose with parents and teachers (academic), they listed another purpose that was important from their perspectives - approval-seeking (to please parents and teachers, and to comply with adult expectations). Furthermore, different from 3rd children (Xu & Corno, 1998), certain children in grades 6-8 (Xu & Yuan, 2003) identified another purpose - self-regulatory (e.g., “it [doing homework] makes you more responsible and independent”).
Based on expectancy-value theory (Eccles & Wigfield, 2002) and extant literature pertaining to homework purpose (e.g., Xu & Corno, 1998; Xu & Yuan, 2003), two recent studies have validated math homework purposes perceived by students (Sun et al., 2020a, 2020b). Sun and her colleagues examined the validity of the Math Homework Purpose Scale (MHPS) based on 585 7th graders (Sun et al., 2020a) and 854 9th graders (Sun et al., 2020b). The results of these studies showed that the MHPS consisted of three subscales: academic, self-regulatory, and approval-seeking. Furthermore, concurrent validity evidence from both studies revealed that academic and self-regulatory purposes were positively associated with math homework effort, completion, and achievement.
Nevertheless, although previous studies using a variable-centered approach provides insights into the direct links of each homework purpose reported by students with other important constructs (e.g., homework behavior and academic achievement), it overlooks or ignores the possibility that (a) children are likely to have multiple purposes for doing homework at the same time, (b) distinct constellations of homework purposes may coexist in the population, and (c) these distinct constellations might relate to differences in other constructs (e.g., homework completion).
To our knowledge, there is only one study that was interested in the analysis of these two questions (Xu, 2023). The goal of this investigation was to identify profiles of students drawn from three purposes of homework: academic, self-regulatory, and approval-seeking. A total of 750 eleventh-grade students in China participated in the study. The results of the latent profile analysis showed a solution of four different groups, or profiles, of students: very low (very low in the three purposes; 5.73%), low (low in the three purposes; 30.40%), medium (moderate in the three purposes; 54.40%), and high (high in the three purposes; 9.47%). The profile membership was significantly related to effort and task completion (with a medium effect size): in general, the higher the purposes, the greater the effort and task completion.
Using a person-centered approach to identify the profiles of homework purposes, this study extends extant literature on homework purpose. These results provide a deeper understanding of how these three homework purposes coexist within students of eleven grade. For example, though the relation between the three homework purposes appeared to be very strong, it was observed that while academic and self-regulation purposes are presented at the same level in the four homework profiles, approval-seeking purpose appear somewhat lower or higher (depending on the profile). These results were interpreted by Xu as partially consistent with the hypothesis that as both academic purpose and self-regulatory purpose reflect self-focused motives, but approval-seeking purpose represent other-focused motive.
Given that there are differences in the relevance of purposes for doing homework in eleventh graders, it is possible that such differences are even greater in younger students. So, as Xu (2023) suggests, it would be beneficial to pursue this line of research involving elementary and middle school students. So, the purpose of the current study is to expand the knowledge provided by Xu (2023), specifically (i) to identify homework purpose profiles in a sample of middle school students (8th graders), regarding the possible combinations of academic, self-regulatory, and approval-seeking purposes, and (ii) to see if they differ in homework effort, completion, and math achievement.
Taking into account the data derived from the study by Xu (2023), and that motivational patterns could already be well developed at 11-12 years (Montero et al., 2001), we expect that the same four homework profiles obtained with eleventh grade students will also be identified in eighth grade students (although the level of purposes within each profile may vary significantly). On the other hand, in line with theoretical expectations (e.g., task value; Wigfield et al., 2015) and related homework literature using variable-centered approach (Epstein & Van Voorhis, 2012; Sun et al., 2020a, 2020b; Xu, 2005), we expect that, in general, the students with a high level of homework purposes would expend more homework effort, complete more homework, and score higher on math achievement.
Method
Participants
Participants were 3,018 8th graders (96 classes; 45.6% female; 100% Han nationality). They came from three different areas in China, including central, southeastern, and southwestern. To reflect a wide range of socioeconomic backgrounds, students were sampled from eight regular public schools, which were randomly selected from nineteen regular public schools allowing us access for our data collection. The mean age for participants was 13.7±0.4 years. Education level was 10.6 years for mothers and 11.4 years for fathers. Regarding math homework practices, 76.9% participants worked on math assignments four or more days a week. They spent a mean of 34 minutes (SD = 25) on math assignments daily. These math homework practices are generally consistent with recent research in China (Xu et al., 2017).
Measures
Math homework purposes . This scale consisted of academic, self-regulatory, and approval-seeking purposes (Sun et al., 2020a, 2020b). Four items measured academic purpose, concerning reinforcement of school learning (e.g., “Doing math homework helps me understand what is going on in class”). Three items measured self-regulatory purpose, concerning promoting desirable self-regulatory attributes (e.g., “Doing math homework helps me learn to work independently”). Three items measured approval-seeking purpose, regarding seeking approvals from teachers, peers, and parents (e.g., “Doing math homework brings me family approval”). Response options for all ten items varied from 1 (strongly disagree) to 4 (strongly agree). In their study with 7th graders, Sun et al. (2020b) reported that math homework purposes consisted of academic purpose (α = .71), self-regulatory purpose (α = .76), and approval-seeking purpose (α = .85). Likewise, in our current investigation with 8th graders, math homework purposes contained academic purpose (α = .76; ω = .76), self-regulatory purpose (α = .85; ω = .85), and approval-seeking purpose (α = .89; ω = .89).
Math homework effort. Three items assessed students' math homework effort, informed by extant literature (Flunger et al., 2017; Xu et al., 2018; Xu, 2020). These items tapped into their initiatives to follow through math assignments (e.g., “I always try to finish my math assignments”; α = .81; ω = .82). Response options ranged from 1 (strongly disagree) to 4 (strongly agree).
Math homework completion. Students responded to one statement regarding homework completion, drawn from relevant studies (Cooper et al., 2006; Yang & Xu, 2018). It asked: “Some students often complete math homework on time, others rarely do. How much of your assigned math homework do you usually complete?” Ratings contained 1 (none), 2 (some), 3 (about half), 4 (most), and 5 (all). This item has been found to give valid information regarding homework completion. For instance, Xu (2017) found that, in line with theoretical predictions, it was positively related to homework expectancy, value, effort, and achievement.
Math achievement. Standardized math achievement was assessed nearly eight months following the administration of the measures (as discussed above). The assessment was aligned with national curriculum (Ni et al., 2011) to assess skills and knowledge (e.g., fraction, axial symmetry, linear function, parallelogram, parallelogram, quadratic radical, triangle, and data analysis). It consisted of short-answer and multiple-choice items, and students were allowed to 120 minutes to complete the test. The reliability estimate was α = .88.
Parent education. Students were asked, “What is the highest level of education completed by your father/guardian?” and “What is the highest level of education completed by your mother/guardian?” Responses were coded: elementary school (6), middle school (9), high school (12), some college (14), college graduate (16), some graduate school (18), and graduate degree (19). As parent education for father and mother were highly related (r = 0.76, p < 0.001), a variable labeled “parent education” was developed by averaging father's education and mother's education.
Prior achievement. We obtained students' grades in math from teachers' school logs at the end of the previous year (grade 7) to measure prior math knowledge. The grades were based on a 5-point letter system, varying from F (fail) to A (excellent). Specifically, they were coded as F (1 point), D (2 points), C (3 points), B (4 points), and A (5 points).
Procedure
We sought and gained approval from families for children to participate in our present study. Several research assistants administered the measures during a typical class, and math teachers were requested to step out of the classroom during the administration. Taken together, the participation rate was close to 90% (88.7%).
Data Analyses
LPA was used to identify underlying latent subgroups of students within the dataset based on academic, self-regulatory, and approval-seeking purposes. All analyses were carried out with robust maximum likelihood estimator in Mplus 8.8, which corrects for non-normality in the measures (Muthén & Muthén, 1998-2012). Because 3018 participants were nested in 96 classes, a design-based correction of standard errors was carried out using analysis code the “type is complex” in Mplus.
Our decision for selecting the optimal number of profiles was based on a combination of fit indices, parsimony, latent profile separation, and interpretability (Flunger et al., 2015; Hickendorff et al., 2018; Nylund et al., 2007). These include Akaike information criterion (AIC), Bayesian information criterion (BIC), sample-adjusted Bayesian information criterion (SSA-BIC), Lo-Mendell-Rubin adjusted likelihood ratio test (LMRT), sample size for each profile, entropy value, and the interpretability of the solutions based on substantive theory or theoretical predication. In general, the solution with smaller AIC, BIC, and SSA-BIC indicates better relative fit. A significant LMRT test indicates that a K profile model fits significantly better in comparison with a K-1 profile model. Profiles that include less than 5% of the sample are viewed unsuitable and not feasible, reflecting excessive profiling extraction (Wolter et al., 2019). Entropy value (from 0 to 1) is used to determine the classification accuracy of the solution (> 0.80 reflecting high separation among profiles; Ullrich-French & Cox, 2020). As the final step to investigate the appropriateness of the solution, we carried out three analyses of variance (ANOVAs) to examine whether there were statistically significant differences among the profiles on each of the measures included in the LPA (academic, self-regulatory, and approval-seeking purposes).
After identifying profiles as a function of three homework purposes, we tested the validity of the classification derived from the LPA, by examining differences across the profile memberships in three external measures of homework effort, completion, and achievement. Taking into account the relevance of socio-family factors as well as the previous performance of the students in school involvement and current and future performance of the students, the education of the parents and the previous performance of the students were included as covariates in this study. This was accomplished by employing the auxiliary variable option in Mplus (Asparouhov & Muthén, 2021). To interpret the effect sizes, we applied the following guidelines (Cohen, 1988), considering η2 = 0.01 (d = 0.20), η2 = 0.059 (d = 0.50), and η2 = 0.138 (d = 0.80) as representing a small, medium, and large effect size.
Results
Table 1 presents descriptive statistics of all measures (i.e., means, standard deviations, skew, and kurtosis). Additionally, it includes Pearson correlations among all measures; all of them were found to be significantly, positively correlated.
Table 1. Descriptive Statistics and Pearson Correlations Among the Measures
| Variable | M | SD | Skewness | Kurtosis | 1 | 2 | 3 | 4 | 5 |
|---|---|---|---|---|---|---|---|---|---|
| 1 Academic purpose | 2.97 | 0.56 | -0.48 | 1.34 | ---- | ||||
| 2 Self-regulatory purpose | 2.79 | 0.67 | -0.26 | 0.31 | .60** | ---- | |||
| 3 Approval-seeking purpose | 2.54 | 0.73 | -0.05 | -0.13 | .49** | .52** | ---- | ||
| 4 Homework effort | 3.19 | 0.60 | -0.74 | 1.34 | .27** | .22** | .14** | ---- | |
| 5 Homework completion | 3.97 | 0.95 | -0.87 | 0.38 | .26** | .21** | .13** | .22** | ---- |
| 6 Math achievement | 63.33 | 26.43 | -0.81 | -0.20 | .25** | .14** | .07** | .24** | .21** |
N = 3018. **p < .01.
Identification of Profiles
The fit of several latent profile models was examined (see Table 2), which was halted in five classes. First, according to the following indices (i.e., AIC, BIC, SSA-BIC, LMRT, and entropy), the four-profile solution yielded a better fit as compared with the two-profile solution and the three-profile solution. Although the four-profile model included one profile less than 5% of the cases (profile 3, n = 119, 3.94%), this profile presented rather distinctive information regarding three homework purposes (i.e., students in this profile had standardized scores on all three homework purposes well over one standard deviation below the mean; see Table 5 and Figure 1). Second, although the five-profile solution yielded somewhat better fit indices than the four-profile solution, it included two profiles less than 5% of the sample (profile 1, n = 119, 3.94%; n = 119; profile 4, 3.94%; see Table 2). In addition, the five-profile solution did not provide better entropy value (0.943) compared with the four-profile model (0.949). With respect to the classification accuracy of the four-profile solution, as displayed in Table 2, the entropy for the solution was 0.949, thus reaching a level of entropy that is viewed as high (e.g., 0.800).
Table 2. Fit Indices for Identifying Latent Profiles (n = 3018)
| Profiles of Math Homework Purposes | ||||
|---|---|---|---|---|
| 2 | 3 | 4 | 5 | |
| AIC | 16514.195 | 15280.857 | 14421.993 | 14151.994 |
| BIC | 16574.318 | 15365.030 | 14530.216 | 14284.266 |
| SSA-BIC | 16542.544 | 15320.547 | 14473.023 | 14214.363 |
| Entropy | .832 | .889 | .949 | .943 |
| LMPT | 1262.015*** | 1203.777*** | 840.635*** | 269.588*** |
| n in each profile | P1 = 2611 P2 = 407 | P1 = 224 P2 = 370 P3 = 2424 | P1 = 786 P2 = 404 P3 = 119 P4 = 1709 | P1 = 119 P2 = 786 P3 = 1702 P4 = 119 P5 = 292 |
| Number of profiles with n ≤ 5% | 0 | 0 | 1 | 2 |
Note: AIC = Akaike's Informational Criterion; BIC = Bayesian Information Criterion; SSA-BIC = Sample-Size Adjusted BIC; LMRT = Lo-Mendell-Rubin adjusted maximum likelihood ratio test. Tech14 option (Parametric bootstrapped likelihood ratio test; BLRT) is not available for the clustering option in Mplus (i.e., TYPE = MIXTURE COMPLEX).
***p < .001.
Furthermore, the findings of the ANOVAs indicated statistically significant differences between the four profiles in the three criterion variables: academic purpose (F [3, 3014] = 597.689; p < .001; η2 = .373; d = 1.54); self-regulatory purpose (F [3, 3014] = 10990.678; p < .001; η2 = .916; d = 6.60); and approval-seeking purpose (F [3, 3014] = 327.941; p < .001; η2 = .246; d = 1.14). The effect size was very large across these three criterion variables (especially for self-regulatory purpose). The results of Scheffé post hoc tests indicated that all four profiles differed significantly from each other on each of the homework purposes, thereby providing further support for the distinctiveness of these homework purpose profiles.
Hence, taking into account the fit indices, sample size for each profile, and the interpretability, the findings of the ANOVAs examining the contribution of the three criterion variables that made up the profiles to the differentiation among profiles, the four-class model seemed to be the optimal solution for our present study.
Description of the Four Profiles
Table 3 displays the mean scores of participants belonging to the four latent profiles. Profile 1 contained 26.04% of the sample (n = 786) and was referred to Low Profile because of their low mean scores on each homework purpose (z = −0.43 to −0.98; see Figure 1). Profile 2 consisted of 13.39% of the sample (n = 404) and was referred to High Profile because of their high mean scores on each homework purpose, with standardized scores about one standard deviation above the means (z = 0.96 to 1.68; see Figure 1). Profile 3 included 3.94% of the sample (n = 119) and was referred to Very Low Profile because of their very low mean scores on each homework purpose (z = −1.40 to −2.49; see Figure 1). Profile 4 was made of a large group of students (56.63%; n = 1702) and was referred to Moderate Profile because their scores on each homework purpose were close to the means (z = 0.03 to 0.23; see Figure 1).
Table 3. Description of Latent Profiles
| Confidence Intervals | ||||
|---|---|---|---|---|
| M | SE | Lower 5% | Higher 5% | |
| Profile 1: Low (n = 786) | ||||
| Academic | 2.73 | 0.03 | 2.69 | 2.77 |
| Self-regulatory | 2.13 | 0.01 | 2.11 | 2.15 |
| Approval-seeking | 2.21 | 0.03 | 2.16 | 2.26 |
| Profile 2: High (n = 404) | ||||
| Academic | 3.66 | 0.03 | 3.61 | 3.71 |
| Self-regulatory | 3.91 | 0.01 | 3.89 | 3.93 |
| Approval-seeking | 3.23 | 0.06 | 3.14 | 3.33 |
| Profile 3: Very Low (n = 119) | ||||
| Academic | 1.98 | 0.10 | 1.81 | 2.15 |
| Self-regulatory | 1.12 | 0.02 | 1.08 | 1.16 |
| Approval-seeking | 1.53 | 0.07 | 1.41 | 1.64 |
| Profile 4: Moderate (n = 1709) | ||||
| Academic | 2.98 | 0.01 | 2.96 | 3.01 |
| Self-regulatory | 2.94 | 0.01 | 2.93 | 2.95 |
| Approval-seeking | 2.60 | 0.02 | 2.56 | 2.63 |
Profile Membership Relations to External Variables of Homework Effort, Completion, and Achievement
The equality of the means of external variables of homework effort, completion, and math achievement was examined across the four profiles. Table 4 includes the mean scores across latent profiles on homework effort, completion, and math achievement. Table 5 includes chi-square statistics for pairwise differences between latent profiles on homework effort, completion, and math achievement.
Table 4. Mean Scores Across Latent Profiles on Homework Effort, Completion, and Math Achievement (n = 3018)
| Profile 1: Low (n = 786) | Profile 2: High (n = 404) | Profile 3: Very Low (n = 119) | Profile 4: Moderate (n = 1709) | Overall chi-square test value (df = 3) | Effect size (d) | |
|---|---|---|---|---|---|---|
| M (SE) | M (SE) | M (SE) | M (SE) | |||
| Homework effort | 3.09b (0.02) | 3.45d (0.03) | 2.86a (0.08) | 3.21c (0.01) | 102.592*** | 0.38 |
| Homework completion | 3.75a (0.04) | 4.36c (0.04) | 3.57a (0.12) | 4.01b (0.02) | 92.976*** | 0.36 |
| Math achievement | 59.02a (1.02) | 71.59c (1.10) | 54.14a (2.84) | 63.99b (0.64) | 64.909*** | 0.30 |
Means with the same superscript in a row are not statistically different at α = .05.
***p < .001
Table 5. Chi-Square Test Statistics for Pairwise Differences Between Latent Profiles
| Profile Comparison | Chi-Square Test Statistic, p-value | |
|---|---|---|
| Homework effort | 1 (Low) vs. 2 (High) | 110.211, (< .001) |
| 1 (Low) vs. 3 (Very Low) | 7.917, (= .005) | |
| 1(Low) vs. 4 (Moderate) | 21.430, (< .001) | |
| 2 (High) vs. 3 (Very Low) | 52.035, (< .001) | |
| 2 (High) vs. 4 (Moderate) | 66.791, ( < .001) | |
| 3 (Very Low) vs. 4 (Moderate) | 19.551, ( < .001) | |
| Homework completion | 1 (Low) vs. 2 (High) | 112.525, ( < .001) |
| 1 (Low) vs. 3 (Very Low) | 1.999, ( = 157) | |
| 1(Low) vs. 4 (Moderate) | 35.874, (< .001) | |
| 2 (High) vs. 3 (Very Low) | 36.773, ( < .001) | |
| 2 (High) vs. 4 (Moderate) | 49.154, (< .001) | |
| 3 (Very Low) vs. 4 (Moderate) | 12.646, ( < .001) | |
| Math achievement | 1 (Low) vs. 2 (High) | 70.496, ( < .001) |
| 1 (Low) vs. 3 (Very Low) | 2.593, ( = .107) | |
| 1(Low) vs. 4 (Moderate) | 16.775, ( < .001) | |
| 2 (High) vs. 3 (Very Low) | 32.954, ( < .001) | |
| 2 (High) vs. 4 (Moderate) | 35.126, ( < .001) | |
| 3 (Very Low) vs. 4 (Moderate) | 11.470, ( = .001) |
Taken together, results revealed that profile membership was significantly related to homework effort, completion, and math achievement, with small to medium effect size. Across homework time, completion, and math achievement, High Profile had significantly higher scores than Moderate Profile, which in turn had significant higher scores than Low Profile. Results further revealed that Low Profile had significant higher scores in homework effort than Very Low Profile. Although the differences in homework completion and math achievement between Low Profile and Very Low Profile were not statistically significant, a clear trend was observed in that Very Low Profile had lower scores in both homework completion and math achievement than Low Profile.
Discussion
Our investigation extends prior research on homework purpose by adopting a person-centered approach to identify the possible combinations of math homework purposes and to examine differences among the empirically derived combinations or profiles. First, the results from PLA confirm the hypotheses formulated based on the data of Xu (2023). Specifically, also four profiles of students were identified: High Profile (high in all purposes; 13.39%), Moderate Profile (moderate in all purposes; 56.63%), Low Profile (low in all purposes; 26.04%), and Very Low Profile (very low in all purposes; 3.94%). These four homework purpose profiles are similar to those obtained in Xu (2023), both in the combination of purposes and in the percentage of students in each profile. In short, the profiles are very similar.
Second, consistent with theoretical expectation and related prior research using a variable-centered approach (Epstein & Van Voorhis, 2012; Sun et al., 2020a, 2020b; Wigfield et al., 2015; Xu, 2005), we found that the profile of students with a high level of homework purposes (i.e., our High Profile learners) were those who expended most homework effort, completed most homework, and scored highest on math achievement. In contrast, the profile of students with a low level of homework purposes (i.e., our Low Profile and Very Low Profile learners) were those who expended least homework effort, completed least homework, and scored lowest on math achievement.
Third, whereas the above findings were consistent with a previous study applying a person-centered approach with high school students (Xu, 2023), the current study extended these findings to middle school students. These findings are particularly noteworthy given that we controlled two important background variables - parent education and prior math knowledge - in our current study, neither variable was controlled in the previous study (Xu, 2023).
Our findings regarding these four profiles and their associations with homework effort, completion, and student achievement suggests that (a) homework purposes perceived by students matter in the homework process, and (b) students with a low or high level of one homework purpose (e.g., academic) are likely to be associating with a low or high level of other homework purposes (e.g., self-regulatory and approval-seeking). Applying a person-centered approach offers a deeper understanding of how these three homework purposes coexist within children, moving beyond stating that these purposes are positively correlated. The homework purpose profiles could help researchers, educators, and parents consider the impact of academic, self-regulatory, and approval-seeking purposes in the homework process.
Hence, it would be beneficial to simultaneously attend to all three purposes (academic, self-regulatory, and approval-seeking). In addition to Warton's proposition (2001) that “if students are to be convinced of the value of homework and invest their time and effort in it, then teachers and parents will need to be aware of the types of work most likely to lead to academic improvement” (p. 157), teachers and families will need to be mindful of the types of homework feedback and support most likely to result in both academic improvement and the development of self-regulatory habits and skills. Such attention is especially important for middle school students relating to their math homework, as student attitude toward homework plays a more and more important role in homework completion and academic achievement (Cooper et al., 1998; Xu, 2022). Yet, as children make transition from elementary to secondary school, math value beliefs tends to decrease (Jacobs et al., 2002; Regueiro et al., 2017; Wigfield et al., 2015) and their attitude toward homework becomes more negative (Xu, 2004).
In particular, our results that students have different homework purpose profiles imply that teacher and parent support ought to be differentiated. Students in High Profile may need less external support, yet encouraging them to articulate and share their homework purposes (e.g., what homework means to them) may benefit students in this profile as well as students in other profiles. For remaining learners (with students in Low Profile and Very Low Profile in particular), it would be helpful for teachers to carefully selecting and assigning high-quality (e.g., to show students the importance and relevance of homework to understand the material covered in class), to make homework assignments more interesting for students (e.g., activity and content interest, Corno & Xu, 2004), and to provide high-quality homework feedback (e.g., useful and positive feedback according to the needs of students in these profiles). It would also be helpful to provide professional development opportunities for teachers, as teacher education programs tend to focus on the quality of classroom instruction (e.g., planning, implementing, and assessing), but not on the quality of homework practice (e.g., regarding homework quality, the quality of homework feedback, and autonomy support; Rosário et al., 2018; Xu, 2016).
Like the vast majority of educational research, this research is not exempt from some limitations. Although the results of this study with eighth-grade students coincide with those also obtained by Xu (2023), with eleventh-grade students, it would be beneficial to pursue this line of research involving elementary school students and in other achievement domains, as there are developmental differences in task value perceived by students across different domains (e.g., language arts and sports; Jacobs et al., 2002). Likewise, the results of our study are likely to be generalizable to other collectivist cultures for the following reason. Academic and self-regulatory purposes represent self-focused motive, whereas approval-seeking purpose reflects other-focused motive (Cooper et al., 2016; Sun et al., 2021; Xu, 2023). The findings regarding non-overlapping profiles of these homework purposes from the present study and the previous study (Xu, 2023) suggest that academic and self-regulatory purposes might have become more other-focused (or less self-focused) in collectivist cultures such as China, where self is frequently defined as its roles for the good of the community and the family (Chen et al., 2006), and where interdependence is more highly valued than independence (Hofstede, 2003). Nevertheless, it would be beneficial to continue this line of research in cross-cultural settings, as homework purposes perceived by children are likely to be influenced by cultural norms and expectations (e.g., concerning the value of homework, effort, hard work, and conformity; Cai, 2003; Sun et al., 2020a). Finally, it would be beneficial to conduct qualitative studies (e.g., involving purposive samples of children from Very Low Profile, Low Profile, Moderate Profile, and High Profile), for example, using focus group methodology (e.g., Rosário et al., 2019a,2019b), to better understand the combination of homework purposes from the student viewpoint in each profile, and consequently what new insights may be gained from this line of research to promote the value of homework for children.
