Relations Between Parents’ Math Anxiety and Children's Math Learning: The Role of Homework Help

ABSTRACT

At the intersection of developmental and educational science, recent studies have focused on the role of parents’ attitudes and beliefs towards math and their homework‐helping behaviors. Specifically, findings from Maloney et al. (2015) suggest that parents’ math‐related anxiety is negatively related to children's gains in math achievement only when parents frequently help with math homework—suggesting that frequent math‐related interactions at home with caregivers who hold negative affect towards math may relate to children's lowered math achievement and, in turn, their math anxiety. To further understand this complex developmental mechanism, the current study aimed to replicate these findings with a larger, and more diverse sample of elementary school children. A sample of 3,018 U.S. kindergarten, first‐, second‐, and third‐grade students (50.48% female) nested across 242 classrooms were assessed on their math achievement and anxiety at the beginning and end of the school year. Parents self‐reported their own math anxiety and frequency of homework help. Contrary to the work of Maloney et al. (2015), we found no significant interaction effect between parent math anxiety and homework help on children's math achievement. Instead, results revealed (a) consistent, negative main effects of homework help frequency across models and (b) inconsistent, negative main effects of parents’ math anxiety across models and samples on children's gains in math achievement across the school year. These findings underscore the role of parents’ math anxiety and homework help as they relate to child math outcomes. However, there remains a need for additional examination of the nature and emotional valence of parents’ homework help and how this process plays a role in children's math development.

Summary

• Previous research (Maloney et al. 2015) found that parents’ math anxiety related to less math learning in children, only when parents frequently helped with homework.

• We did not replicate the interaction effect found by Maloney et al. (2015) across two sets of replication analyses, instead, finding support for main effects.

• Additional research is necessary to understand parent math anxiety and everyday parent–child math‐related interactions—such as homework help—as they predict children's mathematical development.

1. Introduction

Mathematics is a prominent pillar of children's education, especially in the early learning years of elementary school. Parents are often considered to be their children's “first teacher”, and it is common for children to seek academic help from their parents once they start formal education. But how effective is this help if these parents experience math anxiety? Parents’ math anxiety has been noted as a social‐environmental influence on children's math anxiety and achievement (Casad et al. 2015; Maloney et al. 2015). Indeed, findings from Maloney et al. (2015) revealed that high parental math anxiety—when paired with a higher frequency of parents’ homework‐helping behaviors—was negatively associated with children's math achievement and positively associated with children's math anxiety (Maloney et al. 2015). These findings suggested that frequent exposure to math‐anxious parenting during math homework may be detrimental to children's mathematics development. Given the complexity of this developmental mechanism, it is critical to understand whether the Maloney et al. (2015) findings replicate across different samples and settings, ultimately informing how to support positive math attitudes and learning among parents and children. Therefore, the current study aims to replicate the Maloney et al. (2015) study in a large and diverse sample of children, providing important insights into the generalizability of the previous findings as well as offering guidance to families in how best to support their children's math learning and attitudes.

1.1. The Role of Parents in Children's Academic Achievement

Theories of human development place emphasis on the role of parent–child interactions, so much so that these everyday interactions have been referred to as “the engines of development” (Bronfenbrenner and Evans 2000). The microsystem is the most proximal, contextual layer of the bioecological model of human development (Bronfenbrenner 1979), representing children's everyday environments. It is within this context that interactions between the developing child and socialization agents—such as caregivers—have been underscored as a primary mechanism through which children develop culturally salient skills, such as the way in which they approach learning (Rogoff 2003). Similarly, Bandura's (1977) social learning theory provides a framework of child learning through imitation and modeling. Parents who model positive attitudes and behaviors towards formal schooling can expect their children to pick up on and, in turn, imitate them. For example, observational studies have demonstrated positive effects of parents’ values for participation in formal school settings and communication of high academic expectations on children's achievement (Castro et al. 2015).

1.2. Parental Influences on Children's Math Learning

However, not all parental engagement has positive effects on children's achievement. For example, findings from a meta‐analysis revealed that although parents’ involvement in homework was positively associated with elementary school children's reading and language arts skills, it was negatively associated with children's math achievement (Patall et al. 2008). Accordingly, researchers have begun to explore parent‐level factors that may explain this pattern, such as parents’ own views towards mathematics. Indeed, multiple studies have evidenced positive associations between parents’ positive math attitudes and elementary schoolers’ math problem‐solving skills (e.g., Cook et al. 2025) and overall math achievement (e.g., Cui et al. 2019). Importantly, parents’ positive math attitudes have also been found to relate to children's higher levels of math motivation (e.g., Peixoto et al. 2024), underscoring the role of parents’ attitudes in supporting math learning.

This body of work highlights how parents help shape the emotional and motivational context in which children learn math. However, researchers have also examined how negative parental math attitudes relate to children's math experiences. These negative emotions towards math, particularly math anxiety, can have important implications for children's own math anxiety and achievement (e.g., Schaeffer et al. 2018; Tomasetto et al. 2025).

1.3. Math Anxiety and Math Learning

Math anxiety is commonly described as a negative emotional reaction toward math (Ashcraft and Kirk 2001). It is related to, but distinct from, general anxiety, which involves unspecified worry about everyday occurrences. In contrast, math anxiety is specific to math‐related situations. People with general anxiety are more likely to experience math anxiety given their heightened likelihood to perceive situations as anxiety provoking, and correlations between the two constructs are typically moderate to strong (Rubinsten et al. 2018).

Math anxiety is prevalent and has important implications. Estimates suggest that approximately 15%–25% of individuals experience moderate‐to‐high levels of math anxiety (Hart and Ganley 2019; Lau et al. in preparation). Math anxiety is characterized by an increase in heart rate, sweating, nausea, headaches, and even panic when confronted with a math‐related situation (Chernoff and Stone 2014). Math anxiety and math achievement are consistently negatively related across children, adolescents, and adults (Barroso et al. 2021) and both are important for long‐term educational and occupational outcomes, including Science, Technology, Engineering and Mathematics (STEM) career participation (Ahmed 2018; Hart and Ganley 2019; Ritchie and Bates 2013).

There are three commonly cited theories to help explain the negative relation between math anxiety and math achievement. The first is the debilitating anxiety theory, which posits that math anxiety leads to lower math achievement because it disrupts cognitive processing and causes people to avoid math (Ashcraft 2002; Carey et al. 2016; Eysenck et al. 2007). Consistent with this theory, research has shown that math‐related worries consume working memory resources needed for problem‐solving (Pizzie et al. 2020; Ramirez et al. 2013). In addition, although it can be difficult and even impossible to avoid engaging with math early in development, adolescents with high math anxiety avoid cognitive engagement while doing math (Morsanyi et al. 2014), take fewer math courses (Hembree 1990), and avoid STEM careers (Ahmed 2018). These avoidance behaviors prevent the development of adaptive behaviors that could, in time, alleviate anxiety and also lead to reduced opportunities to learn in math, which both impact later math achievement (Arnaudova et al. 2017).

The second theory is the deficit theory, which suggests that earlier poor performance in math fosters negative emotions associated with math, making one more anxious when encountering math situations in the future (Carey et al. 2016; Maloney et al. 2010 2011). Within this framework, children's math anxiety is often rooted in the fear of making mistakes or not understanding something (Ganley and McGraw 2016), which likely stems from previous experiences of failure in math. This theory has been supported by longitudinal studies—conducted with children (Song et al. 2021) and adolescents (Ma and Xu 2004)—in which initial math achievement predicts later math anxiety.

Finally, reciprocal theory suggests that the processes outlined in the aforementioned theories exist simultaneously, with multiple longitudinal studies providing support for this theory (Cargnelutti et al. 2017; Gunderson et al. 2018; Pekrun et al. 2017). Taken together, research suggests that some students will struggle with high math anxiety and low achievement and that these can reinforce each other. Further investigation is necessary to understand contextual predictors of both children's negative feelings towards math and long‐term achievement.

1.4. Linking Parental Math Anxiety to Children's Math Anxiety and Math Achievement

Math‐anxious individuals are likely to exhibit avoidance behaviors in relation to any aspect of math (Ashcraft and Krause 2007). Therefore, parents who experience math anxiety may exhibit this avoidance when helping their children with math homework. For example, DiStefano et al. (2020) found that parents with math anxiety reported more negative emotions while helping their child with math homework. DiStefano and colleagues (2020) further suggest that these negative emotions can have detrimental effects on children's math learning: children pick up on their parents’ negative emotions during homework help and adopt them as their own, forming negative associations with math learning.

However, Oh et al. (2022) suggest another developmental pathway: negative emotions experienced by math‐anxious parents during homework help may result in controlling parenting behaviors, which then interfere with children's math learning. Parents’ controlling behaviors can include leading children when not requested to do so, physically completing homework for the child (instead of assisting), and checking or grading homework when not being requested to do so by the child. Conscious or not, these controlling behaviors may exist as an effort to decrease the amount of time spent on homework (i.e., a form of avoidance), because the quicker that children finish, the less time parents need to endure negative feelings associated with math. However, this results in less time for the child to be exposed to and actively practice math problems, ultimately affecting their math learning. Indeed, this hypothesized mechanism was supported by Akhavein and Finch (2025) who found that parents’ controlling behaviors during homework help mediated the negative relation between parent math anxiety and children's math achievement. Similarly, Herts et al. (2019) found that parents with more math anxiety scored lower on the quality of their math instruction. These recent studies build on foundational work demonstrating that parents’ math anxiety relates to children's math outcomes.

Researchers have increasingly examined how parents’ math anxiety may play a role in children's development of math anxiety and math learning. Maloney et al. (2015) examined 438 first‐and second‐grade children and their parents. Their results demonstrated a significant interaction effect between the frequency of parents’ homework help and parents’ math anxiety in predicting children's end‐of‐year math achievement. This interaction effect revealed that for parents who reported frequently helping their child with math homework, parents’ math anxiety was negatively related to children's end‐of‐year math achievement. In contrast, this relation was not found for parents who reported infrequent homework help. Moreover, this interaction effect was significant after controlling for child gender, grade level, beginning‐of‐year math achievement, and beginning‐of‐year math anxiety. These findings confirmed the researchers’ hypotheses, suggesting that parents’ math anxiety is negatively related to children's math achievement but only when parents frequently assisted with math homework.

Maloney et al. (2015) also controlled for parent education and teacher factors and found the interaction effect remained. They also tested whether the interaction was specific to math achievement, and not a more generalized parenting effect, by including end‐of‐year reading achievement as the outcome (with parallel reading covariates). They did not find a significant interaction for this reading model, indicating their finding was specific to math achievement. Finally, they tested a moderated mediation model, extending the key interaction to also test if lower end‐of‐year math achievement would relate to an increase in math anxiety, consistent with the deficit theory. The authors found that indeed when parents frequently helped with math homework, parents’ math anxiety was related to their children's end‐of‐year math anxiety, and this relation was mediated by the children's end‐of‐year math achievement.

1.5. The Current Study

The Maloney et al. (2015) findings have been very influential in the field of mathematical cognition and in the literature on the role of parenting behaviors in children's socioemotional and cognitive outcomes (e.g., the article has been cited over 900 times to date, with Google Scholar citations per year showing it generally gets cited more than 100 times per year). The original findings also have important policy implications about homework and more generally on the role of parents in children's schooling. Given this impact, we thought it was important to conduct a conceptual replication of the Maloney et al. (2015) key findings in a different sample, to test the robustness and generalizability of the original effect. Therefore, in the current study we aimed to replicate the key interaction effect, and subsequent moderated mediational pathway, found by Maloney et al. (2015) across two samples: a matching sample and a broader sample. In both samples, we test the interaction between parent math anxiety and frequency of homework help in predicting child math achievement, include parallel covariates to Maloney et al. (2015) in a model building process, test for moderated mediation with children's math anxiety as a distal outcome, and examine the specificity of the interaction effect by using reading as an outcome for comparison.

In the matching sample replication, we use a sample of first‐ and second‐graders, mirroring the ages in Maloney et al. (2015). In the broader sample replication, we add kindergarten and third graders to the matching sample, but exclude any children who—as reported by their teacher—were never assigned math homework. In Table S1 in the supplemental materials we detail the methodological similarities and differences between Maloney et al. (2015) and the current study.

2. Methods

2.1. Participants and Procedure

Participants were recruited as part of the Research on Experiences, Attitudes, and Learning in Math (REALM)  Study, a longitudinal study focused on examining how teacher math anxiety is related to math instructional practices and student math attitudes and achievement. The overall study sample consisted of 3,018 kindergarten, first‐, second‐, and third‐grade students and their parents drawn from 242 classrooms located in Florida, USA across 2 cohorts. Analytic samples for the matching sample were 805–851 and for the broader sample were 1,449–1,528. Data collection for the first cohort was during the 2017–2018 school year and for the second cohort was during the 2018–2019 school year. Families with children in participating classrooms received a letter of invitation to participate in the study, and all children who returned consent forms and assented were enrolled, with no criteria for exclusion.

At the time of enrollment, across the full sample of 3,018 children, kindergartners were 5.61 years old (range = 5–7.33), first‐graders were 6.67 years old (range = 5.00–8.42), second‐graders were 7.67 years old (range = 6.25–9.50), and third‐graders were 8.70 years old (range = 7.08–10.50) on average and 50.50% female. The diversity of the sample was representative of the school districts from which they were drawn. Of those who reported racial‐ethnic background information (12% did not report), 4% of children identified as Asian, 22% of children identified as Black, 12% of children identified as Hispanic, 49% of children identified as White, 13% of children identified as Multiracial, and 1% of children identified as Other. Caregivers completing the survey self‐identified their relation to children, with 82% identifying as mothers (biological, adoptive, step, foster), 12% identifying as fathers (biological, adoptive, step, foster), 3% identifying as a parent broadly (biological, adoptive, step, foster), and 3% identifying as a grandparent, aunt, uncle, or other caregiver/guardian. Among caregivers who reported their educational background (12% did not report), 20% of respondents had a high school degree or less, 35% reported having completed some college or an associate's degree, 24% reported having completed a bachelor's degree, and 21% reported having competed a graduate or professional degree (MS, PhD, JD). Children came from schools where, on average, 51% of children received free or reduced‐price lunch (an indicator of low SES). This statistic is similar to levels reported by Maloney et al. (2015) but was slightly lower than rates for the entire state of Florida during the time of the study (62%, FL Dept. of Ed., 2018). Demographic information for the matching and broader analytic samples specifically is in Table S1.

Embedded within consent forms, parents completed a background questionnaire and provided demographic information for themselves and their children. Additionally, parents reported on their own math anxiety and answered a question about how frequently they help their children with their math homework. Children completed surveys and a math achievement assessment at the beginning (September/October) and end (April/May) of the academic year. These surveys and assessments were administered by children's teachers in their classrooms using researcher‐provided instructions. They ensured comprehension by reading the tasks aloud and instructing children to follow along in their survey booklets before responding. Children completed sample items before completing the survey and the math assessment.

2.2. Measures

2.2.1. Parent Math Anxiety

The Single‐Item Math Anxiety Scale (SIMA; Nuñez‐Peña et al. 2014) was adapted to assess parents’ level of math anxiety and was chosen due to limitations on access to parents’ time. Parents were asked to circle their response to the prompt: “On a scale of 1–10, with 10 being the most anxious, how anxious are you about math?”, with 1 being Not Anxious and 10 being Very Anxious. There is evidence from large samples of adults that this single indicator of math anxiety is a good proxy for multi‐item measures of math anxiety, with significant and strong associations found between this one‐item scale and an 11‐item scale (r = .87; Hart and Ganley 2019), a 12‐item scale (r = .84; Hart and Ganley 2019), and a 25‐item scale (r = .77; Nuñez‐Peña et al. 2014).

2.2.2. Parent Homework Help Frequency

In line with the work of Maloney et al. (2015), parents were asked “How often do you help your child with his or her math homework?” and responded on a 6‐point Likert scale (e.g., 1 = Never, 2 = Once a month, 3 = Less than once a week, 4 = Once a week, 5 = 2–3 times a week, 6 = Every day).

2.2.3. Child Math Anxiety

Children completed a revised version of the Math Anxiety Scale for Young Children (MASYC‐R2; Ganley et al. 2023; adapted from Ganley and McGraw 2016), which consisted of 14 items such as “I get worried before I take a math test.” For children in grades first through third, each item was rated on a 4‐point Likert scale (No, Not really, Kind of, or Yes). To facilitate understanding of the rating task for kindergarten children, the Likert scale was presented so that the font size of the options visually depicted the amount of disagreement or agreement with the item. Strong agreement was presented as a large font and moderate agreement was presented as a normal sized font. The Likert scale was No (written in size 36 font), No (written in size 18 font), Yes (written in size 18 font), or Yes (written in size 36 font). Children were instructed to decide between Yes or No and then decide how much they felt that way (i.e., “a little”/“kind of” or “really”) to correspond to the font size. We computed a mean across the 14 items to create their math anxiety score and reliability across items was strong (ordinal Cronbach's α = .93).

2.2.4. Child Math Achievement

Children completed the Elementary Mathematics Student Assessment (EMSA; Schoen et al. 2021) at the beginning and end of the school year. The EMSA is a standardized test of children's math achievement designed to test grade‐level appropriate math knowledge in alignment with Common Core State Standards for Mathematics (CCSS‐M) within the following mathematical domains: 1) counting and cardinality, 2) word problems, 3) number relations, 4) fractions, 5) basic number facts, and 6) multi‐digit computation. The EMSA was administered in a group setting in a paper and pencil format, featuring both selected‐response and constructed‐response items in both numeric and word problem formats. Students’ final scores were operationalized as theta scores, with higher scores indicating higher levels of math knowledge. Using a 2‐parameter logistic model based on item‐response theory, scores were determined through vertical scaling across grade level and performed via a fixed‐item‐parameter approach, with grade 1 specified as the base scale within a common item nonequivalent group design. Any students who did not respond to at least 80% of the items on the EMSA were reported as missing data. Schoen et al. (2021) reported marginal reliability estimates of .78, .80, .83, and .88 on the common metric tests for grades kindergarten through third, respectively.

2.2.5. Child Reading Achievement

First‐ through third‐grade students completed the Test of Silent Reading Efficiency and Comprehension (TOSREC; Wagner et al. 2010), a standardized, group‐administered measure of decoding and reading comprehension. Administered by teachers, students were given three minutes to read sentences and answer “Yes” or “No” to indicate whether the sentence made sense. Final scores were calculated by summing the total number of correctly answered items and subtracting the number of incorrectly answered items, to account for chance. Scores below 0 were assigned a 0. The TOSREC has grade and time of school year versions that were given at the appropriate age and time. First‐grade assessments had 50 items (possible scores = 0 – 50) and the second‐ and third‐grade tests each had 60 items (possible scores = 0 – 60). The TOSREC reliability coefficients exceed .85 across all forms and grade levels and reported reliability coefficients with other reading measures (e.g., Woodcock Johnson Test of Achievement, 3rd edition [WJIII], Group Reading Assessment and Diagnostic Evaluation [GRADE]) exceed .70 (Wagner et al. 2010).

Teachers were given step‐by‐step instructions for administration that included the 3‐minute limit for this task. However, there were a number of teachers who did not set a time limit and therefore most or all of their students had time to complete all of the items on the assessment. This impacted 12% of students at the beginning‐of‐school‐year and 19% at the end‐of‐school‐year. To account for this, we conducted a regression analysis predicting TOSREC scores from whether the teacher used the time limit or not. We then pulled the residualized scores from this analysis and used these scores in our analyses so that the score more closely approximated what they would be if everyone received the time limit.

Given that most kindergarten students are still learning to read, there is no TOSREC for this grade. Therefore, we administered a researcher‐developed measure consisting of four letter identification items (e.g., “Which one is L?”), four sound identification items (e.g., “Which one makes the /G/ sound?”), four word identification items (e.g., “Which one says nap?”), and four sight word identification items (e.g., “Which one says two?”). Across all 16 items, students were given four answer options. Descriptive findings from the first cohort's beginning‐of‐school‐year scores revealed ceiling effects on the letter identification items (all item means > 95%), so we omitted these items in Cohort 2, adding one word identification and one sight word identification item in their place. Final beginning‐of‐school‐year scores for both cohorts were calculated using the 12 shared items only (4 sound, 4 word, 4 sight word). One sight word was dropped (went), as it had an item‐total correlation of .049. The internal consistency (Cronbach's alpha) of the final 11‐item scale was .75. At the end of the school year, students completed the same items in both cohorts, which included 8 word identification and 9 sight word identification items (3 of which and 4 of which overlapped with the beginning‐of‐school‐year items, respectively). The internal consistency (Cronbach's alpha) of the 17‐item scale was .73.

2.3. Analytic Plan

We conducted the same analyses across two overlapping samples: (a) a matching sample to Maloney et al. (2015)—in which the analytic sample consists of just first‐ and second‐graders—and (b) a broader sample—in which data from all children and families (kindergarten—third‐grade) are used in the analysis, excluding children nested in classrooms in which teachers reported assigning no math homework. The inclusion of a broader sample replication allowed us to test these same hypotheses for a broader sample of kindergarten through third‐graders, not just first‐ and second‐graders. For each sample block, first, descriptive statistics were computed for all child and parent variables. Then, bivariate correlations were conducted to determine associations between children's beginning‐of‐year and end‐of‐year child math scores, children's gender, parents’ math anxiety, and parents’ homework help frequency. Finally, a series of regression models were estimated for each sample using the PROCESS (Hayes 2022) macro in SPSS v. 4.2. We used listwise deletion to match the strategy from Maloney et al. (2015).

In both samples, model building practices adhere closely to the research methodology of Maloney et al. (2015). Model 1 predicts children's end‐of‐year math achievement from main effects of parent math anxiety and parent homework help frequency; the interaction effect of parent math anxiety × parent homework help frequency; and covariates of children's beginning‐of‐year math achievement, beginning‐of‐year math anxiety, gender, and grade level. This was the base model of Maloney et al. (2015), testing their research question of whether parents’ math anxiety was related to students’ end‐of‐year math achievement and whether this relation varied as a function of how frequently parents helped with math homework.

Model 2 includes all of the same parameters as Model 1 but also includes parent education as an additional covariate. In Maloney et al. (2015), the authors sought to test if their key interaction results held when controlling for parent math knowledge, which they operationalized as parents’ highest level of education.

Model 3 includes school‐level SES as an additional covariate (note, we diverge from Maloney et al. (2015) in that we did not account for teacher math anxiety and teacher math knowledge in our models). In Maloney et al. (2015), the authors again sought to test their base model finding by examining if the results held when controlling for teacher and classroom factors, measured as teachers’ math anxiety and math knowledge, and school‐level SES.

Model 4 uses end‐of‐year reading achievement as the dependent variable rather than math achievement and includes beginning‐of‐year reading achievement, beginning‐of‐year reading anxiety, grade and gender as covariates. In this model, Maloney et al. (2015) wanted to test the specificity of the interaction on math, rather than more broadly to achievement, represented here as testing if the effect is seen on reading achievement. We diverge from Maloney et al. (2015) as we did not have a reading anxiety measure in our study.

Finally, Model 5 specifies a moderated mediation analysis, to test if the indirect relation between parents’ math anxiety and children's end‐of‐year math anxiety through children's end‐of‐year math achievement is moderated by homework help frequency, while controlling for children's beginning‐of‐year math anxiety, beginning‐of‐year math achievement, gender, and grade level. For Maloney et al. (2015), this model was to test the extended hypothesis that for parents who frequently help their children with math homework, parent math anxiety would be associated with children's math anxiety because of decreased math achievement, consistent with the deficit theory.

For the broader sample, we estimated these same models, but the sample included kindergarten through third‐graders and excluded children who—as reported by their teacher—were never assigned math homework (17% of kindergartners, 8% of first‐graders, 10% of second‐graders, 2% of third‐graders). To elucidate the origin of potential differences in findings (i.e., including kindergarten and third‐graders or dropping children without math homework), full broader sample replication block findings can be found in our Supplemental Material—in which Block 2a includes first‐ and second‐graders, excluding those without math homework, Block 2b includes kindergarten—third‐graders regardless of if they were assigned math homework, and Block 2c is the broader sample included in the manuscript, which includes kindergarten—third‐graders, excluding those without math homework.

3. Results

3.1. Matching Sample Replication of Maloney et al. (2015)

3.1.1. Descriptive Statistics

As can be seen in the last row of Table 1, children's math achievement scores as assessed by the EMSA ranged from –3.43 to 2.41 (M = ‐0.63, SD = 1.13) at the beginning of the academic year and ranged from –2.75 to 2.66 (M = 0.31, SD = 0.97) at the end of the school year. Parents’ math anxiety scores ranged from 1 to 10 (M = 4.76, SD = 2.85), demonstrating variability in experiences of math anxiety with parents reporting moderate levels of math anxiety on average. Notably, although different scales were used across studies, the current sample exhibited slightly higher levels of parent math anxiety on average than that of Maloney et al. (2015), with their participants reporting only “a little anxiety” (M = 2.10, SD = .90, range = 1 – 5) when asked to estimate their anxiety across various math scenarios. On a scale of 1 to 6, parents reported helping their child with math homework two‐to‐three times per week (M = 5.09; SD = 1.19), consistent with Maloney et al. (2015; M = 5.30, SD = 1.30). There were no significant differences in homework help frequency by grade level (M _{first_grade} = 5.11, M _{second_grade} = 5.08; (F(1, 1255) = .16, p = .69). In line with Maloney et al. (2015), because responses on the homework‐help question were left‐skewed (skewness coefficient = ‐1.58), we transformed the responses by squaring them to normalize the distribution (Chambers et al. 1983) before using these values in our analyses.

TABLE 1.

Bivariate Correlations Between Key Variables for Matching Sample Replication.

Variable	Beginning‐of‐Year Math Score	End‐of‐Year Math Score	Beginning‐of‐Year Math Anx	End‐of‐Year Math Anx	Beginning‐of‐Year Reading Score	End‐of‐Year Reading Score	Gender	Parent Math Anxiety	Parent Math HW Help
Beginning‐of‐Year Math Score	—
End‐of‐Year Math Score	.70 ***	—
Beginning‐of‐Year Math Anxiety	−.29 ***	−.29 ***	—
End‐of‐Year Math Anxiety	−.23 ***	−.27 ***	.44 ***	—
Beginning‐of‐Year Reading Score	.41 ***	.39 **	−.20 **	−.12 **	—
End‐of‐Year Reading Score	.31 ***	.41 **	−.17 **	−.14 **	.59 **	—
Gender	−.01	−.02	.08 **	.15 ***	.04	−.00	—
Parent Math Anxiety	−.11 ***	−.11 ***	.06 *	.06 *	−.09 **	−.09 **	.06 *	—
Parent Math Homework Help	−.22 ***	−.24 ***	.15 ***	.07 *	−.15 **	−.12 **	.02	.16 ***	—
Mean	−0.63	0.31	2.04	1.83	0.00	0.00	0.50	4.76	5.09
Standard Deviation	1.13	0.97	0.76	0.70	1.00	1.00	0.50	2.85	1.19
N	1288	1236	1282	1218	1268	1209	1280	1240	1257
Range	−3.43 – 2.41	−2.75 – 2.66	1 – 4	1 – 4	−3.04 – 3.43	−2.73 – 2.56	0 – 1	1 – 10	1 – 6

Anx = Anxiety, HW = Homework. Pairwise deletion was used for correlations.

^* p < .05, ^** p < .01, ^*** p < .001.

3.1.2. Bivariate Correlations

Given the large sample size, even very small correlations (i.e., r = .05) are statistically significant, though many of these small correlations are likely not practically significant. Therefore, we also discuss the magnitude of correlations using effect size guidelines for individual differences research from Gignac and Szodorai (2016), with correlations under .10 considered very small/weak, around .10 being considered small/weak, around .20 being medium/moderate, and .30 or higher being considered large/strong.

Children's beginning‐ and end‐of‐year math achievement scores were strongly positively associated with one another (r = .70, p < .001) and children's beginning‐ and end‐of‐year math anxiety scores were also strongly positively associated with one another (r = .44, p < .001). Children's gender was not associated with math achievement at either time point (rs = −.01, −.02; ps > .05), but it was weakly positively associated with children's math anxiety at both time points. Specifically, girls reported higher math anxiety on average than did boys at both the beginning (r = .09, p = .002; M _boy = 1.99, M _girl = 2.12) and end of the school year (r = .15, p < .001, M _boy = 1.74, M _girl = 1.94). Children's gender was also very weakly positively associated with parents’ math anxiety (r = .06, p = .05) indicating that parents of girls were slightly more likely to report higher levels of math anxiety.

In line with our expectations, there were significant, negative small correlations between parents’ math anxiety and children's math scores at the beginning and end of the school year (r _beginning = −.11, p < .001; r _end = −.11, p < .001), indicating that on average, children with lower math achievement scores across the school year were slightly more likely to have a parent with higher levels of math anxiety. Similarly, parents’ homework help was weakly negatively associated with children's math scores at the beginning (r = −.22, p < .001) and end of the school year (r = −.24, p < .001), indicating that on average, children who exhibited lower math achievement scores had parents who reported helping more frequently with math homework. Moreover, parents’ math anxiety exhibited a small to moderate, positive association with their self‐reported frequency of homework help (r = .16, p < .001).

3.1.3. Moderation Analysis

For Model 1, we tested the moderating effect of homework help frequency on the relation between parent math anxiety and children's end‐of‐year math achievement while controlling for students’ grade, gender, beginning‐of‐year math achievement, and beginning‐of‐year math anxiety. Shown in Table 2, Model 1 results demonstrated a negative main effect of homework help frequency on children's end‐of‐year math achievement (b = −.005, t = ‐2.05, p = .04) and no significant main effect of parent math anxiety on children's end‐of‐year math achievement (b = −.01, t = ‐1.31, p = .19), both of which are consistent with Maloney et al. (2015). However, contrary to findings from Maloney et al. (2015), the negative relation between parent math anxiety and children's end‐of‐year math achievement was not significantly moderated by the frequency of parents’ homework help (b = .0003, t = .40, p = .69). Despite the lack of a significant interaction, we plotted simple slopes for the interaction (Figure 1A) to compare with Maloney et al. (2015)’s Figure 1. The figure shows the main effect of homework help, but not for parent math anxiety, and the similar slopes explain the lack of a significant interaction. Moreover, unlike Maloney et al. (2015), children's beginning‐of‐year math anxiety (b = −.09, p = .001) and grade level (b = −.18, p = .002) each individually contributed a significant amount of variance to the model as covariates

TABLE 2.

Unstandardized Regression Results of Models 1–3 Predicting First and Second Graders’ End‐of‐Year Math Achievement.

	Model 1			Model 2			Model 3
Variable	b	SE	p	b	SE	p	b	SE	p
Constant	1.15 ***	0.11	<.001	0.91 ***	0.16	<.001	1.04 ***	0.17	<.001
Parent Math Homework Help	‐ 0.005*	0.002	.04	−0.004	0.002	.07	−0.004	0.002	.14
Parent Math Anxiety	‐ 0.01	0.01	.19	−0.01	0.01	.56	−0.005	0.01	.59
Parent Math Anxiety × Math Homework Help	0.0003	0.0008	.69	0.0002	0.001	.84	0.0002	0.001	.83
Gender	0.04	0.05	.37	0.04	0.05	.37	0.04	0.05	.36
Grade	‐ 0.18 **	0.06	.002	−0.14 *	0.06	.02	−0.12	0.06	.05
Beginning‐of‐Year Math Achievement	0.62 ***	0.03	<.001	0.60 ***	0.03	<.001	0.59 ***	0.03	<.001
Beginning‐of‐Year Math Anxiety	‐ 0.09 **	0.03	.001	−0.09 **	0.03	.004	−0.09 **	0.03	.003
Parent Highest Education Level				0.04 *	0.02	.03	0.02	0.02	.20
School Level SES							−0.002	0.001	.07
df			7(828)			8(809)			9(808)
R2			.52			.53			.53
F			130.01 ***			113.35 **			101.43 ***
n			836			818			818

Note: Parent Math Anxiety and Parent Homework Help were mean centered before being entered into the model.

Gender: 0 = male, 1 = female; Grade: Grade 1 = 1, Grade 2 = 2; ^*p<.05, ^**p<.01, ^***p < .001.

FIGURE 1.

The Relation between Parents’ Math Homework Help Frequency and End of Year Math Scores at Different Levels of Parents’ Math Anxiety A. Matching Sample. B. Broader Sample. Note. Predicted values are at 1 SD above and below the mean and at mean levels of all other model covariates.

In Model 2, we added caregivers’ self‐reported highest education level to Model 1 as a control variable. Parents’ highest level of education contributed a significant amount of variance to the model (b = .04, p = .03), but in this model, frequency of homework help (b = −.004, p = .07) was nonsignificant in predicting children's end‐of‐year math achievement. In Model 3, we further controlled for school‐level SES, which was not a significant predictor and did not change the overall pattern of findings in Model 2. Finally, as can be found in our Supplemental Material, we constructed a model in which children's end‐of‐year reading achievement was the outcome of interest (Model 4), predicted by parents’ math anxiety, homework helping frequency, and the interaction of these predictors, while controlling for gender, grade level, and beginning‐of‐year reading achievement. Model 4 results demonstrated null main and interaction effects for all primary predictors. Additionally, of the covariates, only children's beginning‐of‐year reading achievement could account for a significant amount of variance in end‐of‐year reading achievement (b = .58, t = 21.07, p < .001).

3.1.4. Moderated Mediation

Despite null findings surrounding moderation effects, to further attempt to replicate the findings of Maloney et al. (2015), we conducted a moderated mediational analysis to test if the indirect relation between parents’ math anxiety and children's end‐of‐year math anxiety through their end‐of‐year math achievement differed as a function of parents’ homework help frequency. Identical to the bootstrapping approach used by Maloney et al. (2015), we interpreted and compared the magnitude of indirect effects at one standard deviation above the mean, at the mean, and below the mean of parents’ homework help frequency. The indirect relation was not significant at the mean of homework help. In addition, the index of moderated mediation—a global indicator of significance of the entire moderated mediational model—was non‐significant (95% CI = −.003, .001), indicating that the indirect relation between parents’ math anxiety and children's math anxiety through children's math achievement does not differ in size as a function of how often parents report helping their children with homework.

3.2. Broader Sample Replication of Maloney et al. (2015)

Given the differences between our dataset and that of Maloney et al. (2015), we were interested in testing these same hypotheses for our entire sample—which included kindergarten through third‐graders, not just first‐ and second‐graders—as well as incorporating a modification that increases the internal validity of our findings: excluding children in classrooms in which the teacher reported not assigning any math homework (Block 2c). Tables outlining results across the intermediary Blocks 2a and 2b are available in the Supplemental Material and we will focus on results from Block 2c below.

3.2.1. Descriptive Findings

Descriptive findings within the broader sample (kindergarten through third‐graders, excluding those without math homework) were comparable to that of the matching replication sample, however, we highlight notable differences between the two samples. First, as can be seen in Table 3, children's math achievement at both time points were slightly lower (M_{begin_achievement} = −.75; M_{end_achievement} = .25) and exhibited a greater spread—standard deviation (SD_{begin_achievement} = 1.60; SD_{end_achievement} = 1.34), which makes sense given the larger age range.

TABLE 3.

Bivariate Correlations Between Key Variables for Broader Sample Replication.

Variable	Beginning‐of‐Year Math Score	End‐of‐Year Math Score	Beginning‐of‐Year Math Anx	End‐of‐Year Math Anx	Beginning‐of‐Year Reading Score	End‐of‐Year Reading Score	Gender	Parent Math Anxiety	Parent Math HW Help
Beginning‐of‐Year Math Score	—
End‐of‐Year Math Score	.84 ***	—
Beginning‐of‐Year Math Anxiety	−.28 ***	−.28 ***	—
End‐of‐Year Math Anxiety	−.17 ***	−.24 ***	.40 ***	—
Beginning‐of‐Year Reading Score	.30 ***	.32 ***	−.17 ***	−.12 ***	—
End‐of‐Year Reading Score	.25 ***	.34 ***	−.16 ***	−.14 ***	.60 ***	—
Gender	−.02	−.01	.10 ***	.14 ***	.02	.03	—
Parent Math Anxiety	−.06 **	−.11 ***	.06 *	.09 ***	−.11 ***	−.14 ***	.07 **	—
Parent Math Homework Help	−.07 **	−.15 ***	.11 ***	.08 **	−.16 ***	−.14 ***	.05 *	.17 ***	—
Mean	−0.75	0.25	2.06	1.88	−0.01	−0.01	0.51	4.86	5.05
Standard Deviation	1.60	1.34	0.76	0.71	1.00	1.00	0.50	2.87	1.29
N	2375	2329	2341	2294	2352	2289	2386	2351	2323
Range	−5.23 – 2.89	−3.75 – 3.62	1 – 4	1 – 4	−3.69 – 3.76	−3.09 – 3.69	0 – 1	1 – 10	1 – 6

Notes. Pairwise deletion was used for correlations.

^* p < .05, ^**p < .01, ^***p < .001.

Similar to the findings from the matching sample replication analyses, parents of kindergarten through third‐graders reported helping their child with math homework two‐to‐three times per week (M = 5.05; SD = 1.30)—and just as in the matching sample replication block, this left‐skewed distribution was transformed prior to being entered into predictive models by squaring responses to normalize the distribution. Parents’ reports of homework help frequency differed significantly by grade level (K = 4.86, first = 5.15, second = 5.10, third = 5.07; F(3, 2319) = 5.36, p = .001). Post hoc comparisons using the Bonferroni correction indicated that parents of first‐, second‐, and third‐graders reported helping their children with homework more often than parents of kindergartners.

3.2.2. Bivariate Correlations

Bivariate correlations are shown in Table 3. In terms of magnitude, direction, and p‐value, they were comparable to those found in the matching sample replication.

3.2.3. Moderation Analysis

Results from Model 1 – testing the moderating effect of parents’ homework help on the relation between parents’ math anxiety and children's math achievement—revealed a slightly different pattern to those from both the matching sample replication and Maloney et al. (2015): parents’ math anxiety had a significant, negative main effect on children's end‐of‐year math achievement (see Table 4; b = −.02, p = .004; t = ‐2.90). Besides this difference, the pattern of findings mirrored that of Block 1, with a significant negative main effect of homework help frequency (b = −.01, t = ‐4.30, p < .001), consistent with Maloney et al. (2015), but a null Parent Math Anxiety × Homework Help interaction effect (b = .00, t = 1.05, p = .29), and children's beginning‐of‐year math anxiety and grade level remained significant covariates in the model. Despite the lack of a significant interaction, we plotted simple slopes for the interaction to compare with Maloney et al. (2015; Figure 1B). Slopes show the main effect of homework help and parent math anxiety, and the similar slopes explain the lack of a significant interaction. After controlling for parents’ education in Model 2, and after adding school‐level SES as a covariate in Model 3, this same pattern of findings held. As can be found in our Supplemental Material, Model 4 replication results with reading achievement showed, unlike the matching sample, significant negative main effects of both math homework help and parent math anxiety, but consistent with the matching sample and Maloney et al. (2015), there was not a significant interaction.

TABLE 4.

Unstandardized Regression Results of Models 1–3 Predicting Kindergarten through Third Graders’ End‐of‐Year Math Achievement.

	Model 1			Model 2			Model 3
Variable	b	SE	p	b	SE	p	b	SE	p
Constant	1.01 ***	0.07	<.001	0.92 ***	0.10	<.001	0.94 ***	0.12	<.001
Parent Math Homework Help	−0.01 ***	0.002	<.001	−0.01 ***	0.002	<.001	−0.01 ***	0.002	<.001
Parent Math Anxiety	−0.02 **	0.01	.004	−0.02 *	0.01	.03	−0.01 *	0.01	.03
Parent Math Anxiety × Math Homework Help	0.001	0.001	.29	0.001	0.001	.28	0.001	0.001	.28
Gender	0.05	0.04	.18	0.05	0.04	.19	0.05	0.04	.18
Grade 0	−0.22 ***	0.06	<.001	−0.23 ***	0.06	<.001	−0.23 ***	0.06	<.001
Grade 2	−0.26 ***	0.06	<.001	−0.23 ***	0.06	<.001	−0.23 ***	0.06	<.001
Grade 3	0.01	0.07	.93	0.03	0.07	.68	0.03	0.07	.63
Beginning‐of‐Year Math Achievement	0.66 ***	0.02	<.001	0.66 ***	0.03	<.001	0.65 ***	0.02	<.001
Beginning‐of‐Year Math Anxiety	−0.08 **	0.03	.002	−0.08 **	0.03	.001	−0.08 **	0.03	.001
Parent Education				0.02	0.01	.18	0.02	0.01	.27
School Level SES							−0.0003	0.001	.72
df			9(1481)			10(1452)			11(1451)
R2			.71			.71			.71
F			406.48 ***			360.27 ***			327.33 ***
n			1491			1463			1463

Note: Parent Math Anxiety and Parent Homework Help were mean centered before being entered into the model.

Gender: 0 = male, 1 = female; Grade: Kindergarten = 0, Grade 1 = 1, Grade 2 = 2, Grade 3 = 3;*p<.05, **p<.01, ***p < .001.

3.2.4. Moderated Mediation

In contrast to the findings from the matching sample, the indirect relation between parents’ math anxiety and children's math anxiety through children's math achievement was statistically significant at the mean of homework help. However, consistent with the findings in the matching sample replication block, the index of moderated mediation was non‐significant (95% CI = −.000, .001), indicating that the size of the indirect relation between parents’ math anxiety and children's math anxiety through children's math achievement does not differ as a function of how often parents report helping their children with homework—even when considering kindergarten through third‐graders and excluding children who were never assigned math homework from analyses.

4. Discussion

The current study sought to replicate findings from Maloney et al. (2015), in which their results demonstrated a significant interaction effect between the frequency of parents’ homework help and parents’ math anxiety as it predicted children's end‐of‐year math achievement. Specifically, they found that for parents who reported frequently helping their child with math homework, parents’ math anxiety was negatively related to children's end‐of‐year math achievement, which was then in turn, negatively related to children's end‐of‐year math anxiety, consistent with the deficit theory (Carey et al. 2016). Moreover, this effect was found above statistical controls of parent level of education, child gender, grade, beginning‐of‐year math achievement, and beginning‐of‐year math anxiety, and teachers’ math knowledge, teachers’ math anxiety, and school‐level SES. When following the same model building process as Maloney and colleagues (2015), our results did not replicate their primary interaction effect. However, we did find some, albeit inconsistent, evidence for independent, noninteractive negative main effects for both homework help frequency and parent math anxiety, consistent with Maloney et al. (2015).

4.1. Relations Between Parent Math Anxiety, Homework Help Frequency, and Child Math Achievement

In both matching sample and broader sample replication blocks of analyses, parents’ homework help frequency was negatively associated with children's change over time in math achievement: children with parents who reported helping their child with homework more frequently demonstrated fewer gains in math achievement across the year compared to their peers who received less frequent homework help. These findings are consistent with prior meta‐analytic work linking parents’ homework involvement to lower math achievement (Patall et al. 2008) as well as to Maloney and colleagues’ (2015) findings. We find that, based on zero‐order correlations between parent math homework help and children's math achievement at the beginning of the year, children with lower math achievement are more likely to be receiving additional help on homework from their parents than their high achieving peers. However, our finding that parent homework help impedes growth in math achievement, irrespective of the starting levels of achievement, still suggests that homework help is leading to poorer math outcomes regardless of parent math anxiety. This pattern is consistent with prior work suggesting that the negative emotional valence of math homework interactions (e.g., DiStefano et al. 2020) and controlling parenting behaviors during math homework help (Akhavein and Finch 2025; Oh et al. 2022) may impede children's math learning regardless of parents’ math anxiety.

In our matching sample replication, our results indicated that parents’ math anxiety did not predict children's change over the academic year in math achievement. It is worth noting that Maloney and colleagues (2015) did not find this pattern either, as demonstrated by null main effects of parent math anxiety in their primary model, though their null main effect was qualified by homework help frequency. Our broader sample replication block of models, using a full sample of kindergarten through third‐graders, and only selecting those who received homework, did support a significant relation of parents’ math anxiety with children's change over time in math achievement, replicating findings from multiple other studies examining this association (Becker et al. 2022; DiStefano et al. 2020; Herts et al. 2019; Oh et al. 2022).

4.2. Interaction Between Parent Math Anxiety and Homework Help on Child Math Achievement

The key finding of Maloney et al. (2015) was the moderating role of homework help frequency on the association between parents’ math anxiety and students’ gains in math achievement showing that parent math anxiety was only related to child math achievement when parents reported frequently helping with math homework. Neither our matching sample nor our broader sample replication findings were consistent with the pattern of results found by Maloney and colleagues (2015). Even with our large sample, we consistently did not detect a statistically significant moderation effect across many iterations of the model and sample decisions.

As discussed in the previous section, our results more consistently show two independent roles for parent math anxiety and parent homework help frequency such that, in our broader sample (not the matching sample), parent math anxiety is negatively related to children's growth over the school year across all levels of homework help. Similarly, homework help frequency is negatively related to children's math achievement regardless of parent math anxiety levels. Thus, our findings showed some evidence that these relations are occurring, but they do not rely on one another or suggest a specificity to the relations.

It is important to consider discrepancies between the original research and our conceptual replication in terms of contextual and methodological differences between the studies (See Table S1 for details about similarities and differences). For example, Maloney et al. (2015) collected data shortly following the implementation of Common Core State Standards (CCSS) in the state of Illinois, which introduced some math concepts earlier and emphasized higher cognitive demand (Clements et al. 2017; Dingman et al. 2013; Porter et al. 2011). Comparatively, Florida was using the CCSS‐based Florida Standards during our study period (2017‐2019), but the political climate around CCSS differed across states. Accordingly, families and caregivers were observed voicing concerns and negative attitudes toward CCSS on social media websites (Otten and de Araujo 2015), perhaps due to the perceived difficulty they faced when assisting their children with “Common Core Math Problems”. In Florida, public criticism of CCSS was extreme enough that it led to the replacement of these standards in 2020. Therefore, these contextual differences, such as variation in parent familiarity with curricula or attitudes about Common Core math, may have shaped parent attitudes and their average comfort with helping their children with math homework. This may have resulted in less effective homework help across our whole sample as opposed to only for moderate to high‐math anxious parents as was found by Maloney et al. (2015).

In addition, in the current study, many of the measures used were different for both parents and children, which could impact results. For example, Maloney et al. (2015) used the Applied Problems subtest from the norm‐referenced Woodcock‐Johnson III Tests of Achievement (Woodcock et al. 2001), and the current study used the Elementary Student Mathematics Assessment (EMSA; Schoen et al. 2021), which includes similar items to the Woodcock‐Johnson Applied Problems test, but also includes basic computations and algebraic reasoning. In addition, the Woodcock‐Johnson assessment was individually administered by a researcher, whereas the EMSA was administered to the whole class by the students’ teacher. Differences in how parent math anxiety and homework help interact to impact math achievement could be attributed to the contents and administration of each measure. Additional measures differed, including the parent and child math anxiety measures, which could pick up on different aspects of these constructs.

4.3. Homework Help as a Moderator of the Indirect Relation Between Parent Math Anxiety and Child Math Anxiety Through Child Math Achievement

The test of the moderating role of homework help frequency on the indirect relation between parent math anxiety and child math anxiety through child math achievement resulted in null findings as well. This is not surprising given the already null interaction effect from the initial models. Interestingly, in the matching sample, there was no evidence for the indirect relation at all and for the broader sample, there was a significant indirect relation, but it was not dependent on homework help frequency, consistent with other work showing a relation between parent and child math anxiety (Casad et al. 2015; Maloney et al. 2015). Both our study and Maloney et al. (2015) measure math achievement and math anxiety at the same time, thus not establishing temporal precedence for these variables. Therefore, results from a mediation analysis should be interpreted with caution, especially in light of reciprocal theories of children's co‐development of math anxiety and math achievement (Carey et al. 2016). Indeed, as these reciprocal theories suggest, the co‐development of math anxiety and achievement likely unfolds over time in ways that a mediation over a single year cannot fully capture. Thus, there remains a need for research that explicitly tests the associations between parent math anxiety, homework help, and children's math anxiety and achievement using additional waves of data collection to more effectively tease these causal relations apart.

4.4. Limitations

Although one strength of the current study was a relatively larger, more diverse sample than that of Maloney et al. (2015), our findings are only generalizable to the state of Florida in the USA and the historical time period from which our data originate (e.g., pre‐COVID, before the current state standards). Another limitation of the current study, and Maloney et al. (2015), was the examination of frequency of homework help, rather than indicators of quality or duration of parent–child homework help interactions—such as parents’ controlling behaviors (Akhavein and Finch 2025; Bronstein et al. 2005; Oh et al. 2022). As main effects from the current sample demonstrated a negative relation between homework help frequency and children's math achievement, it remains unclear how parents’ homework helping behaviors may play a role in the socialization of mathematics cognition.

There are also some measurement limitations in the current study. Parent math anxiety was assessed via a single question (Nuñez‐Peña et al. 2014), whereas Maloney et al. (2015) benefitted from using a 25‐item index (Alexander and Martray 1989). While this single‐item measure has demonstrated strong associations with longer measures of math anxiety (Hart and Ganley 2019; Nuñez‐Peña et al. 2014), it may nonetheless limit the precision with which parents' math anxiety is captured. As math anxiety is a large, broad construct, future studies may benefit from assessing how parents exhibit (or do not exhibit) math anxiety during homework help. In addition, given the developing reading skills of kindergarten students, there were some decisions that were made to ensure that the measures were developmentally appropriate that required different measures (reading achievement) or response scales (math anxiety) for the kindergarten students. Thus, there may be comparability issues when combining across grade levels for the analyses, including these measured for the broader sample.

4.5. Implications and Future Directions

Unlike the work of Maloney et al. (2015)—that did not report information about the racial‐ethnic background of students—the current study's findings can be generalized to families of various cultural backgrounds given the racial diversity of the children in this study. Indeed, some researchers have explored differences in parents’ attitudes and involvement with regard to children's schooling across different racial‐ethnic groups, as these attitudes and behaviors originate from parents’ perceptions of home–school connections (e.g., parent–educator communication, school as a resource, school‐readiness practices) as a function of cultural capital (Miller et al. 2014). Accordingly, one next step for math anxiety researchers may lie in understanding the prevalence of math anxiety in marginalized communities—and its subsequent role in children's development—as caregivers in these communities may receive less support from schools due to diminished cultural capital.

Examining not only frequency of homework help but also the quality of these parent–child interactions would provide important insights into how parents can best support their children in their math learning and achievement. The negative relation between homework involvement and math achievement observed here, in Maloney et al. (2015), and in prior work (Patall et al. 2008) suggests that frequency alone is insufficient to understand how parent homework help shapes children's math learning. Findings from previous studies suggest that poor quality interactions occurring during homework help (DiStefano et al. 2020, Oh et al. 2022) may result in reduced math learning and children developing negative associations with math. Future studies could implement experimental designs to fully understand the causal role of parents’ homework help on children's math outcomes, such as training parents in the use of autonomy‐supportive parenting practices within the context of math homework (Oh et al. 2022) or increasing educators’ communication with parents to better equip caregivers to help their children with homework (Lin et al. 2019). Moreover, examining a broader scope of parental involvement with regard to children's mathematical development—in addition to parental math anxiety—would provide a more nuanced understanding of the intergenerational transmission of mathematical attitudes and beliefs. Indeed, a recent meta‐analysis by Daucourt et al. (2021) revealed that parents’ attitudinal traits related to math (e.g., attitudes, expectations, and anxiety) have been shown to play a more pronounced role in children's math achievement compared to other forms of parent involvement.

Regarding other pathways that may underlie the intergenerational transmission of math anxiety, some research has highlighted the role of genetic predispositions underlying math anxiety. Wang et al. (2014) found that 40% of the variance in math anxiety from a sample of twins could be explained by genetic factors and 9% of the total variance in math anxiety could be attributed to genetic predispositions for general anxiety. Moreover, multivariate analyses revealed that children's math anxiety was not only predicted by independent genetic factors associated with math‐based problem solving, but also by nonfamilial environmental risk factors. In line with theories emphasizing the interaction between genetic, neural, behavioral, and environmental influences on human development (Gottlieb 1991; van Bergen et al. 2014), future work would benefit from exploring gene–environment interactions regarding the development of math cognition and anxiety (Rutter et al. 2006). In addition, future work should always consider that parents not only impact their children's outcomes through socialization processes, but also by sharing genetic influences (Hart et al. 2021). Therefore, any correlation between parent behaviors and their children's outcomes cannot be concluded to be simply due to environmental transmission.

Implications of the current study's findings are applicable to parents’ efforts to support their children's math learning and achievement. Understanding the role of parental homework help in children's math achievement may provide a clearer path for parents who wish to assist in improving their children's math grades and learning. Similarly, knowing more about how parental math anxiety may affect their children's math learning will allow math‐anxious parents to better help their children academically.

5. Conclusion

The current study's findings did not replicate the main findings of Maloney et al. (2015)—namely, we found no evidence that parent math anxiety related negatively to children's math achievement only when they helped with homework. We also did not replicate their finding that the mediation pathway from parent math anxiety to child math anxiety through student math achievement was only significant for those with high frequency of homework help. We did, however, find an overall negative relation between parent homework help and children's later math achievement; similar to what Maloney et al. (2015) found for solely higher math anxious parents, we found to be true for all parents. We also found evidence for an overall relation between parent math anxiety and children's math achievement in our broader sample, but not our matching sample. This study was able to further test previously accepted findings with a larger and more diverse sample as well as a sample with a broader age range and with only children who receive homework. The primary results were consistent across these sampling choices, highlighting the robustness of these findings. Given the nature of the similarities and differences between the current study's findings and those presented by Maloney et al. (2015), additional research is necessary, especially those that employ naturalistic observation or qualitative approaches. Our paper provides a starting point for researchers to delve deeper into the complex interplay between parents’ math attitudes and anxiety, the nature of their interactions during homework help, and children's early development of mathematical competencies.

Author Contributions

Elyssa A. Geer: writing – review and editing, investigation, methodology. Sara A. Hart: writing – review and editing, conceptualization, methodology, data curation, supervision, formal analysis. Madison B. Poisall: formal analysis, writing – original draft. Connie Barroso: writing – review and editing, investigation, methodology. Colleen M. Ganley: conceptualization, methodology, investigation, funding acquisition, writing – review and editing, formal analysis, project administration, data curation, supervision, visualization, resources, validation. Olivia K. Cook: writing – original draft, formal analysis, data curation, supervision. Rachel A. Conlon: writing – review and editing, investigation.

Funding

This work was supported by the Institute of Education Sciences, U.S. Department of Education, through Grant R305A170463 to Florida State University. The opinions expressed are those of the authors and do not represent the views of the Institute or the U.S. Department of Education. Research reported in this publication was supported by the Eunice Kennedy Shriver National Institute Of Child Health & Human Development of the National Institutes of Health under Award Number P50HD052120. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This research was undertaken, in part, thanks to funding from the Canada Excellence Research Chairs Program to Sara A. Hart.

Conflicts of Interest

The authors have no conflicts of interest to report related to the research in this manuscript.

Supporting information

Supporting information: desc70261‐supp‐0001‐SuppMat.docx

Data Availability Statement

All data, code, output, and corresponding notes are available on LDbase at https://ldbase.org/projects/ef459f47-897f-4610-916a-0771f1c9900c.