Can Writing Attitudes and Learning Behavior Overcome Gender Difference in Writing? Evidence From NAEP

Abstract

Based on eighth-grade writing assessment data from the 1998 (N = 20,586) and 2007 (N = 139,900) National Assessment of Educational Progress (NAEP), this study examines the relationships among students’ writing attitudes, learning-related behaviors, and gender in relation to writing performance. Overall, the effects of attitudes were slightly larger than the effects of learning behaviors on writing performance, and gender differences were more prominent in attitudes than learning behaviors related to writing. Perhaps the most surprising finding from the 2007 NAEP data was that females with the most negative attitudes toward writing outperformed males with the most positive attitudes (i.e., writing scores based on two measures of attitudes: females, 157 and 161; males, 151 and 149). Overall, a similar pattern was observed with learning behaviors and gender differences in writing scores. Furthermore, medium effect sizes of gender difference in writing scores (females scoring substantially higher than males) were present even though the students reported to be at the same level in terms of writing attitudes and learning behaviors. The present study demonstrates that gender disparity in students’ writing performance is persistent and strong; it cannot be explained by gender differences in attitudes or behavior alone or in attitudes and behavior combined.

Keywords

National Assessment of Educational Progress self-concept usefulness of writing interest in writing time on homework large-scale assessment

Researchers in recent years have called for the need to examine motivational factors to explain why many students fail to acquire writing proficiency (e.g., Bruning & Horn, 2000; Graham, Berninger, & Fan, 2007). Two recent meta-analytic studies revealed that the majority of school-based writing instruction was for process and scaffolding of writing and explicit teaching of writing skills and strategy (Graham & Perin, 2007; Rogers & Graham, 2008). In fact, among nine writing treatments identified as “effective” in Rogers and Graham’s meta-analysis (2008), no attitudinal or motivational factors were included except for goal setting. Similarly, while it was noted that schools’ writing instruction has improved over the last 25 years with respect to explicit teaching of writing skills and strategies for various subject areas (Applebee & Langer, 2009, 2011), there was no mention of how successful schools have been in nurturing students to develop deep interest in composition requiring extended writing.

There has been, however, compelling empirical evidence that supports the importance of self-regulatory behaviors and motivation in students’ writing development. On the basis of a meta-analysis of 123 documents, Graham and Perin (2007) reported that among 17 broad categories of writing intervention treatments, the largest effect size (with the weighted mean effect size of 1.14) was found in the instruction strategies designed to encourage “self-regulated writing” (i.e., teaching with emphasis on self-regulatory aspects of writing, such as goal setting, monitoring, reinforcing, and managing strategies). In spite of such strong empirical evidence, we have relatively little knowledge about the extent to which such motivational, attitudinal, and behavioral factors potentially influence writing competence (cf., Bruning & Horn, 2000). There are very few studies that take into account a wide range of students’ attitudinal and behavioral tendencies in relation to writing performance, which is the focus of the present study.

The main variables of the present study are student-level attitudinal and behavioral factors in writing achievement. Much of the research on these factors has been done in the domain of reading and mathematics learning. Thus, a reasonable starting point is to identify key student-level variables that are known to have a strong relationship to students’ reading and mathematics learning (or academic achievement in general) and then examine them in relation to writing achievement. A recent review article (Lee & Shute, 2010) integrated findings from over 60 years of research on achievement-related psychological constructs and identified two student-level variables as the most relevant in explaining academic achievement: a group of motivational/attitudinal variables, on one hand, and a group of learning behavioral variables, on the other. The present study adopts this framework and employs attitudes, typically measured by motivation or interest (e.g., Pintrich & De Groot, 1990), and learning behaviors (e.g., Zimmerman, 1990) as the key student-level factors. Additionally, students’ demographic information, such as gender, race/ethnicity, and parents’ education level, is considered in terms of contextual variables.

Specifically, this study has three main objectives. First, it examines the roles that attitudinal and behavioral factors play in students’ writing performance. To provide a more complete picture, demographic and language factors are also examined. The second objective is to investigate gender differences in writing. Girls’ higher writing achievement has been widely recognized in large-scale and locally based empirical studies (Organization for Economic Cooperation and Development, 2007; Salahu-Din, Persky, & Miller, 2008). Gender is probably the most extensively studied demographic variable in relation to writing performance. Nonetheless, writing researchers have only scratched the surface of why this problem persists. The third objective of this article is to further investigate the interrelationships between attitudinal and behavioral factors and gender with respect to writing performance. The next few sections summarize the research findings concerning relationships between these factors and writing development among school-age students.

Attitudes and Writing Performance

One of the major concerns expressed by writing educators is that students generally have negative attitudes toward writing (cf. Graham et al., 2007). Students tend to express anxiety, fear of failure, or a feeling of lack of control when facing writing tasks (Bruning & Horn, 2000). Writing apprehension (i.e., wishing to avoid) is a common feeling for both boys and girls (Pajares & Valiante, 2001). Even highly competent students experience some level of stress and anxiety when writing (Bruning & Horn, 2000).

Many studies indicate that students with more positive attitudes toward writing tend to show better writing skills (e.g., Bruning & Horn, 2000). Students who are interested in writing tend to be less frustrated when facing challenges in writing tasks, welcome constructive feedback, and challenge themselves to do many writing tasks (Lipstein & Renninger, 2007). Several studies also showed that students’ perceived usefulness of writing is closely related to writing achievement (e.g., Lipstein & Renninger, 2007). When students perceive writing as a personal tool to express themselves, they are motivated to write more (Bruning & Horn, 2000), observe their writing progress metacognitively, and tend to set specific goals (Pintrich & De Groot, 1990).

Self-belief (e.g., self-efficacy or self-concept) in writing is another salient component of writing attitudes. One’s self-belief is associated with affects such as enjoyment, intense anxiety, or apprehension before or during writing (Clark & Dugdale, 2009; McCarthy, Meier, & Rinderer, 1985). Several empirical studies have shown strong positive associations between self-belief in writing and writing scores. For instance, writing self-efficacy was the only motivational construct (e.g., Pajares & Valiante, 1999) and the strongest variable (McCarthy et al., 1985) in predicting students’ writing performance. In another study, writing self-efficacy was the only significant predictor of writing performance in a model that included the perceived usefulness of writing, self-regulation, and writing apprehension (Pajares, Miller, & Johnson, 1999). Thus, students’ attitudes toward writing seem to be manifested in the level of interest, perceived usefulness, and self-beliefs in writing, with self-belief variables potentially having stronger relations than the other two attitudinal variables to writing performance.

Learning Behavior and Writing Performance

Researchers have linked students’ writing achievement to various learning-related behaviors. For instance, a close relationship was found between students’ tendency to read for fun and writing performance at school (e.g., Clark & Foster, 2005), especially in vocabulary and grammar test scores (Krashen, 1995). The amount of reading done outside of school has also been related to students’ writing performance; generally, regular readers were better writers (Krashen, 1995).

Students’ learning environment at home (e.g., time spent watching television or discussion with parents about schoolwork) was also found to predict writing performance. Among many other dimensions of parental involvement variables (e.g., communication with the school, participation in school activities, and supervision at home), discussion with one’s children about schoolwork was the strongest predictor of students’ academic achievement (Ho & Willms, 1996), including writing performance (Fantuzzo, McWayne, Perry, & Childs, 2004). Cooper and associates (Cooper, Lindsay, Nye, & Greathouse, 1998) also showed that students’ report on the amount of homework they did and parents’ report on the amount of time their children spent on homework were both positively correlated to students’ writing performance in a statewide assessment. Comstock (1995) reported, on the basis of more than 500,000 U.S. sixth and twelfth graders, a negative relationship between students’ writing achievement and the amount of television viewed per day.

Students’ writing competence has also been positively related to their tendency to use the computer to write or communicate via e-mail (e.g., Goldberg, Russell, & Cook, 2003). The majority of young people have favorable views about using computers for writing, with the reported benefits of better concentration, ease of constructing longer messages, and greater creativity (Clark & Dugdale, 2009). A meta-analysis by Goldberg et al. (2003) analyzed 99 articles published between 1992 and 2002 and concluded that students’ use of computers was linked to an increased number of revisions and better quality and greater quantity of writing. Shang (2007) also points out that writing e-mail with classmates is related to improvements in grammatical accuracy and sentence complexity among learners of English as a foreign language. Competent writers see writing as personal communication with others and enjoy using a variety of methods to write (Bruning & Horn, 2000). Thus, writing e-mail can be seen as one vehicle for students to develop their personal voices and share their views with others.

Gender and Writing Performance

Higher literacy skills among girls are shown in most countries (Organization for Economic Cooperation and Development, 2007) as well as in English-speaking countries (Salahu-Din et al., 2008). Girls also have more positive attitudes and feelings toward writing, in comparison to boys (e.g., Graham et al., 2007; Pajares & Valiante, 1999). They like and value writing more than boys do. Yet, males tend to perceive little value or usefulness in writing and thus are less satisfied with writing activities (Hansen, 2001). Different patterns of achievement goal orientations have also been found, with males showing performance goal orientations and females having mastery goal orientations for writing (Pajares & Valiante, 2001).

Studies on gender differences have been inconclusive, however, in terms of attitudes involving self-evaluation (e.g., self-efficacy). Girls generally believe that their writing styles are more valued in school than are boys’ (Hansen, 2001) and that they can write better than their male classmates (Pajares et al., 1999). However, when boys and girls have expressed their own levels of self-efficacy in writing, the results have been mixed. For instance, no difference was found in the writing self-beliefs between boys and girls (Hansen, 2001; Pajares & Valiante, 1999), especially after controlling for writing aptitudes (Pajares et al., 1999). Pajares and Valiante (2001), however, reported stronger writing self-concept and writing self-efficacy among girls. In general, girls have substantially higher writing scores than boys, but their confidence levels do not necessarily match their ability levels; both males and females tend to overestimate their abilities, but the differences between abilities and confidence are much greater for males (see Stankov & Lee, 2008).

As for learning behavior, girls tend to demonstrate more “desirable” behaviors, which are also associated with better writing performance. For example, girls are more likely than boys to read for fun in their own time and talk about what they read (Clark & Foster, 2005), as well as discuss their schoolwork with parents (Fantuzzo et al., 2004). Females also tend to engage more in using a variety of writing platforms, including blogging, writing e-mail and text messages, and utilizing social network sites (e.g., Clark & Dugdale, 2009). However, boys tend to spend less time doing homework and more time watching television (Vereecken, Todd, Roberts, Mulvihill, & Maes, 2005), both of which have been negatively associated with academic achievement, including writing (Comstock, 1995; Cooper et al., 1998).

Purpose of This Study

Research thus far has established independent relationships of gender, attitudes, and behavior to writing performance. However, relatively little information is available about how these factors interdependently function in relation to students’ writing performance. For instance, studies have indicated that gender is a strong factor in writing attitudes and writing performance and that writing attitudes (including self-beliefs) tend to significantly relate to writing performance. Many studies with such findings imply that gender differences in writing attitudes may partly explain the persisting gender gap in writing. This suggestion has yet to be tested by empirical data. The present study investigates this issue along with other questions:

What are the relationships among gender, attitudes, and behaviors in the context of students’ writing performance? For example, are the patterns of attitudes and behavior consistent across gender?

What are the relative strengths of relationships between writing performance and attitudes, behavior, and gender on the National Assessment of Educational Progress (NAEP) assessments? For example, would writing performance have stronger associations with either attitudes or behavior? Which aspects of attitudinal or behavioral indicators are important in writing performance? Is the relationship between writing and attitudes or behavior particularly stronger for either gender group?

How important are attitudinal or behavioral factors in writing performance in the broader context, for example, in comparison to the relationships between demographic and language factors to writing performance?

Can gender disparities in writing performance be attributed to gender differences in attitudes and behavior? In other words, does the relationship between gender and writing achievement remain strong above and beyond attitudinal/behavioral factors?

In sum, this study focuses on gender and attitudinal and behavioral factors and their relationship to writing performance in the NAEP assessment.

Method

Data Sources

This study employed eighth-grade data from the 1998 (N = 20,586) and 2007 (N = 139,900) NAEP writing assessments. The NAEP uses nationally representative samples. Thus, the analysis based on NAEP data produces sample-based nationally representative estimates. In this sense, population estimates based on NAEP samples can be generalized to students in a given grade level living in the United States. Furthermore, the NAEP employs a multistratified sampling technique to ensure nationally representative proportions for given subgroup populations at a grade level. The distributions of student subgroups (by gender, race/ethnicity, parental education levels, and other demographic factors) are proportional to the national subgroup populations; thus, the NAEP results presented in this study can be generalized to the particular subgroups of eighth graders in the United States. In typical U.S. public and private school systems, eighth grade is in the upper grade of middle school.

Variables

Writing achievement

The writing test scores of the NAEP 1998 and 2007 assessments were used as the measure of writing achievement. The metrics for the NAEP writing scores are in the range of 0 and 300, with a mean of 150 and a standard deviation of 35. The observed mean scores at Grade 8 were 150 in 1998 (Greenwald, Persky, Campbell, & Mazzeo, 1999) and 156 in 2007 (Salahu-Din et al., 2008). Both the 1998 and 2007 writing assessments consisted of 20 writing tasks, with 7 narrative, 7 informative, and 6 persuasive purpose writing styles. Each student was given 50 minutes for either one or two (shorter) writing tasks.

Explanatory statements/variables

The main explanatory variables/statements in this study are gender and four attitudinal and eight behavioral statements. This information was gathered from the questionnaire that students have to complete as part of the NAEP assessments. The NAEP student questionnaire asks about students’ demographics and other relevant factors for learning and classroom practices. The attitudinal questions asked students to (strongly) agree or (strongly) disagree with statements such as “Writing helps me share my ideas” and “Writing things like stories or letters is one of my favorite activities” (2007 assessment) and whether students think that they are “good at writing” and whether they “like to write” (1998 assessment). Students’ engagement in learning behaviors employed in this study asked students how often they “read for fun in their own time,” “talk with friends or family about something they have read,” and “write e-mail to friends or family” and “About how many pages a day do you have to read in school and for homework?” (2007 assessment). An additional four behavioral questions drawn from the 1998 writing assessment included the amount of time per day spent on homework, whether students discussed their studies at home, the amount of television or video watched on a school day, and the number of pages read in school and for homework. Students were asked to select one of four or five response categories for each attitudinal and behavioral statement. It was expected that higher writing scores would be associated with the response categories indicating more positive responses (e.g., agree than disagree) and more desirable responses (e.g., doing every day than never or hardly ever) on the positive aspects of attitudes and behavioral statements.

Also examined in relation to the NAEP writing scores were demographic variables (parental education level, race/ethnicity, and eligibility for the National School Lunch Program¹) and language factors (students’ status as English-language learners and use of a language at home other than English). The information relating to race/ethnicity was based on school records (and not student responses on the questionnaires). The appendix presents the response categories provided for the attitudinal and behavioral statements and students’ demographic background information.

Analysis

The NAEP oversamples schools with a high concentration of certain subgroups of students to make sure that the data have sufficient numbers of students in particular subgroups. This oversampling needs to be adjusted by sampling weights, which adjust for the nonresponses of schools and students (Hahs-Vaughn, 2005) and the lower sampling rates of students attending small nonpublic schools (Lee, Grigg, & Dion, 2007). The NAEP Data Explorer is a web-based data analytic tool for the NAEP data. It produces NAEP estimates with the sampling weights so that the results accurately represent the estimates of the respective population subgroups (see Hahs-Vaughn, 2005, for a discussion of sampling weights). The NAEP Data Explorer was used for the analyses reported in this study.

Interpreting NAEP results

Due to the large sample size of the NAEP data, a small score difference is likely to be statistically significant at the conventional statistical significance level of .05 (Salahu-Din et al., 2008). Thus, statistical significance test results on the NAEP data do not have the same type of functionality as they would when employed in other “typical” sample sizes of social science analyses. In the NAEP, a score difference of 1 or 2 points is likely to be statistically significant at a 0.05 level for the total sample group, and a score difference of 3 or 4 points is likely to be statistically significant between two subgroup populations (e.g., by gender). Thus, practical (standardized or unstandardized) group score differences (e.g., expressed by Cohen’s d) are the more appropriate statistics for the NAEP data. In fact, the results reported in this article are all statistically significant at a 0.05 significance level (since the group score differences were about 10 to 25 points); thus, their significance test results are not reported.

Effect size

An effect size is a statistical measure to indicate the strength of the relationship between two variables in the sample-based estimates (Cohen, 1988, 1992). It represents the estimated magnitude of the relationship without inferring a “true” relationship in the population. It is highly recommended that effect sizes be reported for empirical findings in the social sciences (Wilkinson & APA Task Force on Statistical Inference, 1999). It may be particularly appropriate for presenting the NAEP results, since the analysis focus of the present study is not so much on whether the observed relationships can be applied to a “wider” population or have occurred due to chance (i.e., tests for significance) but to generalize findings to U.S. eighth graders in general. Effect sizes can be biased estimates unless the data were sampled in a way that is appropriate for presenting the observed findings and inferring implications accordingly. In this sense, effect size estimates are suitable statistics for the data, since the NAEP employs careful and elaborate stratified sampling.

The analyses reported in the present study are based on a standardized measure of effect, represented by Cohen’s d values, and an unstandardized measure of effect, calculated by the raw differences between subgroup means. Cohen’s d is defined as the difference between two means divided by a standard deviation of either population (Cohen, 1988) or by a pooled standard deviation of two independent samples (Hartung, Knapp, & Sinha, 2008). Cohen’s d effect sizes applied to the percentages is calculated as the difference between two proportions divided by a pooled standard deviation of two independent samples. Thus, the Cohen’s d estimates based on the same size of difference in the means or proportions will differ depending on the pooled standard deviation of a particular “pooled” subgroup. Because NAEP data are based on large sample sizes, the pooled standard errors were used in this study, which are available on the official NAEP Data Explorer website (http://nces.ed.gov/nationsreportcard/naepdata/dataset.aspx). Cohen’s d values typically range between 0 and 2, although it can reach infinity. A Cohen’s d size of 0.2 to 0.3 represents a small effect are considered large effects (Cohen, 1988).

Another way to interpret these findings is to look at the subgroup (writing) scores associated with their responses to the given (attitudinal or behavioral) statements and then compare them to the cutoff scores in the NAEP criteria (i.e., NAEP achievement levels and rating criteria). The NAEP assessment defines three achievement levels: basic, proficient, and advanced. The cutoff scores for each achievement level in the 2007 writing assessment were 114, 173, and 224, respectively (Salahu-Din et al., 2008). The NAEP also distinguishes students’ writing ability according to five levels of rating criteria: insufficient (77), uneven (106), sufficient (147), skillful (213), and excellent (254). Students performing in the middle range of the basic level are likely to produce a “sufficiently” good piece of writing, but only the top range of the proficient level is able to demonstrate “skillful” writing (Salahu-Din et al., 2008). If, for example, the male students who strongly agreed that they like to write scored 147 in the 2007 assessment, they could not reach the NAEP’s proficient writing achievement level in spite of their report of liking of writing.

Results

Attitudes Toward Writing and Gender

Table 1 presents the percentages of the total group and male and female students who answered positively (strongly agree or agree) on the four statements about attitudes toward writing. Overall, students demonstrated lukewarm attitudes toward writing: not positive but not particularly negative either. About half the students showed positive attitudes in three of four statements. Although a relatively small portion of the students said that writing was one of their favorite activities (35%), many of them understood that writing is helpful for sharing their ideas with others (61%).

Table 1.

Percentages of Students With Positive Attitudes Toward Writing: National Assessment of Educational Progress, 1998 and 2007.

Attitude	Total	Male	Female	Cohen’s d on Gender
Like to write (1998)	52	38	65	0.5
Good at writing (1998)	51	46	55	0.2
Writing helps share ideas (2007)	61	53	70	0.3
Writing is one of my favorite activities (2007)	35	22	46	0.5
Average Cohen’s d				0.4

Note: Percentages are based on a nationally representative sample and thus close to population estimates for U.S. eighth graders.

Responses to the attitudinal statements showed a clear pattern that the female students expressed more positive attitudes toward writing than their male counterparts (Table 1). This gender difference showed small to medium effect sizes (Cohen’s d values ranging from 0.2 to 0.5). The liking component (“Like to write” and “Writing is one of my favorite activities”) showed the largest gender differences in both years of assessment (d = 0.5). Yet, there was a less prominent gender difference on the attitude concerning self-evaluation (i.e., “Good at writing,” d = 0.2), although a greater number of female students (55%) showed a positive writing self-concept.

Learning Behaviors and Gender

Table 2 presents the percentages of students who reported to be at the lowest (i.e., most negative; denoted as L in the table) and highest (i.e., most positive; denoted as H) response categories for the eight types of learning behaviors. As many as 41% of the students (in the total group) reported having discussions at home with their parents about their schoolwork almost every day, and 38% of them reported writing e-mail to their friends or family almost every day. However, 37% reported that they never talked with their friends about what they read, and 33% of them reported that they never read for fun at all.

Table 2.

Percentages of Students Reporting the Lowest and Highest Categories on Learning Behaviors: National Assessment of Educational Progress, 1998 and 2007.

	Total		Male		Female		Cohen’s d on Gender
Learning Behavior	L	H	L	H	L	H	L	H
Discussion of studies at home (1998)	20	41	24	36	16	45	0.2	0.2
Pages read in school and for homework (1998)	31	15	33	15	29	14	0.1	0
Time per day on homework (1998)	7	28	8	22	6	33	0.1	0.2
Amount of television/video watched on school day (1998)	13	3	15	4	12	3	0.1	0.1
Write e-mails to friends or family (2007)	26	38	33	29	18	48	0.3	0.4
Talk with friends about what you read (2007)	37	9	42	7	31	11	0.2	0.1
Read for fun in your own time (2007)	33	19	41	14	26	24	0.3	0.3
Pages read in school and for homework (2007)	28	19	30	19	27	19	0.1	0.0
Average Cohen’s d							0.2	0.2

Note: Percentages are based on a nationally representative sample and thus close to population estimates for U.S. eighth graders. The numbers under the L and H columns represent the percentages for those who endorsed the lowest (the most negative, the least desirable) and highest (the most positive, the most desirable) categories for a given statement (see appendix).

Overall, the females were engaged in more “desirable” learning behaviors than their male counterparts, showing small to medium effect sizes in five of eight learning behavioral statements (see the last two columns of Table 2). The few exceptions where there were no or little gender differences involved the number of pages read in school and for homework and the amount of television or video watching (d = 0 and 0.1). Yet, the gender differences were fairly salient on e-mail writing frequencies (d = 0.3 and 0.4) and reading for fun (d = 0.3). For instance, as many as 41% of the male students reported that they never read for fun; it was only 26% for the female students, yielding a medium Cohen’s effect of 0.3. The gender differences on behavioral factors were fairly consistent for the least and most “desirable” behavioral groups with an average Cohen’s d value of 0.2 across the eight statements, meaning that male-female differences on the pattern of learning behaviors were apparent regardless of their level of behavioral engagement in learning.

Writing Achievement, Attitudes Toward Writing, and Gender

The 1998 and 2007 writing scores by gender are shown in Figure 1. As can be seen, there was a gender score gap of 20 points in both years, with girls demonstrating higher writing ability. The NAEP 2007 writing scores were calculated for the students who either strongly disagreed or strongly agreed on the four attitudinal statements (Table 3). Their associated Cohen’s d values are also presented. As expected, students who reported more positive attitudes toward writing demonstrated better performance in writing. The effect sizes of attitudes on writing performance varied across the aspects of attitudes being asked, but they were mostly in medium and large effects (see the columns for “Cohen’s d on Attitude” in Table 3). Self-concept in writing (“good at writing”) showed the largest effect (d = 0.9 on the total group) on the writing scores, whereas students’ endorsement of whether writing was one of their favorite activities showed the smallest effect (d = 0.5 on the total group). This pattern of effect sizes was consistent in each gender group, meaning that the relationships between writing attitudes and writing performance were fairly strong within each gender group, although the effects were slightly stronger for the female students (with differences in the Cohen’s d of 0.1 and 0.2).

Figure 1.

National Assessment of Educational Progress 1998 and 2007 writing scores by gender.

Table 3.

Writing Scores by Attitudes Toward Writing: National Assessment of Educational Progress, 1998 and 2007.

	Total			Male			Female			Cohen’s d on Gender
	SD	SA	d	SD	SA	d	SD	SA	d	SD	SA
Like to write (1998)	128	160	0.8	125	147	0.6	138	166	0.8	0.4	0.5
Good at writing (1998)	120	157	0.9	116	148	0.9	130	165	0.9	0.4	0.5
Writing helps share ideas (2007)	144	165	0.6	139	151	0.3	157	172	0.4	0.5	0.6
Writing is one of my favorite activities (2007)	149	165	0.5	144	149	0.2	161	171	0.3	0.5	0.6
Average Cohen’s d on attitude			0.7			0.5			0.6	0.5	0.6

Note: SD = strongly disagree; SA = strongly agree; d = Cohen’s d on attitude.

The effect sizes of gender on the writing scores are also presented in the last two columns of Table 3 (under “Cohen’s d on Gender”). These effect sizes were based on the writing score differences between the male and female students who reported being on the same categories (i.e., strongly disagree or strongly agree) on the attitudinal statements. As can be seen, there was a medium effect size for gender, favoring the female students, on the writing scores (d = 0.4-0.6). In other words, the female students still scored substantially higher than the male students, although they reported to be on the same level of writing attitudes. It appears that the male students did not transfer their positive attitudes to their writing performance as effectively as the females. The gender effects on writing performance were slightly stronger for those with positive attitudes (an average Cohen’s d of 0.6) than for those with negative attitudes (an average Cohen’s d of 0.5). Also noteworthy was that the female students with negative attitudes outperformed their male counterparts with positive attitudes in the 2007 data (i.e., writing scores based on two measures of attitudes: females, 157 and 161; males, 151 and 149; p < .01).

Writing Achievement, Learning Behaviors, and Gender

The NAEP 2007 writing scores are presented for the students who were at the lowest and highest categories on the eight learning behavioral statements (Table 4). As expected, substantially higher scores were found for students who reported to be engaged in “desirable” behaviors more frequently. The amount of time per day spent on homework showed the largest effect on writing performance, with a Cohen’s d of 0.9 (on the total group), followed by whether students read for fun in their own time (0.7) and then by whether they discussed schoolwork at home (0.6). This pattern was fairly consistent across genders. In addition, there was a fairly large effect of the amount of time spent watching television on the writing scores for females only (0.7).

Table 4.

Writing Scores by Learning Behaviors: National Assessment of Educational Progress, 1998 and 2007.

	Total			Male			Female			Cohen’s d on Gender
Behavior	L	H	d	L	H	d	L	H	d	L	H
Discussion of studies at home (1998)	137	159	0.6	130	149	0.6	147	166	0.6	0.5	0.5
Pages read in school and for homework (1998)	142	155	0.4	133	145	0.4	152	165	0.4	0.6	0.6
Time per day on homework (1998)	129	160	0.9	123	150	0.8	142	167	0.8	0.6	0.5
Amount of television watching (1998)	132	150	0.5	126	139	0.4	141	164	0.7	0.5	0.6
Write e-mails to friends or family (2007)	149	159	0.3	144	147	0.1	159	167	0.2	0.4	0.6
Talk with friends about what you read (2007)	148	161	0.4	141	147	0.2	158	169	0.3	0.5	0.6
Read for fun in your own time (2007)	147	172	0.7	140	159	0.6	157	179	0.6	0.5	0.6
Pages read in school and for homework (2007)	147	162	0.4	137	152	0.4	157	172	0.4	0.6	0.5
Average Cohen’s d on behavior			0.5			0.4			0.5	0.5	0.6

Note: The numbers under the L and H columns represent the scores for those who endorsed the lowest (the most negative, the least desirable) and highest (the most positive, the most desirable) categories for a given statement (see appendix). d = Cohen’s d on Behavior.

Table 4 also presents the Cohen’s d values for gender effects on writing scores among those who reported the same categories (i.e., the least or most desirable) of the learning behaviors (see the last two columns of Table 4). Large gender gaps in the writing scores in favor of females (approximately 15 to 20 score points with effect sizes of 0.5 or 0.6) were evident even when both gender groups reported the same levels of learning-behavioral engagement. For instance, among the students who said that they read for fun in their own time almost every day, the female students scored 179, whereas the male students scored only 159, resulting in a gender effect size of 0.6. This persisting gender difference among students who were at the same levels of learning behaviors implies that gender disparities in writing are hardly explained by gender differences in the reported patterns of learning behavior. Furthermore, by comparing the scores of the L column for the males and the scores of the H column for the females in Table 4, we can see that the female students who reported having the least desirable behavioral engagement showed higher writing scores, statistically and practically, than the male students who reported the highest levels of learning behavior in five of eight behavioral statements.

In terms of overall effect sizes, attitudinal factors (0.7 in Table 3) appear to have stronger effects on writing performance than learning behavioral factors (0.5 in Table 4). This pattern was observed across both gender groups. Additionally, the female students’ scores were slightly more differentiated by their attitudes and behaviors (0.6 for attitudes and 0.5 for behaviors) than the male students’ scores (0.5 for attitudes and 0.4 for behaviors). Again, there were fairly consistent sizes of effects between gender and writing performance among students who reported the same levels of writing attitudes (0.5 and 0.6; see the last two columns of Table 3) or learning behavioral engagement (0.5 and 0.6; see the last two columns of Table 4).

Writing Achievement, Attitudes Toward Writing, and Learning Behaviors

The scores (and their Cohen’s effect sizes) rowwise in Table 5 provide estimates for the effects of attitudes on writing performance among students who reported the same levels of learning behaviors. Overall, higher scores were obtained by those with more positive attitudes toward writing across different levels of learning engagement behaviors (average Cohen’s d of 0.4 and 0.5). This means that students’ attitudes have a fairly substantial relationship with writing performance above and beyond their learning engagement levels. There were slightly greater effects among students whose behaviors were on the positive side (d = 0.5-0.6) than those who reported to never or hardly ever engage in the learning behaviors (d = 0.1-0.5). It is interesting to note that among the students who never or hardly ever read for fun, their writing scores did not differ much (144 vs. 148) regardless of whether writing was one of their favorite activities or not.

Table 5.

Writing Scores by Attitudes Toward Writing and Learning Behavior: National Assessment of Educational Progress, 2007.

	Writing Helps Share Ideas			Writing Is One of My Favorite Activities
Learning Behavior	SD	SA	d	SD	SA	d	Average Cohen’s d on Behavior
Write e-mails to friends or family
Never or hardly ever	140	155	0.4	144	157	0.4
Almost every day	146	167	0.6	151	167	0.5
Cohen’s d on behavior	0.2	0.3	—	0.2	0.3	—	0.3
Talk with friends about what you read
Never or hardly ever	140	155	0.4	145	154	0.3
Almost every day	149	167	0.5	146	168	0.5
Cohen’s d on behavior	0.3	0.3	—	0	0.4	—	0.3
Read for fun in your own time
Never or hardly ever	138	153	0.5	144	148	0.1
Almost every day	159	177	0.5	160	179	0.5
Cohen’s d on behavior	0.6	0.6	—	0.5	0.8	—	0.6
Pages read in school and for homework
Never or hardly ever	138	155	0.5	142	156	0.4
Almost every day	149	170	0.6	153	171	0.5
Cohen’s d on behavior	0.3	0.4	—	0.3	0.4	—	0.4
Average Cohen’s d on attitudes			0.5			0.4

Note: SD = strongly disagree; SA = strongly agree; d = Cohen’s d on attitude.

Examining the scores columnwise in Table 5 provides estimates for the effects of learning behaviors on writing scores among students who shared the same levels of writing attitudes. Again, more “desirable” behavior was associated with higher scores among those who endorsed the same level of attitudes toward writing. Their effects were, however, relatively small (d = 0.0-0.4), with the exception of “reading for fun,” which had medium to large effects on the writing scores. For instance, there was a score difference of 31 points (d = 0.8) between the students who reported reading for fun almost every day (179) and those who never or hardly ever read for fun (148), even though both groups strongly agreed that writing was one of their favorite activities.

Writing Achievement, Attitudes Toward Writing, Learning Behaviors, and Gender

The writing scores and their associated effect sizes presented in Tables 6 and 7 represent empirical evidence for the relationship between gender and writing achievement among students who shared the same response categories for the attitudinal and behavioral factors. The scores shown in Table 6 are for those with the most positive attitudes and most desirable learning behaviors; the scores in Table 7 are for those with the most negative attitudes and least desirable learning behaviors. As can be seen, the medium size of gender effects on the writing scores was persistently shown across the four learning behavioral types and two attitudinal statements, regardless of whether the students were at the positive (Table 6) or negative (Table 7) end of endorsement for the attitudinal and behavioral statements. For instance, among the students who shared the same strong view that writing was one of their favorite activities and reported to read for fun almost every day, there was still a 24-point score difference between the male (160) and female (184) students, with a Cohen’s d effect size of 0.7 (Table 6). Similar patterns were observed for those who reported having negative attitudes and low behavioral engagement (Table 7).

Table 6.

Writing Scores by Attitudes, Behaviors, and Gender Among the Students Endorsing the Highest (Most Positive) Categories on the Attitudinal and Behavioral Variables: National Assessment of Educational Progress, 2007.

	Writing Helps Share Ideas				Writing Is One of My Favorite Activities
Learning Behavior	Total	Male	Female	d	Total	Male	Female	d
Write e-mails to friends or family	167	150	173	0.7	167	149	171	0.6
Talk with friends about what you read	167	152	174	0.6	168	151	174	0.6
Read for fun in your own time	177	161	183	0.6	179	160	184	0.7
Pages read in school and for homework	170	157	177	0.6	171	156	177	0.6

Note: d = Cohen’s d on gender.

Table 7.

Writing Scores by Attitudes, Behaviors, and Gender Among the Students Endorsing the Lowest (Most Negative) Categories on Attitudinal and Behavioral Variables: National Assessment of Educational Progress, 2007.

	Writing Helps Share Ideas				Writing Is One of My Favorite Activities
Learning Behavior	Total	Male	Female	d	Total	Male	Female	d
Write e-mails to friends or family	140	138	151	0.4	144	141	156	0.4
Talk with friends about what you read	140	136	153	0.5	145	140	157	0.5
Read for fun in your own time	138	135	150	0.5	144	139	156	0.5
Pages read in school and for homework	138	133	151	0.5	142	136	154	0.5

Note: d = Cohen’s d on gender.

Writing Achievement, Attitudes Toward Writing, Learning Behaviors, and Demographic Background Information

To illustrate the effects of gender and attitudinal and behavioral factors on writing performance in the broader context, the variables of demographic background information and language factors were linked to writing performance (Tables 8 and 9, based on the NAEP 1998 and 2007 writing assessments, respectively). Perhaps not surprising, the students’ status of learning English as a second language (i.e., whether they were English-language learners whose first language is not English) showed the largest effect sizes on writing performance in both years of writing assessment (1.3 in 1998, 1.1 in 2007). What is surprising is that the next-largest effects (in the 1998 assessment) were found for the attitudinal factor of whether the students considered themselves to be good at writing (0.9) and the behavioral factor of amount of time spent per day on homework (0.9). These effects were larger than the effect sizes for the score differences between White and Black students (0.8), between the students whose parents did not finish high school and those whose parents graduated from college (0.8), and between the students who were at the national poverty level and eligible for the National School Lunch Program and those who were not (0.7). Among the six attitudinal and behavioral statements in the 1998 assessment, the largest effect sizes on writing performance were found for amount of time spent on homework and being “good at writing,” followed by “liking to write.”

Table 8.

Writing Scores of Subgroups by Student-Level Variables Employed in This Study: National Assessment of Educational Progress, 1998.

Variable	Subgroup	Score	Cohen’s d
Demographic information
Gender	Male	160	0.6
Gender	Female	140	0.6
Lunch eligibility	Eligible	132	0.7
Lunch eligibility	Not eligible	157	0.7
Race/ethnicity	Black	131	0.8
Race/ethnicity	White	157	0.8
Parent education	Less than high school	130	0.8
Parent education	Graduated college	159	0.8
Status on ELL	ELL	107	1.3
Status on ELL	Not ELL	151	1.3
Language other than English use at home	All or most of the time	141	0.3
Language other than English use at home	Never	152	0.3
Attitudes toward writing
Like to write	Strongly disagree	128	0.8
Like to write	Strongly agree	160	0.8
Good at writing	Strongly disagree	120	0.9
Good at writing	Strongly agree	157	0.9
Learning behaviors
Discussion of studies at home	Never or hardly ever	137	0.6
Discussion of studies at home	Almost every day	159	0.6
Number of pages read	5 or fewer	142	0.4
Number of pages read	More than 20	155	0.4
Time per day on homework	Have but don’t do	129	0.9
Time per day on homework	More than 1 hour	160	0.9
Amount of television or video watching	More than 6 hours	132	0.5
Amount of television or video watching	None	150	0.5

Note: ELL = English-language learner.

Table 9.

Writing Scores of Subgroups by Student-Level Variables Employed in This Study: National Assessment of Educational Progress, 2007.

Variable	Subgroup	Score	Cohen’s d
Demographic information
Gender	Male	146	0.6
Gender	Female	166	0.6
Lunch eligibility	Eligible	141	0.7
Lunch eligibility	Not eligible	164	0.7
Race/ethnicity	Black	141	0.7
Race/ethnicity	White	164	0.7
Parent education	Less than high school	139	0.8
Parent education	Graduated college	166	0.8
Status on ELL	ELL	120	1.1
Status on ELL	Not ELL	158	1.1
Language other than English use at home	All the time	146	0.3
Language other than English use at home	Never	158	0.3
Attitudes toward writing
Writing helps	Strongly disagree	144	0.6
Writing helps	Strongly agree	165	0.6
Writing is my favorite activities	Strongly disagree	149	0.5
Writing is my favorite activities	Strongly agree	165	0.5
Learning behaviors
Writing e-mails	Never or hardly ever	149	0.3
Writing e-mails	Almost every day	159	0.3
Talking with friends about what you read	Never or hardly ever	148	0.4
Talking with friends about what you read	Almost every day	161	0.4
Read for fun	Never or hardly ever	147	0.7
Read for fun	Almost every day	172	0.7
Number of pages read	5 or fewer	147	0.4
Number of pages read	More than 20	162	0.4

Note: ELL = English-language learner.

Similarly, in the 2007 NAEP writing assessment, the largest effects were found for English-language learner status (1.1) and parental education (0.8). The behavioral indicator of reading for fun had the same effect size (0.7) as the effects found in the scores between White and Black students and between the students who were eligible for the National School Lunch Program and those who were not. The two attitudinal statements showed about the same effect size on writing as gender (0.6). Among the six attitudinal and behavioral statements in the 2007 assessment, “reading for fun” showed the largest effect size on writing performance, followed by two attitudinal statements.

As can be seen from Tables 8 and 9, the effect sizes for the variables/statements on writing performance were fairly consistent between the two assessment years (1998 and 2007). The gender effect size was 0.6 in both assessments, and all the other demographic information variables showed very similar effect sizes between the two years. The effect sizes of attitudinal and behavioral factors on writing achievement were also in a similar range. Overall, the behavioral indicators showed slightly smaller effect sizes than attitudes, with the exception of “reading for fun” (in 2007) and “amount of homework time per day” (in 1998). In sum, the NAEP writing scores in the 1998 and 2007 assessments highlight the importance of attitudes and behavior in students’ writing performance; the effect sizes were comparable to those for demographic information and language factors, which are known to have strong relationships to students’ academic achievement.

Discussion

The main objectives of this study were to investigate the relationships among gender, attitudes, and behavior and to compare their relative effects on writing performance. As expected, the female students demonstrated more positive attitudes toward writing and reported engaging in learning behaviors more frequently than the male students. The only attitudinal aspect showing relatively less gender difference was self-concept in writing. The female students exhibited stronger writing self-concept in the present study, as was the case in Pajares and Valiante’s study (2001). However, boys showed overconfidence in comparison to their actual abilities. Overall, gender differences appear to be more prominent in attitudes than learning behaviors.

Analysis of the relationships between attitudes and writing performance showed medium to large effect sizes (Table 3). It is noteworthy that gender differences in the writing scores were persistently shown (with medium effect size) among students who shared the same level of attitudes (Table 3). Furthermore, the gender effects were slightly higher for those with positive attitudes than those with negative attitudes, which means that the writing score differences between females and males were larger when they both held positive attitudes than when they reported having negative attitudes toward writing. The relationship between attitudes and writing scores was stronger for females than males, indicating that the female students’ scores were more reflective of their attitudes than the male students’ scores. Perhaps the most surprising finding was that the female students with the most negative attitudes outperformed the male students with the most positive attitudes in the 2007 assessment (Table 3).

Fairly similar patterns were observed in the relationship between learning behavior and writing scores. Overall, there was a medium size of gender effects, with an average Cohen’s d of about 0.4 to 0.5 among the students who shared the same level of learning engagement (Table 4). Females outperformed the males with writing score differences of 15 to 25 points when they both reported to be at the same level of learning behavior. Furthermore, the female students who reported to be at the lowest categories of learning behaviors performed better than the male students who reported to be at the highest categories of the same learning behaviors in five of eight statements (Table 4). Apparently, positive attitudes and “good” learning behavior among the male students did not transfer to their writing skills as effectively as with the female students.

The current study indicates that a good proportion of gender differences in writing performance is not explained by gender differences in either attitudes or behavior or in attitudes and behavior taken together. When attitudes and learning behaviors were considered together in relation to writing performance, the gender effects were still present with medium effect sizes. This shows that although the students were at the same level in terms of writing attitudes and learning behaviors, the female students scored substantially higher than the male students (Tables 6 and 7). For instance, the female students with positive attitudes and behaviors had writing scores around or at the proficient level (scores higher than 173), whereas the male students with the same high levels of attitudes and behavioral engagement showed writing scores corresponding to the lowest achievement level (basic) in the 2007 assessment framework. Pajares and Valiante (1999) had noted that there was no interaction effect between gender and self-efficacy in predicting writing performance. This suggests that the relationship between gender and writing does not differ by students’ level of self-efficacy. This previous finding, along with the findings in this study, suggests that researchers should look beyond attitudinal and behavioral factors to understand how gender plays a strong role in writing performance. One such attempt was made by Pajares and Valiante (2001), who demonstrated that the gender effects on writing remain insignificant when students’ masculinity and femininity are taken into account.

As can be expected, some attitudinal and behavioral statements distinguished writing scores more effectively than others. In general, the effects of attitudes (Table 3) were slightly larger than the effects of behaviors (Table 4) on writing performance. However, among six attitudinal and behavioral statements measured in the 1998 assessment, the largest effect sizes on writing performance were found for the amount of time spent on homework and whether students thought they were good at writing. Again, among six attitudinal and behavioral statements measured in the 2007 assessment, the behavioral indicator of “reading for fun” showed the largest effect on writing performance. Self-belief (i.e., writing self-concept) was the most effective in distinguishing the writing scores from among four attitudinal statements (Table 3). This is in line with previous studies showing that self-belief is an important and perhaps the strongest predictor of writing performance (see McCarthy et al., 1985; Pajares & Valiante, 1997, 1999; Pajares et al., 1999). Overall, the effects of attitudes and behaviors on writing performance were quite comparable to the effects observed between students’ demographic information (e.g., race/ethnicity, parental education) and writing performance (see Tables 8 and 9), which highlights the importance of attitudes and behaviors on writing achievement.

Contrary to the concerns expressed by some writing researchers (e.g., Bruning & Horn, 2000), the findings of the present study did not indicate a serious or widespread problem in students’ attitudes toward writing. The majority of students (61%) were aware of the usefulness of writing, and about half of the students viewed themselves as being good at writing (51%). Students’ generally positive attitudes toward their own abilities have been noted in studies conducted in other countries as well. For example, according to Clark and Dugdale (2009), students in the United Kingdom were also split equally between positive and negative writing self-concepts. Since the 1980s, researchers (e.g., McCarthy et al., 1985) have noted students’ inaccurate evaluations of their own cognitive abilities. This concern continues to be expressed in recent research (e.g., Lee, 2009; Stankov & Lee, 2008). To counter the prevalence of overconfidence, McCarthy et al. (1985) argued that self-evaluation as writers should include an assessment of written work. Stankov and Lee (2008) advocated the use of a bias score (i.e., one’s confidence score deducted from the actual cognitive scores) as a more accurate index of one’s self-evaluation of cognitive and noncognitive abilities. The extant literature, however, has not addressed the extent to which this type of overconfidence and inaccurate information play a role in one’s cognitive abilities or learning. Perhaps the issue is, as Bruning and Horn (2000) have indicated, that writers need to have strong self-beliefs and perseverance to overcome difficulties and frustrations that typically accompany the writing process. If that is the case, then there is a troubling sign: Only 10% to 15% of students in this study showed such strong positive attitudes toward writing.

It is noteworthy that there was a relatively small effect for the relationship between the frequency of e-mail writing and students’ writing scores, although these two have a common component of “writing.” The existence of this relatively weak connection between the frequency of writing e-mail and writing scores implies a disconnect between the qualities of formal writing (typically used in schools and assessments) and less formal types of writing (such as texting). Even young people—and not just contemporary researchers (see Lee & Stankov, 2012)—do not perceive informal writing as “writing” (Pew Internet & American Life Project, 2008). We know, however, that young people enjoy communicating more via texting, blogging, or social networking sites than through traditional forms of writing, such as essays or letters (Clark & Dugdale, 2009). And yet, most schools have not addressed the disconnect between their writing curricula and the type of writing that students enjoy and frequently use. As long as this problem continues, fewer students will enjoy writing for and in school (Clark & Dugdale, 2009). Clearly, there is a great need for teachers and schools to know how to make the connection between formal and informal writing. Future writing research will have to suggest strategies for doing so since virtually no research has been conducted into how and what types of skills typically used in informal writing can be transferred to more formal written language and how teachers can help students to turn their enjoyment of informal writing into the practices of more formal writing.

The present study also shows remarkable consistency in the results garnered from the NAEP 1998 and 2007 data. In fact, it has been well documented that there has been a relative stability (and lack of substantial increase) in students’ writing achievement measured by the long-term NAEP projects (cf. Applebee & Langer, 2009; Salahu-Din et al., 2008). One such consistent finding includes a gender gap in writing performance, favoring female students. Thus, the better writing performance of girls is not new information. However, the present study noted that the writing gender gap has stagnated at about 20 points (on a scale of 300). This is larger than the gender gap typically found in the NAEP reading scores (i.e., about 10 points on a scale of 500; see Lee, Grigg, & Patricia, 2007). Researchers have made suggestions that may bring about improvement in boys’ engagement and skills in writing. Although most young students view writing primarily as a feminine activity (Parajes & Valiante, 2001), boys do not see it as inherently belonging only to females and thus something they cannot overcome (Hansen, 2001). It has also been shown that boys have a stronger desire to succeed in writing and show their writing competence to others than girls (Parajes & Valiante, 2001). In this light, active interventions for boys to get them more involved in writing may be worthwhile. Some suggestions include embracing reading and writing genres that boys might prefer in English classrooms and exposing boys and girls to a wide range of reading and writing literacies so that they have opportunities to write in the forms they prefer (Hansen, 2001). Once students are exposed to a variety of writing styles and have the opportunity to write in different styles and genres, their writing may improve. English teachers should also be cognizant of the types of writing assessments that they use, because school writing tends to favor the writing style that girls tend to prefer (i.e., expressive).

The extant literature on school-based writing interventions indicates that there have been hardly any programs that are specifically targeted to address gender issues in writing. As shown in recent meta-analyses on the effectiveness of writing instruction and intervention (Graham & Perin, 2007; Rogers & Graham, 2008), the main treatment has been on planning/drafting, prewriting activities, grammar, text structure, sentence combining, summarization, goal setting, strategies for idea development, explicit teaching of self-regulation in writing, and editing skills. Teachers of writing should also consider strategies to motivate students to engage in extended writing rather than asking for short answers or to fill in blanks (Applebee & Langer, 2011). Teachers can encourage students to self-direct their writing process, which involves choosing a topic and genre, considering an audience for their work, deciding on the amount of technology to be used, and managing and monitoring writing progress (cf, Applebee & Langer, 2011; Graham & Perin, 2007). Teachers can also help students to realize that writing is, in fact, a labor-intensive activity that requires long hours of concentration and hard work. It also involves the effective management of affective and cognitive strategies (Graham et al., 2007). Writers have to be “patient, persistent, and flexible” (Bruning & Horn, 2000, p. 26). Some researchers argue for using authentic writing tasks in school (i.e., writing with real purposes and for personal enjoyment) as a way to maximize students’ persistence and patience when writing (Clark & Dugdale, 2009;). Authentic writing may also enhance positive feelings toward writing, which would help students engage in more writing activities (Tunks, 2010) and build their self-beliefs in writing (Bruning & Horn, 2000). A number of other instruction-based strategies for Grades 3 to 12 writing are presented and summarized in detail in Graham and Perin (2007, pp. 466-467) and Rogers and Graham (2008, pp. 899-900).

Conclusion

Strong associations between students’ writing attitudes and learning behaviors and their writing performance have important policy implications, as these characteristics (i.e., attitudes and behavior) are relatively malleable (as opposed to more static variables; e.g., parental education levels or race/ethnicity). Many strategies have been suggested by researchers, which can be implemented for classroom use. Meanwhile, issues relating to gender in writing remain unresolved and may be more resilient than what some of the previous literature may have suggested. The present study shows that the gender disparities in writing persist and remain strong even when boys and girls share the same levels of attitudes and behavior. Male students with positive attitudes and behaviors performed even worse than the female students with negative attitudes and behaviors in half of the statements examined. Future studies should look into gender disparities in writing that go beyond issues relating to attitudes and behavior and look into ways to make the connection between formal and informal writing.

Footnotes

Appendix

Response Categories for Attitudinal/Behavioral Statements and Demographic Information

Variable (Survey Year)	Responses
Gender (1998, 2007)	Male, female
Lunch eligibility (1998, 2007)	Eligible, others
Race/ethnicity (1998, 2007)	White, Black, Hispanic, Asian/Pacific Islander, American Indian, unclassified
Parent education (1998, 2007)	Did not finish high school, graduated high school, some education after high school, graduated college, unknown
Status on English-language learner (ELL; 1998, 2007)	ELL, not ELL
Language other than English at home (1998, 2007)	Never, less than half time, about half the time, all or most of the time (1998); never, once in a while, half the time, all or most of the time (2007)
Good at writing (1998), Like to write (1998), Writing helps (2007), Writing is one of my favorite activities (2007)	Strongly disagree, disagree, agree, strongly agree
Time per day on homework (1998)	Don’t usually have it, have but don’t do, 1/2 hour or less, 1 hour, more than 1 hour
Discuss studies at home (1998), Read for fun (2007), Talking with friends about what you read (2007), Writing emails (2007)	Never or hardly ever, once or twice a month, once or twice a week, almost every day
Amount of television or video watching (1998)	Less than 2 hours, 3 hours, 4 hours, 5 hours, more than 6 hours
Number of pages read (1998, 2007)	5 or fewer, 6-10, 11-15, 16-20, more than 20

Acknowledgements

I would like to thank Christin Rekha Jonathan at National Institute of Education, Singapore, for her assistance in the process of preparing the literature review section of this article.

Declaration of Conflicting Interests

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

Funding

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

Notes

Author Biography

Jihyun Lee is a Senior Lecturer in the School of Education of the University of New South Wales, Sydney, Australia. She researches on the relationships between attitudes and behaviors in students’ academic learning. Main constructs of her studies involve interest, self-beliefs, engagement, confidence, and enjoyment in learning.

References

Applebee

A. N.

Langer

J. A.

(2011). A snapshot of writing instruction in middle schools and high schools. English Journal, 98, 18-28.

Bruning

Horn

(2000). Developing motivation to write. Educational Psychologist, 35(1), 25-37.

Clark

Dugdale

(2009). Young people’s writing: Attitudes, behavior and the role of technology. London: National Literacy Trust.

Clark

Foster

(2005). Children’s and young people’s reading habits and preferences: The who, what, why, where and when. London: National Literacy Trust.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.

Cohen

(1992). A power primer. Psychological Bulletin, 112, 155-159.

Comstock

(1995). Television and the American child. In Hedley

Antonacci

Rabinowitz

(Eds.), Thinking and literacy: The mind at work (pp. 101-123). Hillsdale, NJ: Erlbaum.

Cooper

Lindsay

J. J.

Nye

Greathouse

(1998). Relationships among attitudes about homework, amount of homework assigned and completed, and student achievement. Journal of Educational Psychology, 90, 70-83.

Fantuzzo

McWayne

Perry

M. A.

Childs

(2004). Multiple dimensions of family involvement and their relations to behavioral and learning competencies for urban, low-income children. School Psychology Review, 33, 467-480.

10.

Goldberg

Russell

Cook

(2003). The effect of computers on student writing: A meta-analysis of studies from 1992 to 2002. Journal of Technology, Learning, and Assessment, 2(1), 1-52.

11.

Graham

Berninger

Fan

(2007). The structural relationship between writing attitude and writing achievement in first and third grade students. Contemporary Educational Psychology, 32, 516-536.

12.

Graham

Perin

(2007). A meta-analysis of writing instruction for adolescent students. Journal of Educational Psychology, 99, 445-476.

13.

Greenwald

E. A.

Persky

H. R.

Campbell

J. R.

Mazzeo

(1999). The NAEP 1998 writing report card for the nation and the states (NCES Report No. 1999-462). Washington, DC: National Center for Education Statistics.

14.

Hahs-Vaughn

D. L.

(2005). A primer for using and understanding weights with national datasets. Journal of Experimental Education, 73, 221-248.

15.

Hansen

(2001, July). Boys and writing: Reluctance? Reticence? Or rebellion? Paper presented at the annual meeting of the Education Research Network, Spetses, Greece.

16.

Hartung

Knapp

Sinha

B. K.

(2008). Statistical meta-analysis with application. Hoboken, NJ: Wiley.

17.

S.-C. E.

Willms

J. D.

(1996). Effects of parental involvement on eighth-grade achievement. Sociology of Education, 69, 126-141.

18.

Krashen

S. D.

(1995). Free voluntary reading: Linguistic and affective arguments and some new applications. In Eckman

Highland

Lee

Mileham

Weber

(Eds.), Second language acquisition: Theory and pedagogy (pp. 187-202). Mahwah, NJ: Erlbaum.

19.

Lee

(2009). Universals and specifics of math self-concept, math self-efficacy, and math anxiety across 41 PISA 2003 participating countries. Learning and Individual Differences, 19, 355-365.

20.

Lee

Grigg

Dion

(2007). The nation’s report card: Mathematics 2007 (NCES Report No. 2007-494). Washington, DC: National Center for Education Statistics.

21.

Lee

Grigg

Patricia

(2007). The nation’s report card: Reading 2007 (NCES Report No. 2007-496). Washington, DC: National Center for Education Statistics.

22.

Lee

Shute

V. J.

(2010). Personal and social-contextual factors in K–12 academic performance: An integrative perspective on student learning. Educational Psychologist, 45, 185-202.

23.

Lee

Stankov

(2012). Large-scale online writing assessments: New approaches adopted in the National Assessment of Educational Progress (NAEP). In Grigorenko

E. L.

Mambrino

Preiss

(Eds.), Writing: A mosaic of new perspectives (pp. 371-384). New York: Psychology Press.

24.

Lipstein

R. L.

Renninger

K. A.

(2007). “Putting things into words”: The development of 12-15-year-old students’ interest for writing. In Boscolo

Hidi

(Eds.), Writing and motivation (pp. 113-140). Oxford, England: Elsevier.

25.

McCarthy

Meier

Rinderer

(1985). Self-efficacy and writing: A different view of self-evaluation. College Composition and Communication, 36, 465-471.

26.

Organization for Economic Cooperation and Development. (2007). PISA 2006: Science competencies for tomorrow’s world. Paris: Author.

27.

Pajares

Miller

M. D.

Johnson

M. J.

(1999). Gender differences in writing self-beliefs of elementary school students. Journal of Educational Psychology, 91(1), 50-61.

28.

Pajares

Valiante

(1997). The predictive and mediational role of the writing self-efficacy beliefs of upper elementary students. Journal of Educational Research, 90, 353-360.

29.

Pajares

Valiante

(1999). Grade level and gender differences in the writing self-beliefs of middle school students. Contemporary Educational Psychology, 24, 390-405.

30.

Pajares

Valiante

(2001). Gender differences in writing motivation and achievement of middle school students: A function of gender orientation? Contemporary Educational Psychology, 26, 366-381.

31.

Pew Internet & American Life Project. (2008). Writing, technology and teens. Washington, DC: Author.

32.

Pintrich

P. R.

De Groot

E. V.

(1990). Motivational and self-regulated learning components of classroom academic performance. Journal of Educational Psychology, 82(1), 33-40.

33.

Rogers

L. A.

Graham

(2008). A meta-analysis of single subject design writing intervention research. Journal of Educational Psychology, 100, 879-906.

34.

Salahu-Din

Persky

Miller

(2008). The nation’s report card: Writing 2007 (NCES Report No. 2008-468). Washington, DC: National Center for Education Statistics.

35.

Shang

H. F.

(2007). An exploratory study of e-mail application on FL writing performance. Computer Assisted Language Learning, 20(1), 79-96.

36.

Stankov

Lee

(2008). Confidence and cognitive test performance. Journal of Educational Psychology, 100(4), 961-976.

37.

Tunks

K. W.

(2010). Attitudes toward writing in fifth grade students preparing for state mandated writing exam. International Journal of Applied Educational Studies, 8, 1-9.

38.

Vereecken

C. A.

Todd

Roberts

Mulvihill

Maes

(2005). Television viewing behavior and associations with food habits in different countries. Public Health Nutrition, 9(2), 244-250.

39.

Wilkinson

, & APA Task Force on Statistical Inference. (1999). Statistical methods in psychology journals: Guidelines and explanations. American Psychologist, 54, 594-604.

40.

Zimmerman

B. J.

(1990). Self-regulated learning and academic achievement: An overview. Educational Psychologist, 21, 3-17.