Does accuracy and confidence in working memory performance relate to academic achievement in NAPLAN, the Australian national curriculum assessment?

Abstract

Objective

The aim of this study was to examine how accuracy and confidence in working memory performance relates to academic achievement as assessed in the Australian national curriculum assessment (National Assessment Program—Literacy and Numeracy (NAPLAN)).

Method

A total of 150 Australian schoolchildren enrolled in Year 4 participated in the study. Accuracy and confidence in performance of a working memory task were assessed. Associations between these working memory measurements and scores in each of the NAPLAN domains (numeracy, reading, persuasive writing, grammar, and spelling) were examined, separately for males and females.

Results

Accuracy in working memory performance was associated with achievement in all five NAPLAN domains, in both males and females. Confidence in working memory performance was also related to achievement, but the pattern of results differed for males and females. For females, significant associations were found between confidence and achievement in numeracy, reading, writing, and spelling. For males, confidence was associated with achievement in numeracy only. Females outperformed males in persuasive writing. There was a non‐significant trend for males to outperform females in numeracy.

Conclusion

The strong links between working memory and achievement highlight the importance of early detection of working memory problems. Further research is needed to examine the extent to which the pattern of results generalises to other year levels.

Key words:

• academic achievement
• confidence
• literacy
• NAPLAN
• numeracy
• working memory

What is already known about the topic?

• Working memory ability is related to academic achievement.

• Confidence is related to academic achievement in adults.

• Studies of gender differences in confidence have produced mixed results.

What this topic adds?

• Working memory ability is related to achievement in all five domains of the Australian National Assessment Program—Literacy and Numeracy (NAPLAN).

• Confidence in working memory performance is related to both males' and females' achievement in NAPLAN numeracy.

• Confidence is related to achievement in the NAPLAN literacy domains, for females but not for males.

Working memory and academic achievement

Working memory has been described as a complex cognitive mechanism used to temporarily store and process information (Miyake & Shah, 1999). It is important in the performance of tasks such as learning, comprehension, and reasoning (Baddeley, 2003). Children with working memory problems tend to struggle to keep up in class and generally display poor academic progress (Alloway, Gathercole, Kirkwood, & Elliot, 2009).

An association between working memory and achievement in numeracy and literacy has been demonstrated in a number of studies (e.g., Alloway & Alloway, 2010; Alloway & Copello, 2013; Alloway & Passolunghi, 2011; Bull, Espy, & Wiebe, 2008; Dumontheil & Klingberg, 2012). Much of the research examining the relationship between working memory and academic achievement has been conducted in England, using key stage (KS) attainment levels to measure achievement in English, mathematics, and science (GOV.UK, 2015). These studies have primarily used working memory measures based on Baddeley's tripartite model of working memory, which consists of a domain‐free central executive and two domain specific slave systems, the phonological loop and the visuospatial sketchpad (Baddeley, 2003; Baddeley & Hitch, 1974). The phonological loop is primarily responsible for briefly storing and rehearsing verbal information and the visuospatial sketchpad is responsible for processing visual and spatial information. Using this framework, Gathercole and Pickering (2000) found poor KS1 performance in the domains of English, mathematics and science was related to impaired performance on working memory measures that are assumed to tap the central executive and visuospatial sketchpad, but found no relationship between verbal working memory and academic achievement. St Clair‐Thompson and Gathercole (2006) also found KS1 performance in mathematics was associated with visuospatial working memory, but in contrast to Gathercole and Pickering, they found performance in English was associated with verbal working memory. Gathercole, Pickering, Knight, and Stegmann (2004) also reported that achievement in each of the three main KS domains was associated with working memory capacity in Year 3 children. By Year 10, the relationship held for mathematics and science, but not for English.

Although a number of studies have examined the link between working memory and academic achievement, relatively few studies have been conducted using national curriculum assessments outside of England (e.g., De Smedt, Verschaffel, & Ghesquiere, 2009; Lundberg & Sterner, 2006). Only one study of which we are aware has investigated the relationship between working memory and academic achievement in a national curriculum assessment in an Australian sample. Cameron, Glyde, and Dillon (2014) examined the association between working memory capacity and performance in the Australian national curriculum test, the National Assessment Program—Literacy and Numeracy (NAPLAN, The Australian Curriculum, Assessment and Reporting Authority (ACARA), 2013). The NAPLAN assesses academic performance in mathematics, reading, writing and language.

Cameron et al. (2014) assessed the relationship between working memory and NAPLAN literacy (collapsed over reading, persuasive writing, grammar, and spelling) and numeracy in children in Years 3 and 5. They found weak correlations between working memory assessed by their sentence processing task, and both numeracy and literacy. Moderate correlations between working memory assessed by a reverse digit span task, and numeracy and literacy were also reported. No difference between males and females was found in either of their working memory measures. They also found no difference between males and females in their NAPLAN literacy composite. They did, however, find males performed better than females in numeracy.

Confidence and academic achievement

Metacognition refers to the processes involved in thinking about one's own thinking (Kleitman & Gibson, 2011). Metacognitive awareness is thought to begin to develop around 4–6 years of age, followed by the development of metacognitive knowledge (Veenman & Spaans, 2005). Metacognitive skills (i.e., the ability to monitor one's performance and adapt learning strategies) develop around 11–12-years of age (Veenman, Wilhelm, & Beishuizen, 2004).

Metacognitive skills are often examined by tapping an individual's confidence in performing a task. Confidence may be assessed prior to the individual undertaking a task (e.g., How well do you think you will do in the science exam?), which is often referred to as self‐efficacy. Confidence may also be assessed by a confidence rating taken following the completion of a test or a test trial. For example, a participant may be asked to rate their confidence that their answer to a test question was correct on a scale (e.g., ranging from I'm just guessing to I'm sure I got it right). In the present study we use the term confidence in the latter sense, to refer to one's confidence in a test answer.

Research on the role of confidence in academic achievement has mainly focused on adolescent and adult populations, but investigation into the relationship in younger children has increased in recent years. For example, Kleitman and Gibson (2011) found confidence is related to academic achievement in Year 6 students (mean age of 12-years) and Kasperski and Katzir (2013) found confidence is related to reading comprehension in Year 4 students. Kleitman and Moscrop (2010) concluded that confidence is a stable construct in children as young as 9 years old, both within and across cognitive test domains (numeracy, vocabulary, and Raven's Progressive Matrix). More recently, Freeman, Karayanidis, and Chalmers (2017) reported that Year 4 children's confidence in their performance on a working memory test was related to academic achievement in numeracy assessed using the Wechsler Individual Achievement Test (WIAT‐II, Wechsler, 2007).

Studies examining gender differences in confidence have reported conflicting results. For example, while Pallier (2003) reported higher confidence in males than in females in performance of perceptual and cognitive tasks (see also Cooke‐Simpson & Voyer, 2006, who used a mental rotation task), other researchers have found no gender differences. As Lundeberg, Fox, and LeCount (1992) point out, gender differences in confidence may be domain specific. In a study of undergraduate psychology students, Lundeberg et al. found males were more confident than females in the numeracy domain, but not in tasks assessing learning and memory.

Studies in children also suggest gender differences in confidence vary with the context and the task. For example, Kleitman and Moscrop (2010) found no difference between males and females in confidence in numeracy and vocabulary, whereas Kleitman and Gibson (2011) reported that males tended to be more confident than females in a composite measure of academic achievement, although the difference was only marginally significant. Sex differences in each of the assessed domains (reading, vocabulary, and mathematics) were not reported.

Present research

The aim of the present research was to examine the relationship between working memory accuracy and confidence and academic achievement in Australian schoolchildren. NAPLAN scores were used to index academic achievement in five domains (numeracy, reading, persuasive writing, grammar, and spelling). Working memory accuracy and confidence were assessed using the Working Memory Power Test (WMPT) for children. The WMPT is the only working memory test for children of which we are aware that assesses both confidence and accuracy in working memory performance. The test items consist of non‐verbal stimuli (line drawings of animals that are familiar to young children) and the task requirements are written in simple English, making it suitable for children from different language backgrounds. The WMPT is easy to administer and can be completed online with minimal supervision.

Based on previous findings reported by Cameron et al. (2014), we predicted there would be a positive relationship between WMPT accuracy and NAPLAN numeracy scores. While we expected there would be a positive relationship between WMPT accuracy scores and NAPLAN literacy scores, no specific predictions were made for the individual NAPLAN literacy domains as these scores were combined to form a composite variable in Cameron et al.’s study; results for each of the individual NAPLAN domains were not reported. In line with previous work examining metacognitive measures of working memory as a predictor of academic achievement, we expected children’s ratings of confidence in their working memory performance would be positively correlated with attainment in numeracy, but not reading and spelling (Freeman et al., 2017).

We also provide a preliminary investigation of whether the relationship between working memory confidence and NAPLAN scores, and working memory accuracy and NAPLAN scores differs for males and females. No specific predictions were made as we are unaware of any previous studies examining the relationship between working memory performance (both confidence and accuracy) and NAPLAN scores in males and females, separately.

Differences between males and females in NAPLAN scores and working memory performance (both accuracy and confidence) were also examined. Based on Cameron et al.’s (2014) findings, we expected that males would outperform females in NAPLAN numeracy, but perhaps not in literacy. As noted above, Cameron et al. found no difference between males and females in their composite literacy score, but results for the individual NAPLAN domains were not reported. Based on Cameron et al.’s finding of no significant difference between males and females in accuracy of performance in their working memory tasks, we expected similar results for working memory accuracy in the present study. No specific prediction regarding sex differences in confidence in working memory performance were made as we are unaware of any previous research examining sex differences in children’s confidence in working memory performance and research in other domains has produced inconsistent results.

METHOD

Participants

Participants were 150 children (75 males, 75 females) enrolled in Year 4 of primary school in NSW, Australia. Mean age at time of testing was 9 years 11-months for males and 9 years 10 months for females (standard deviation (SD) = 5 months for both males and females). All children spoke fluent English and were predominately from low and middle socioeconomic status (SES) backgrounds. Written consent was obtained from both the school and the child’s parent. Permission to access results from the previous year’s NAPLAN test was obtained from the parent, with these provided by the school. Verbal assent was obtained from the child prior to the commencement of testing. Ethics approval to conduct the research was obtained from the Human Research Ethics Committee of the University of Newcastle.

Measures

The WMPT (Lewis Cadman Consulting Pty Ltd, Sydney, Australia)

The WMPT is an online assessment used to measure children’s working memory performance. It is designed so that children can complete it independently or with minimal supervision. It assesses accuracy and confidence in performance. The WMPT has five levels of increasing difficulty. In the Memorise condition, the child is presented with three pictures (e.g., pig, cat, duck) and asked to remember the order of the pictures as they are displayed on the screen. They are then presented with an answer screen, which shows the same three pictures in each possible order and asked to select the option representing the correct sequence of pictures. Subsequent levels of the WMPT involve mentally swapping the order of the pictures, for example, ‘Swap 2 and 3’ and then selecting the option that represents the correct sequence of pictures after the swap has been made (i.e., pig, duck, cat in the example above). The levels progress from 1 to 4 swaps, with item complexity increasing as more consecutive swaps are required. There are 5 trials at each level for a total of 25 items. After each test item, the child is asked to rate how confident they are that they answered the preceding question correctly using a 4‐point scale (1 = I guessed the answer through to 4 = I definitely got it right). Internal consistency (Cronbach’s alpha = .85) and concurrent validity of WMPT test scores have been reported (Chalmers & Freeman, 2017). The test takes approximately 20-min to complete.

National Assessment Program—Literacy and Numeracy

The NAPLAN is a standardised Australian national curriculum assessment administered annually to students in Grade 3, 5, 7, and 9 (ACARA, 2013). It assesses academic achievement in numeracy and literacy. The numeracy component assesses skills in number, space, algebra, function and pattern, and measurement. The literacy component consists of three subtests: reading, which assesses comprehension; persuasive writing, which examines spelling and punctuation, sentence and text structure, and vocabulary; and language conventions, which assesses spelling, punctuation, and grammar. The NAPLAN standards are reported in bands. For Year 3, Band 1 indicates students are below the national standard, Band 2 indicates students are at the national standard, Bands 3 and above indicate students are above the national standard. Achievement in each of five domains is reported (numeracy, reading, persuasive writing, grammar, and spelling).

Procedure

Children were tested individually during their normal school day. Testing took place in a quiet space at their school. Children completed the WMPT on a laptop computer using a mouse to record their responses. The Memorise, 1‐Swap and 2‐Swap conditions started with two practice problems. Responses were automatically scored. The child’s Year 3 NAPLAN results were provided by the school (with the parent’s permission).

Data analysis

WMPT accuracy was calculated as the number of correct trials (out of 25) and converted to a percent correct. WMPT confidence ratings were summed over all 25 trials. Independent samples t‐tests were used to examine differences between males and females in working memory accuracy and confidence scores. Differences between males and females in NAPLAN scores were examined using the Mann–Whitney U‐test. Spearman correlations were used to examine relationships between working memory and NAPLAN scores. Fischer’s z‐test was used to test the significance of the difference in the strength of the associations between males and females. Cohen’s (1988) guidelines were used to interpret the strength of significant correlations. The statistical assumptions for these tests were met. An alpha level of .05 was used for all statistical analyses.

RESULTS

Descriptive statistics for males and females in WMPT accuracy and confidence scores are presented in Table 1. There was no significant difference between males and females in either accuracy, t(148) = 0.07, d = .12, p = .482, or confidence, t(148) = 1.63, d = .27, p = .105, in working memory performance. Significant correlations between working memory confidence and accuracy scores were found for males, r_s(75) = .278, p = .016, females, r _s(75) = .383, p = .001, and overall (i.e., collapsed across gender), r_s(150) = .331, p < .001. Higher confidence was associated with higher accuracy, with the strength of the associations ranging from small to moderate.

Table 1. Descriptive statistics for males and females in performance in the Working Memory Power Test for Children (WMPT) and in the National Assessment Program—Literacy and Numeracy (NAPLAN)

Measure	Males			Females
	Mean	SD	Range	Mean	SD	Range
WMPT
Accuracy	62.72	18.19	20–96	64.84	18.67	20–100
Confidence	74.98	12.56	41–100	71.59	12.87	25–99
NAPLAN
Numeracy	4.79	1.30	2–7	4.43	1.27	1–7
Reading	5.01	1.54	1–7	5.25	1.42	2–7
Persuasive writing	4.51	.91	2–6	4.83	.80	3–7
Grammar	5.01	1.38	2–7	5.21	1.27	2–7
Spelling	4.64	1.43	1–7	4.68	1.21	2–7

Descriptive statistics for males and females in each of the NAPLAN domains (numeracy, reading, persuasive writing, grammar, and spelling) are also presented in Table 1. Females had significantly higher scores than males in persuasive writing, U = 2,316, d = .31, p = .042. Males had higher scores than females in numeracy, but in contrast to previous findings (Cameron et al., 2014) the difference was not significant, U = 2,350, d = .29, p = .072. There were no significant differences between males and females in performance in any other NAPLAN domain (reading, U = 2,590, d = .14, p = .393; grammar, U = 2,610, d = .13, p = .434; and spelling, U = 2,745, d = .04, p = .793).

Relationship between working memory accuracy and NAPLAN scores

Spearman correlations examining the relationship between working memory accuracy and each of the NAPLAN domains are presented in Table 2. For males and females combined, significant positive correlations between working memory accuracy and academic achievement were found in all five NAPLAN domains. Higher working memory accuracy scores were associated with higher NAPLAN scores. The largest correlation between working memory accuracy and NAPLAN scores was in numeracy, followed by reading, persuasive writing, grammar, and spelling. Each of these correlations was of medium strength according to Cohen’s (1988) guidelines.

Table 2. Spearman correlations (for males, females, and combined) examining accuracy in performance in the Working Memory Power Test for children and in the National Assessment Program—Literacy and Numeracy (NAPLAN)

NAPLAN domain	Combined	Males	Females
Numeracy	.498^***	.493^***	.528^***
Reading	.417^***	.369^**	.463^***
Persuasive writing	.319^***	.253^*	.413^***
Grammar	.317^***	.323^**	.330^**
Spelling	.309^***	.229^*	.402^***

Note.

* p < .05,

** p < .01,

*** p < .001.

Spearman correlations between working memory and NAPLAN scores were also computed for males and females separately. As shown in Table 2, for females, there was a strong positive correlation between working memory accuracy and NAPLAN numeracy scores, and medium correlations between working memory accuracy and all four literacy NAPLAN domains. For males, there were moderate correlations between working memory accuracy and NAPLAN numeracy, reading, and grammar, and weak correlations between working memory accuracy and persuasive writing and spelling. While the pattern of significant correlations differed for males and females, the results of Fischer’s z‐test showed the differences in the strength of the associations between males and females were not significant.

Relationship between working memory confidence and NAPLAN scores

Spearman correlations between working memory confidence and NAPLAN scores are presented in Table 3. Collapsed across males and females, the only significant correlation was between working memory confidence and NAPLAN numeracy scores. Higher confidence in working memory performance was moderately correlated with higher numeracy scores.

Table 3. Spearman correlations (for males, females, and combined) examining confidence in performance in the Working Memory Power Test for children and in the National Assessment Program—Literacy and Numeracy (NAPLAN)

NAPLAN domain	Combined	Males	Females
Numeracy	.303^***	.309^**	.271^*
Reading	.157	.079	.253^*
Persuasive writing	.152	.093	.270^*
Grammar	.140	.125	.164
Spelling	.104	−.024	.239^*

Note.

* p < .05,

** p < .01,

*** p < .001.

When considered separately for males and females, the same pattern of results was found for males. That is, the only significant correlation was a moderate correlation between working memory confidence and NAPLAN numeracy scores. For females, small but significant correlations were found between working memory confidence and scores in NAPLAN numeracy, reading, persuasive writing and spelling. Higher confidence was associated with higher NAPLAN scores in each of these four domains. The results of Fischer’s z‐test showed that while the pattern of significant correlations differed for males and females, the differences in the strength of the associations between males and females were not significant.

DISCUSSION

The aim of this study was to investigate the relationship between working memory and academic achievement as assessed by the Australian national curriculum assessment, NAPLAN. Both accuracy and confidence in working memory performance were examined. Gender differences in the working memory and academic achievement measures and whether the relationship between these constructs differs for males and females were also assessed. We first discuss the results for the analyses examining the relationship between working memory confidence and accuracy scores and academic achievement. The findings relating to gender differences in working memory and gender differences in academic achievement scores are then discussed. Limitations and future directions are also presented.

Relationship between working memory and academic achievement

The present results were consistent with our predictions; working memory accuracy, as measured by the WMPT, was associated with academic achievement, as measured by the NAPLAN. Higher working memory accuracy scores were associated with higher NAPLAN scores in each of the five NAPLAN domains (numeracy, reading, writing, grammar, and spelling), all with moderately strong correlations. These results were consistent with previous research showing that working memory is related to academic achievement in numeracy (Dumontheil & Klingberg, 2012; Lee, Ning, & Goh, 2014; Raghubar, Barnes, & Hecht, 2010) and literacy skills (Gathercole et al., 2004; Jarvis & Gathercole, 2003), including reading comprehension (Cain, Oakhill, & Bryant, 2004; Kasperski & Katzir, 2013), and writing and spelling skills (Bourke, Davies, Sumner, & Green, 2014; Niedo, Abbott, & Berninger, 2014; Ormrod & Cochran, 1988).

The results are also consistent with Cameron et al.’s (2014) finding of a relationship between working memory and achievement in numeracy and literacy assessed in the NAPLAN. Additionally, we have shown that the relationship between working memory is present in all four of the NAPLAN literacy domains (reading, persuasive writing, grammar, and spelling) as opposed to a composite literacy score.

When the relationship between working memory accuracy and achievement was examined separately for males and females, some qualitative differences in the strength of the associations (using Cohen’s, 1988, guidelines) were observed. For females, there was a strong positive correlation between working memory and numeracy, and moderate correlations in each of the four literacy domains. For males, there were moderate positive correlations between working memory accuracy and numeracy, reading and grammar, and small correlations between working memory accuracy and both writing and spelling. However, when the magnitude of the correlations for males and females were compared (using Fischer’s z‐test), the differences were not statistically significant. It is possible that working memory ability plays a greater role in academic achievement in females than in males, but further research with a larger sample size is needed before a strong conclusion can be drawn.

When the relationship between confidence in working memory performance and academic achievement was examined (for males and females combined), there was a moderately strong correlation between confidence and numeracy. Higher confidence was associated with higher scores in numeracy. There were no significant associations between confidence and academic achievement in any of the literacy NAPLAN domains. For males, the same pattern of results was found; a moderate association between confidence and numeracy only. For females, however, while higher confidence was still related to higher scores in numeracy, it was also related to higher scores in reading, writing and spelling.

The finding of a significant association between confidence and numeracy, but not reading or spelling, is consistent with previous findings by Freeman et al. (2017) in which academic achievement was assessed using the WIAT‐II (Wechsler, 2007), an individually administered standardised achievement test. In that study, confidence in performance in the working memory task was related to Numerical Operations, but not to Word Reading or Spelling. While the relationship between confidence and numeracy held for males in the present study, the different pattern of results for females (i.e., significant associations between confidence and achievement in reading, writing, and spelling, as well as numeracy) is a new finding.

The significant associations between working memory accuracy scores and achievement in numeracy, reading, persuasive writing, grammar, and spelling suggest children with a deficit in working memory are likely to experience difficulties in achieving in one or more of these academic abilities. The association between confidence in working memory performance and academic achievement also highlights the importance of confidence in children’s achievement academically. It is important, therefore, that we have easy‐to‐administer assessments available to allow the early detection of a working memory deficit or a lack of confidence. The WMPT is easy to administer and can be completed online, with minimal supervision. It is suitable for children from different language backgrounds and abilities. The WMPT therefore has the potential to be used as a screening test for the early detection of a working memory deficit or a lack of confidence.

Gender differences in working memory

Consistent with some previous studies that have examined gender differences in working memory performance in children (e.g., Alloway, Gathercole, & Pickering, 2006; Cameron et al., 2014), we found no difference between males and females in working memory accuracy scores. We also found no difference between males and females in confidence in their working memory performance. While previous research on gender differences in confidence has produced mixed results, with some studies reporting no differences (e.g., Kleitman & Moscrop, 2010) and others suggesting higher confidence in males than females in some domains but not others (Lundeberg et al., 1992), we are unaware of any previous studies specifically examining gender differences in children’s confidence in their performance of a working memory task.

Gender differences in academic achievement

The present findings regarding gender differences in NAPLAN scores differ from Cameron et al.’s (2014) findings. While they found males performed significantly better than females in the numeracy component, the difference between males and females in the present study failed to reach significance. The age range of children in Cameron et al.’ study (from 8 years 4 months to 12-years 4 months) was larger than in the present sample (8 years 8 months to 10 years 10 months) which could account for the different findings, if we assume gender differences in numeracy increase with age. Their overall sample size was larger (202 vs. 150 in the present study) which could also explain the difference in significance of the results across the studies. Whereas Cameron et al. found no gender differences in literacy (collapsed across reading, persuasive writing, grammar, and spelling scores), the females in our study outperformed the males in persuasive writing. As Cameron et al. did not report results for the individual NAPLAN literacy domains, it is unclear as to whether the different findings between studies are due to their use of a combined literacy score or to some other factor.

SUMMARY AND CONCLUSIONS

The present study examined the relationship between working memory ability and academic achievement in children enrolled in Year 4 in Australian primary schools. To date, only one other study of which we are aware (i.e., Cameron et al., 2014) has examined this association using the Australian national curriculum measure and an Australian cohort (i.e., children in Years 3 and 5). As previous research (e.g., Alloway & Copello, 2013; Bull et al., 2008; Gathercole et al., 2004) has shown the relationship between working memory and academic achievement may vary across age groups, at least in some domains, further research is needed to establish whether the relationship between working memory and the NAPLAN is stable across the school years.

The present results show that accuracy in working memory performance is related to academic achievement in all five NAPLAN domains and this relationship holds for both males and females. We have also shown that confidence in working memory performance is related to academic achievement, but the pattern of results differs for males and females. Whereas confidence was related only to numeracy in males, for females, significant relationships between confidence and numeracy, reading, writing and spelling were found. Further research is needed to determine whether these findings hold for children in other year levels.

In conclusion, the present study is the first study to assess the relationship between academic achievement assessed using the NAPLAN and both accuracy and confidence in performance on a working memory test (the WMPT). While the results contribute new knowledge about the relationships among these assessments, further research is needed to establish the generality of the results to other age groups and study populations.

ACKNOWLEDGEMENTS

We would like to thank the children, parents, and schools whose participation made this research possible.