Step to it – can physical activity improve kids’

cognition? A six-month longitudinal study

Adrian McPherson

Auckland University of Technology

Scott Duncan
Auckland University of Technology
Lisa MacKay
Auckland University of Technology
Jule Kunkel
Auckland University of Technology

Research Article

Keywords: Physical activity, cognition, academic performance, school, children, causation

Posted Date: December 17th, 2024

DOI: https://doi.org/10.21203/rs.3.rs-5425164/v1

License:   This work is licensed under a Creative Commons Attribution 4.0 International License.
Read Full License

Additional Declarations: No competing interests reported.

Page 1/21
Abstract
Background
Identifying the relationships among physical activity, cognition, and academic performance in children is
important for targeted public health and education initiatives. However, the majority of research has
been cross-sectional in nature; we have a limited understanding of the causal direction of these
associations. Therefore, aim of this study was to utilise longitudinal data to explore causal relationships
among physical activity, cognition and academic performance in elementary school children.

Methods
Data were sourced from 675 New Zealand children aged 5–11 years. Weekday home, weekday school,
and weekend physical activity was measured by multiple pedometer step readings, cognition by four
measures from the CNS Vital Signs assessment, and academic performance from the New Zealand
Ministry of Education Assessment Tools for Teaching and Learning (asTTle) reading and maths scores.
Measures were taken at baseline, two months, and six-month intervals. Data were analysed for 632
students identified with data for at least half of the 27 variables. A generalised linear mixed model was
used to investigate changes in physical activity, cognition and academic performance over those three
time periods while adjusting for gender, school, age, and socioeconomic status.

Results
No significant relationships were identified between physical activity and three of the cognitive domains.
However, significant, positive relationships were observed between physical activity change at two-
months and (1) composite memory change at six-months, (2) maths proficiency change at two-months,
and (3) math proficiency change at six-months. Regression coefficients suggest that a child who doubles
step count - a 100% increase in PA - will affect a 3.7% improvement in maths proficiency after two
months, and after six months affect a 2.6% improvement in maths proficiency and a 4.7% improvement
in composite memory.

Conclusions
This six-month longitudinal analysis identified that an increase physical activity led to small but
significant improvements in composite memory and maths proficiency. The small associations suggest
that substantial improvements in PA would be required to generate meaningful improvements in
cognition and academic achievement. However, timeframes longer than six-months are recommended
to identify long-term changes.

Page 2/21
Background
Walking is one of the easiest, accessible forms of exercise. Almost obsessively, people monitor step
counts aiming for a magical 10,000 steps a day for fitness. And with the growing body of research
showing high levels of physical activity (PA) have been linked with cognitive benefits [1–9], can the
simple act of walking and counting steps to promote PA have positive academic performance and
learning outcomes for children?

Exercise has exciting potential for improving both physical and cognitive abilities for children, but there is
limited information on how the relationship between PA and cognition interacts, particularly causation.
Cross-sectional studies show consistent relationships between PA and cognition [2, 10–12]. Positive
relationships have also been identified between PA and academic performance in school settings [11,
13–16]. Our previous investigation on this subject group used structured equation modelling to
demonstrate PA has independent relationships with both cognition and academic performance but could
not ascribe causation [16]. Do smart children exercise, or does exercise make children smarter? To
determine causation, changes in subjects’ performance and abilities need to be measured over time.

Findings of four key longitudinal studies are detailed below. The Vanves study was completed in 1950,
Paris, France [17]. Academic instruction was reduced by 26%, with a range of interventions added
including PA in afternoons, but children were calmer, more attentive, and school results were comparable
to other schools. The Trois Rivieres study analysed the effect of one hour extra PE for students taught by
a specialist PE teacher over a six year period [18]. The control group received 13–15% more academic
instruction than the experiment group. In the first year, the control group had higher average grades, but
in Grades 2–6, the experiment group had higher grades, significantly in years 2, 3, 5, and 6 [18]. A
secondary analysis of the ‘The Early Childhood Longitudinal Study (ECLS), Kindergarten Class of 1998 to
1999’ comparing children in low, medium and high activity groups also found girls in the high activity
group had a small benefit in mathematics and reading, but there was no positive or negative association
for boys [19]. In the last longitudinal study considered, Physical Activity Across the Curriculum (PAAC)
was a 3-year cluster randomized controlled trial in 24 elementary schools in Kansas, USA, with a primary
focus of decreasing BMI and improving physical health of students and a secondary aim to assess
changes in academic achievement [3]. The experiment classes engaged in 90 minutes additional PA per
week and were found to score significantly better than the control group for reading, writing,
mathematics and oral language skills [3].

In further studies, Haapala et al studied 635 children aged 11–13 years, and found positive baseline
correlations between Moderate to Vigorous Physical Activity (MVPA) and grade point average (GPA) [20].
MVPA was assessed by self-reported responses to the question, “Over the past 7 days, on how many
days were you physically active for a total of at least 60 minutes per day?” Students were assessed
almost two years later, when it was found MVPA was associated with better GPA in boys, but not girls.
That finding differs with studies showing cognitive improvement with increased PA. The authors state
contrasts may be due to their study controlling for academic achievement at baseline which is the

Page 3/21
strongest predictor of academic achievement at follow up [20]. That may be correct, but our previous
study found the relationship between PA and academic performance independent of the child’s cognitive
ability [16]. The lack of relationship between MVPA and GPA identified by the authors may be due to the
poor reliability of self-reported data and the timeframe between the analyses. Also, although measures
were taken at two timepoints, there were no analyses of subjects’ change over time, so this was
effectively two cross-sectional studies and not a longitudinal study.

In a study of 902 Danish school aged 7–12 years over a three-year period, academic performance was
measured in maths and Danish, and PA was measured using an accelerometer for at least four full days
at four time points [21]. Interestingly, they found both MVPA and sedentary time were directly associated
with academic performance. It was theorised that sedentary time was associated with study time for
this population. Wickel measured MVPA of 1364 children at baseline then again six years later [22]. At
the six-year point, children’s cognition was also measured. Because there were no baseline cognitive
measures, the study could only identify cross-sectional relationships between MVPA and cognition at the
second timepoint. Against the theory PA improves cognition, the Wickel found sedentary time was
positively associated with executive function and increases in PA were inversely related with executive
function. However, the lack of baseline cognitive data and six-year period between MVPA assessments
impact the ability to draw long-term conclusions. Hence, the author also acknowledges a gap in the
knowledge base about the associations among PA, sedentary time and cognition in scientific literature.

Petrigna et al, completed a systematic review to see if learning through movement improves academic
performance in primary children [23]. They identified 54 articles finding a range of simple PAs had
positive associations with maths, attention, and other academic scores. However, all of the interventions
considered were integrated as part of the overall classroom curriculum making it difficult to ascribe
relationships, and there were no longitudinal studies so it was not possible to identify causation.

A thorough meta-analyses specifically investigating the relationship between PA and cognition in

children was completed by Donnelly et al [24] in 2016. The authors started from 6,237 articles but using
the 27 point Downs and Black checklist that considers methodology rigor [25], only identified 137 articles
suitable to consider [24]. The Downs and Black checklist considers methodological strengths including
validity characteristics, clarity of hypothesis and outcome measure details, participant compliance, and
study power [25]. The review found PA has a positive influence on cognitive function as well as brain
structure and function but noted limitations on conclusions due to weaknesses including a lack of
information about estimates of random variability in the outcome data, statistical power not being
stated, larger sample sizes needed, and lack of randomised controlled trials [24]. They only identified two
longitudinal studies that were robust enough to be considered in their review [24]. Particularly, they
advise more research is necessary to establish causality, to determine mechanisms, and to investigate
long-term effects [24].

It is essential to consider the differences of measures and methodology to understand the PA, cognition,
academic performance relationship. For example, a 2020 meta-analysis compared the effects on

Page 4/21
cognition of closed skill exercise (CSE) such as running and swimming with open skilled exercise (OSE)
where environment and movement needs to be continually adapted such as basketball and soccer on
children and adults [12]. Cross-sectional studies found OSE was superior to CSE with a small effect
regarding cognitive performance, inhibition and cognitive flexibility. Among the four intervention studies
examined, no significant differences were observed between OSE and CSE. Findings only partially
supported the hypothesis that OSE is superior to CSE in terms of executive function. Further to the CSE
OSE debate, an Italian study compared the effect of 30 minute bouts of CSE with OSE on short-term-
memory on 125 children aged 7–10 years [26]. OSE had positive effects on STM for the full sample and
of CSE had positive effects on STM for children 9–10 years, but not for children aged 7–8 years.

These longitudinal studies show PA likely has causal links with cognition, but each study has
shortcomings that limit conclusions on causation. It is difficult to generalise the Vanves findings
because experimental sample was small, it is not clear how the control group was matched in terms of
size and SES, and the treatment included more than just PA [17]. The ECLS study found differences
between boys and girls [19], but none of the other studies consider gender effects. SES is recognised as
one of the main influences on children’s academic success [13, 27, 28], but none of the papers adjust for
SES.

Therefore, the aim of this study was to explore causal relationships between PA, cognition, and
academic performance over a six-month longitudinal period for primary school children aged 7–10
years.

Methods
Participants
A total of 675 participants (326 male, 349 female) were part of an eight-week randomised controlled
trial: Healthy Homework was a curriculum-based, classwork and homework schedule designed to
promote PA and healthy eating [29]. Full details of the Healthy Homework programme are described in
its pilot study [30]. All measurements were taken at baseline, immediately post-intervention, and six-
month post-intervention. The study comprised eight control and eight experimental schools. For the
purposes of this study, data from both groups were used for analysis – see later under ‘Statistical
Analysis’ for rationale and statistical consideration. Eligibility criteria for the schools were as follows: a
school with more than 100 students, location within Auckland or Dunedin cities, and a contributing, full
primary, or composite structure that included at least one class each of students in school years 3–5. A
total of 16 primary schools from Auckland (n = 10) and Dunedin (n = 6) were selected to participate in the
study. Socioeconomic decile ratings of participating schools ranged from 3 to 10 (median [IQR] = 8 [6,
9]). Decile is a New Zealand Ministry of Education (MoE) socioeconomic rating system for school
funding based on SES with 1 being low and 10 being high. Decile is a rating of the whole school, and not
specific to individual students. It is common to have children from a range of SES within one school, with
socioeconomic decile being an average representation of the school’s surrounding area. Students were

Page 5/21
selected to participate from one Year 3, one Year 4, and one Year 5 class from each school; simple
random sampling was used in instances where there were two or more classes per year. All children in
each participating class were invited to take part in the evaluation (i.e., no formal inclusion or exclusion
criteria). Written parental consent and assent was obtained for children to participate in the study.
Ethical approval was obtained from the Auckland University of Technology Ethics Committee (10/159).
The Healthy Homework study only measured changes in PA and diet and found the programme was
successful resulted in substantial and consistent increases in PA and had limited effects on body size
and fruit consumption [29]. That original study obtained data on academic performance and cognitive
ability, but those were never analysed. This second analysis investigates longitudinal changes and
relationships between PA, cognition and academic performance.

Measures
PA was assessed using sealed NL-1000 pedometers (New Lifestyles Inc, Lee’s Summit, MO) over five
consecutive days (three weekdays, two weekend days). Research has established the validity of these
NL-1000 pedometers for measuring steps in children [31]. NL-1000 pedometers have a multiday memory
that automatically categorizes data according to the day of the week which enables step count for
weekdays and weekends to be collected [32]. Pedometers were used to gain three measures of PA:
average weekday steps at home, average weekday steps at school, and average steps at weekend.

Image 1. NL-1000 pedometer used in the study to record children’s steps.

The cognitive abilities of children were measured using CNS Vital Signs (CNSVS): a standardised
cognitive screen assessment suitable for participants aged 7–90 years [33]. CNSVS is a web-based
assessment battery with seven tests that are scored individually and combined to give scores in nine
different areas. Four of the nine CNSVS domains were considered for this study: Composite Memory
(recognize, remember, and retrieve words and geometric figures), Executive Function (recognize rules,
categories, and manage or navigate rapid decision making), Psychomotor Speed (perceive, attend,
respond to complex visual-perceptual information and perform simple fine motor coordination), and
Reaction Time (react, in milliseconds, to a simple and increasingly complex direction set) [34]. The other
domains could not be used because of the difficulty in administering the Complex Attention Test, and the
four remaining domains used combinations of the same base assessment.

Academic performance was measured using the New Zealand Ministry of Education electronic
Assessment Tools for Teaching and Learning (e-asTTle). The e-asTTle assessments have more than
2,000 curriculum-based assessment items standardised on over 50,000 students covering curriculum
levels 2—4 to assess student’s achievement and progress in reading, writing and mathematics and the
New Zealand native language Māori equivalents of panui, tuhituhi, and pangarau [35–38]. Measures are
norm-referenced and used to evaluate children’s progress through the school year [36]. Teachers create
their own multi-choice assessment as the e-asTTle software generates a test that selects the best set of
items meeting the teacher’s content and difficulty constraints [37]. For the purpose of this research, a
Page 6/21
research team conducted both the reading and maths assessments, which were done using pen and
paper with a time limit. Researchers marked total scores (0–12), and results were entered into a
computer by research assistants. Testing was completed within 10 minutes. The e-asTTle software
converts raw scores into measures that align with a child’s curricular needs [37]. Raw scores were
sufficient for the current analyses because they give a measure of academic performance for students in
relation to peers of the same school year. Demographic information was obtained from the school
records and included gender, age, school, ethnicity and decile.

Study Protocol
Two pedometers were assigned to each child: one clearly labelled ‘School’ and the other ‘Home’. The
‘School’ pedometer was worn during school hours, while the ‘Home’ pedometer was left inside; a
collection tray in the classroom. At the end of the school day, each child placed their ‘School’ pedometer
in the tray and attached their ‘Home’ pedometer. Parents were given instructions how to attach the
‘Home’ pedometer to the child when he/she got up in the morning and take it off when before going to
bed at night. Upon arrival at school the next day, the teacher reminded the children to switch over their
pedometers again. Pedometers were issued to children and height and weight measures taken on one a
separate day within a month by trained researchers. For each school, CNSVS and e-asTTle baseline
measures were collected by a team of researchers on one day. CNSVS assessment was completed
before the e-asTTle test, with at least 30 minutes between the two. The CNSVS assessment was
conducted in groups using school computer facilities or libraries and assisted by at least three
researchers. Group sizes and types of computers depended on the facilities and computers provided by
the school. Researchers introduced the test beforehand while each instruction for each test appeared on
the screen before each test started. CNSVS was introduced for the research purposes and not part of
routine school assessment practice. Thus, as it is not part of the students’ normal education practice
and procedures, they may have struggled with it being an unfamiliar task and not necessarily had
difficulty with the cognitive demands and content. Researchers were available for the children in case
they did not understand the instructions or if children clicked it away too quickly. The e-asTTle
assessments were introduced and explained by the researchers. While the attitude questions were read
out by the researchers, waiting for all children to go through them and ensuring that they understand
them, the reading test was then conducted before the math test with a time limit of ten minutes.
Students are used to e-asTTle assessments through the year as part of their normal school routines.

Statistical Analysis
All variables were checked for normality, skewness and outliers. Three students were identified to have
special needs and removed from the analysis because the cognitive and academic measures are not
specific enough to cater for their needs and abilities. The distribution of the CNSVS composite memory
item was skewed positively, but that reflects what is to be expected in the general population thus data
were not transformed [33]. The other CNSVS measures were normally distributed. The two asTTle

Page 7/21
variables were normally distributed with no problematic outliers. One problematic outlier was identified
with weekday steps which was clearly a data entry mistake. As all other variables were appropriate for
the subject, the weekday steps value was removed and a new value was imputed later as part of the
missing values analysis (MVA). The final analyses used the total weekly steps, which was gained by the
formula: (mean weekday steps home x 5) + (mean weekday steps school x 5) + (mean weekend steps x
2).

The extent of missing values was assessed on the full study cohort. To minimize loss of data, subjects
with data for at least half of variables included in the final model were retained. Our study sample was
reduced from 675 to 632. Details of missing data from the 675 and 632 subjects is included in Appendix
1. An MVA was completed on the three pedometer step readings. Data were not found to be Missing
Completely at Random (MCAR; Little's MCAR test: Chi-Square = 684.058, DF = 595, Sig. = 0.007). The
researchers inspected the data visually and could not see any patterns for missing IV data, so data was
presumed to be Missing at Random (MAR). Expectation Maximisation (EM) was then used to impute
missing values for the IVs. An MVA was then completed on the four CNSVS measures and two asTTle
measures. The data were not found to be MCAR (Little's MCAR test: Chi-Square = 1358.221, DF = 1172,
Sig. = 0.000). Based on inspection of missing data patterns, data are assumed to be MAR. Missing data
for CNSVS and asTTle measures is due to a child not being present in class when the test was being
taken. EM was then used to impute missing values. In a detailed study, Dong and Peng found as long as
data are MAR, EM data imputation produced statistically significant results to p < .001 when removing
20%, 40% and 60% data from a complete dataset of 432 subjects [39].

This present study did not evaluate differences between experiment and control groups. Whilst the
intervention of the experiment group aimed to improve diet and PA, there were no interventions directed
at cognition and academic performance. This present study sought to investigate changes over time in
the new areas of cognition and academic performance, and their relationship to PA. The original study
followed RCT methodology including randomly assigning schools to experiment and control group. Thus,
adjusting school effects in our analyses account in for any potential effect of both school clusters and
the RCT intervention impact on the experiment group. Furthermore, adjusting for school effects also
addresses any possible actual or placebo benefit caused differences subjects’ performance.

Changes in the total weekly PA, the six cognitive domains, and the two academic outcomes were
analysed over the two-month and six-month periods using generalised linear mixed models (GLMMs).
GLMM was the most appropriate methodology for analysis because we measured change in the
population as a whole, and did not analyse differences between groups. Physical activity change over
two months was compared with cognitive and academic change over both two and six months (12
models in total). The GLMM analysis adjusted for fixed (age, gender, decile) and random (subjects
nested in schools) effects. Although decile is different from SES as stated in methods, adjusting for
decile has the same effect as adjusting for SES. All analyses were completed using IBM SPSS 24
(Armonk, NY: IBM Corp). Anthropometric data of subjects was obtained and used in the initial analyses,

Page 8/21
but analyses showed no difference in findings with or without that data. The final analyses did not
include anthropometric data to increase model parsimony.

Results
Assumptions
Demographic data for the full 632 students from this analysis (48.7% male) aged 5.2–10.8 years residing
in New Zealand were available for analyses (Table 1). There was an even spread of children across the
three school years (3: 32.3%, 4: 34%, 5: 33.7%). The majority of students were of New Zealand European
ethnicity (69.9%). Students were from schools of predominantly high socioeconomic decile. Tables 2 and
3 show the with mean, median, SD and inter-quartile range for step counts and cognitive/academic data
(respectively) for the 632 students considered in this analysis, with EM imputations for missing data.
Table 2 and Fig. 1 show total average steps for students was consistent for the three timepoints and that
weekend day average steps were lower than weekday average steps.

Page 9/21
 Table 1
Sociodemographic characteristics of the study sample.
Age

Male Female Total

N M + SD Min + N M + SD Min + N M + SD Min +

Max Max Max

School 98 7.74, ± 6.65, 106 7.72, ± 6.48, 204 7.73, ± 6.48,

Year 3 0.56 9.21 0.67 9.25 0.62 9.25

School 104 8.70, ± 7.60, 111 8.72, ± 5.21, 215 8.71, ± 5.22,
Year 4 0.60 9.86 0.66 9.89 0.64 9.89

School 106 9.64, ± 8.11, 107 9.74, ± 6.88, 213 9.69, ± 6.88,
Year 5 0.52 10.84 0.65 10.8 0.59 10.8

Total 308 8.72, ± 6.65, 324 8.73, ± 5.22, 632 8.73, ± 5.22,
0.96 10.8 1.05 10.8 1.00 10.8

Ethnicity

Male Female TOTAL

Māori 19 (6.2%) 25 (7.7%) 44 (7%)

Pacific 12 (3.9%) 11 (3.4%) 23 (3.6%)

Island

Asian 34 (11.0%) 70 (21.6%) 104 (16.5%)

Other 9 (2.9%) 12 (3.7%) 21 (3.3%)

NZ 234 (76%) 206 (63.6%) 440 (69.6%)

European

Total 308 324 632

Decile

Decile 3 23 (7.5%) 20 (6.2%) 43 (6.8%)

Decile 4 9 (2.9%) 11 (3.4%) 20 (3.2%)

Decile 5 13 (4.2%) 38 (11.7%) 51 (8.1%)

Decile 6 45 (14.6%) 52 (16%) 97 (15.3%)

Decile 7 46 (14.9%) 48 (14.8%) 94 (14.9%)

Decile 8 60 (19.5%) 47 (14.5%) 107 (16.9%)

Decile 9 47 (15.3%) 40 (12.3%) 87 (13.8%)

Decile 10 65 (21.1%) 68 (21%) 133 (21%)

Page 10/21
Age

Total 308 324 632

School year

Year 3 98 (31.8%) 106 (32.7%) 204 (32.3%)

Year 4 104 (33.8%) 111 (34.3%) 215 (34%)

Year 5 106 (34.4%) 107 (33%) 213 (33.7%)

Total 308 324 632

Table 2
Descriptive statistics of the three step count measurements.
Baseline

Mean Median SD 25th %ile 50th %ile 75th %ile

Weekday (home) 5118 4995 2027 3896 4996 6054

Weekday (school) 5447 5219 2118 3988 5219 6410

Weekend day 7507 7235 3448 5337 7235 8785

Daily average* 9691 9490 2914 7728 9490 11211

Two-months

Mean Median SD 25th %ile 50th %ile 75th %ile

Weekday (home) 5288 5214 2193 3972 5214 6288

Weekday (school) 5944 5796 2397 4400 5796 5796

Weekend day 8279 7899 3433 6244 7899 9774

Daily average 10388 10121 3139 8264 10121 12195

Six-months

Mean Median SD 25th %ile 50th %ile 75th %ile

Weekday (home) 5235 5214 2179 3833 5214 6081

Weekday (school) 5997 5796 2190 4734 5796 6891

Weekend day 7909 7899 3225 6107 7899 9200

Daily average 10283 10121 3028 8357 10121 11971

Page 11/21
 Table 3
Descriptive statistics of the dependent variables of four cognitive domains and two
academic domains for 632 students considered in final analyses including EM imputed
data for missing values.
Baseline

Mean Median SD 25th %ile 50th %ile 75th %ile

Composite Memory 73 76 27.6 54 76 95

Executive Functioning 95 96 15.3 84 96 105

Psychomotor Speed 93.9 94.9 13 85 94.9 102

Reaction Time 95.6 95.5 18.9 83 95.5 107

Reading Proficiency 5.08 5.83 3.03 2 5.83 8

Maths Proficiency 6.06 6 3.2 3.57 4 6

Two-months

Mean Median SD 25th %ile 50th %ile 75th %ile

Composite Memory 75.7 78 25.9 58 78 95

Executive Functioning 103 102 16 92 102 114

Psychomotor Speed 97.3 97 13.6 89 97 106

Reaction Time 96.6 96 18.2 86 96 108

Reading Proficiency 6.14 6 2.59 4 6 8

Maths Proficiency 6.56 6.66 2.87 4 6.66 9

Six-months

Mean Median SD 25th %ile 50th %ile 75th %ile

Composite Memory 79 79 22.7 67 79 95

Executive Functioning 104 103 15 96 103 115

Psychomotor Speed 97.4 96.2 12.3 89.5 96.1 105

Reaction Time 94.2 95.5 16.6 84 95.4 104

Reading Proficiency 6.34 6 2.42 5 6 8

Maths Proficiency 7.39 7 2.38 6 7 9

Table 4 shows the mean change for cognitive and academic domains at the two-month and six-month
intervals from the 12 generalised mixed models, adjusted for 2-month physical activity change, age, sex,
socioeconomic status (decile), and school clustering. The β coefficient indicates the percentage change
Page 12/21
to each domain associated with a 1% increase in PA at two-months. Significant, positive relationships
were observed between PA change and composite memory change at six-months (0.021), and nearing
significance for change in composite memory at two-months (0.051). PA change and maths proficiency
change were significant at two-months (0.019) and six-months. at six-months (0.034). No other
associations were significant.

Table 4
Associations between changes in physical activity at two-months with changes
in cognitive/academic outcomes at two-months and six-months.
Domain Mean % change (LCL, UCL) β (LCL, UCL) P

Composite Memory change

at 2 months 19.5 (14.0, 25.4) 0.048 (0.000, 0.097) 0.051

at 6 months 29.7 (22.9, 36.8) 0.047 (0.007, 0.087) 0.021

Reaction Time change

at 2 months 3.55 (1.68, 5.58) -0.009 (-0.157, 0.139) 0.904

at 6 months 1.73 (-0.114, 3.72) -0.028 (-0.176, 0.121) 0.714

Psychomotor Speed change

at 2 months 4.68 (3.62, 5.77) -0.013 (-0.260, 0.234) 0.917

at 6 months 5.10 (3.83, 6.37) -0.012 (-0.240, 0.216) 0.921

Executive Function change

at 2 months 9.31 (7.92, 10.5) 0.123 (-0.083, 0.330) 0.242

at 6 months 11.0 (9.60, 12.5) 0.138 (-0.055, 0.331) 0.162

Reading Proficiency change

at 2 months 72.9 (63.1, 84.0) 0.011 (-0.014, 0.037) 0.385

at 6 months 84.8 (72.2, 98.0) 0.010 (-0.014, 0.034) 0.402

Maths Proficiency change

at 2 months 41.5 (32.9, 50.5) 0.037 (0.006, 0.068) 0.019

at 6 months 70.3 (58.5, 82.4) 0.026 (0.002, 0.050) 0.034

β = standardised coefficient; LCL = lower 95% confidence limit, UCL = upper 95% confidence limit using
bias-corrected bootstrapping.

Discussion

Page 13/21
To our knowledge, this is the first study investigating the effects of changes in PA on change in cognitive
ability and academic performance in school children over a six-month period. Our findings suggest that
small gains in specific areas of cognition and academic function – namely composite memory and
maths – can be obtained with increased PA. That is acknowledged by Ziv et al, who say findings
regarding the effects of exercise on cognitive performance are also usually modest [40]. Although the
gains indicated are small, success is a series of small victories, and they could represent a meaningful
impact for children and their learning. For example, a 1% increase in PA after two months was
associated with a 0.037% increase in maths proficiency, and after six months was associated with a
0.047% increase in composite memory and 0.026% increase for maths proficiency. Thus, if students
doubled their PA (100% increase) with a simple, closed skill exercise activity as walking, that would
theoretically affect a 3.7% or 2.6% increase in maths proficiency and a 4.7% improvement in composite
memory.

The present results concur with other studies that have found increased PA is associated with
improvement in executive function, memory and maths [14, 41]. Other longitudinal studies have indicated
that PA has a positive impact on maths and reading scores [3, 19]. Importantly, the present analyses
adjusted for potential confounding factors such as gender [19], age [10, 11], and the impact of SES
through the socioeconomic decile differences between schools [13, 27, 28]. This study found that the
significant relationships PA had on composite memory and maths proficiency is independent of such
confounding factors.

Studies have highlighted possible methodological flaws that may bias research to support the PA-
cognition relationship [40] [42, 43]. For example, randomised control trials (RCT) may have a tendency to
have higher performing subjects in the experiment group and pretest and post-test equivalence need to
be verified [42]. As our study used data from control and experiment groups from a RCT trial and
adjusted for differences between subjects and schools, neither of these were an issue. Ciria et al, say
there is a preference for RCT studies which are seen as a gold-standard to ascertain causal links, but
other sources of empirical evidence, such as observational or epidemiological studies, should also be
considered in their ability to determine causation [43]. Our study is an observational longitudinal study
with rigorous methodology that identified small but valid changes inferring causation. Ziv et al note with
the placebo effect, it is important control and experiment groups have similar expectations of input [40].
Our experiment and control groups had similar input with all students given pedometers to motivate
them. Again, to counter such a bias, our study pooled data and adjusted from both groups, thus negating
such concerns. Publication bias finds studies with large and positive changes are more prevalent [43],
whereas the effects of PA on cognitive performance are usually modest [40]: "We believe this
exponential accumulation of low-quality evidence has led to stagnation rather than advance in the field
hindering the discernment of the real existing effect." [43]

Our study did not identify any significant relationships between increased PA and the three other
cognitive tested or reading proficiency. Most other longitudinal studies that investigated the relationship
between PA and cognition analyse change over periods longer than six-months [3, 18, 19, 44]. It is

Page 14/21
possible that two-months and six-months were not a long enough time span to notice gradual cognitive
changes. In addition, the measures used in this study have potential limitations. Pedometers give a valid
and reliable indicator of overall volume of physical activity and have been used widely among student
populations [31], but do not consider the intensity of the steps or time of day. High intensity aerobic
activity and activity immediately prior cognitive assessment have been linked to greater cognitive
function and academic performance [2, 4, 11]. Furthermore, pedometers do not monitor other aspects of
fitness that have been linked to cognitive function such as acute effects of activity, cardiorespiratory
fitness, resistance exercise, or combinations of open skill exercise and activity [11, 13, 15]. Similarly, the
two e-asTTle measures are well researched and robust, but additional school-based assessments such
as writing could provide greater insights to children’s academic performance. The CNSVS measures
used in this study give a good insight to cognitive function, but as a screen assessment CNSVS may not
have been sensitive enough to detect changes in some areas. Thus, a significant relationship was only
identified in composite memory. Students were not familiar with the CNSVS assessment and thus
results may reflect this unfamiliarity rather than difficulty with the cognitive demands and content. A last
possible limitation is use of school decile as a measure of SES. It would have been better to have the
SES for each individual child and not the school as a whole. Lastly, the lack of relationships may simply
mean there were no relationships with increased PA and those domains with this population in this
study.

Identifying causation is one of the key questions in the PA/cognition field. Are smart children active or
does being active make children smart? The results of this study provide some evidence that the more a
child increases PA, the greater the improvement in memory and maths after six months. To examine the
relationships further, future studies could consider wider ranges of PA, robust paediatric cognitive
assessment, alternative academic performance measures. Further, as this study demonstrated small
cognitive and academic gains over a short period, future studies should be completed over longer
timeframes which will give greater opportunity to identify how changes in PA can make larger
quantifiable changes in cognition and academic performance.

Conclusions
This six-month longitudinal study provides some support for the theory that increased PA improves
cognition and academic performance in children. The analysis identified after adjustment for age, sex,
socioeconomic status, and school clustering, increased PA was associated with small but significant
improvements in composite memory and maths but not for executive function, psychomotor speed,
reaction time, or reading proficiency. While this reinforces that PA may have a role to play in children’s
learning, the relatively small magnitude of the associations suggests that substantial improvements in
PA would be required to generate meaningful improvements in cognition and academic achievement.
Further research into the long-term effects of PA on brain function would provide additional information
regarding the potential benefits of increasing PA in school children.

Page 15/21
Abbreviations
CSE: Closed skill exercise

OSE: Open skill exercise

CNSVS: CNS Vital Signs

GLMM: General Linear Mixed Model

MAR: Missing at random

MCAR: Missing completely at random

MNAR: Missing not at random

MVA: Missing values analysis

MVPA: Moderate to Vigorous Physical Activity

PA: Physical Activity

PE: Physical education

RCT: Randomized control trial

SES: Socio-economic status

Declarations
Acknowledgements

The authors would like to thank all participating schools, children, and parents for contributing their time
and effort to this study. I would also like to thank my son, Campbell McPherson, for his assistance in the
data preparation.

Funding

This study was funded by a project grant from the Health Research Council of New Zealand (10/207).
The funder had no involvement in the design of the study, the collection, analysis or interpretation of the
data, or in writing the manuscript.

Availability of data and materials

The datasets generated and/or analysed during the current study are not publicly available due to
confidentiality but are available from the corresponding author on reasonable request.

Page 16/21
Authors’ contributions

AM Study design, analysis and manuscript preparation.

SD Study design, data curation, and manuscript review.

LM Contribution to analysis and manuscript review.

JK Study design, data collection, and manuscript review.

All authors read and approved the final manuscript.

Ethics approval and consent to participate

Written informed parental consent and personal assent was obtained from each participant for the
collection and use of the data in future publication. Ethical approval for the study was obtained from the
Auckland University of Technology Ethics Committee (10/159).

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests

References
1. Best, J.R., Effects of physical activity on children’s executive function: Contributions of experimental
research on aerobic exercise. Developmental Review, 2010. 30: p. 331-351.
2. Chaddock, L., et al., A Review of the Relation of Aerobic Fitness and Physical Activity to Brain
Structure and Function in Children. Journal of the International Neuropsychological Society, 2011.
17: p. 975-985.
3. Donnelly, J.E., et al., Physical Activity Across the Curriculum (PAAC): A randomized controlled trial to
promote physical activity and diminish overweight and obesity in elementary school children.
Preventive Medicine, 2009. 49: p. 336-341.
4. Hillman, C.H., K. Kamijo, and M. Scudder, Review: A review of chronic and acute physical activity
participation on neuroelectric measures of brain health and cognition during childhood. Preventive
Medicine, 2011. 52(Supplement): p. S21-S28.
5. Mahar, M.T., Impact of short bouts of physical activity on attention-to-task in elementary school
children. Preventive Medicine, 2011. 52(Supplement): p. S60-S64.
6. Amika S Singh, E.S., Vera van den Berg, Léonie Uijtdewilligen, Renate H M de Groot, Jelle Jolles, Lars
B Andersen, Richard Bailey, Yu-Kai Chang, Adele Diamond, Ingegerd Ericsson, Jennifer L Etnier, Alicia

Page 17/21
 L Fedewa, Charles H Hillman, Terry McMorris, Caterina Pesce, Uwe Pühse, Phillip D Tomporowski,
Mai J M Chinapaw, Effects of physical activity interventions on cognitive and academic performance
in children and adolescents: a novel combination of a systematic review and recommendations
from an expert panel. British Journal of Sports Medicine, 2019. 53: p. 640–647.
7. Yue Xue, Y.Y., Tao Huang, Effects of chronic exercise interventions on executive function among
children and adolescents: a systematic review with meta-analysis. British Journal of Sports
Medicine, 2019. 0: p. 1-9.
8. Spyridoula Vazou, C.P. and a.A.S.-O. Kimberley Lakes, More than one road leads to Rome: A narrative
review and metaanalysis of physical activity intervention effects on cognition in youth. Internationl
Journal of Sport and Exercise Psychology, 2019. 17(2): p. 153-178.
9. Sandra Amatriain-Fernández, M.E.G.-N., Henning Budde, Effects of chronic exercise on the inhibitory
control of children and adolescents: A systematic review and meta-analysis. Scandinavian Journal
of Medicine and Science in Sports, 2021. 31: p. 1196–1208.
10. Khan, N.A. and C.H. Hillman, The Relation of Childhood Physical Activity and Aerobic Fitness to
Brain Function and Cognition: A Review. Pediatric Exercise Science, 2014. 26(2): p. 138-146.
11. Hillman, C.H., et al., Aerobic Fitness and Cognitive Development: Event-Related Brain Potential and
Task Performance Indices of Executive Control in Preadolescent Children. Developmental
Psychology, 2009. 45(1): p. 114-129.
12. Hao Zhu, A.C., Wei Guo, Fengshu Zhu, Biye Wang, Which Type of Exercise Is More Beneficial for
Cognitive Function? A Meta-Analysis of the Effects of Open-Skill Exercise versus Closed-Skill
Exercise among Children, Adults, and Elderly Populations. Applied Sciences, 2020. 10(8): p. 1-15.
13. Dwyer, T., et al., Relation of academic performance to physical activity and fitness in children.
Pediatric Exercise Science, 2001. 13(3): p. 225-237.
14. Davis, C.L., et al., Exercise Improves Executive Function and Achievement and Alters Brain Activation
in Overweight Children: A Randomized, Controlled Trial, Health Psychology. Health Psychology,
2011. 30(1): p. 92-98.
15. Chang Hung, C. and C. Jui-Fu, The Relationship between Physical Education Performance, Fitness
Tests and Academic Achievement in Elementary School. International Journal of Sport & Society,
2011. 2(1): p. 65-73.
16. McPherson A, M.L., Kunkel J, Duncan S, Physical activity, cognition and academic performance: an
analysis of mediating and confounding relationships in primary school children. BMC Public Health,
2018. 18(1).
17. Shephard, R.J., Curricular Physical Activity and Academic Performance. Pediatric Exercise Science,
1997. 9(2): p. 113.
18. Shephard, R.J., et al., Academic skills and required physical education: The Trois Rivieres
Experience. Canadian Association for Health, Physcial Education and Recreation - Research
Supplement, 1994.

Page 18/21
19. Carlson, S.A., et al., Physical education and academic achievement in elementary school: data from
the early childhood longitudinal study. American Journal of Public Health, 2008. 98(4): p. 721-727.
20. Haapala, E.A.H., H.L.; Syvaoja, H.; Tammelin, T.H.; Finni, T.; Kiuru, N., Longitudinal associations of
physical activity and pubertal development with academic achievement in adolescents. Journal of
Sport and Health Science, 2020. 9(3): p. 265-273.
21. Lima, R.A.P., K.A.; Moller, N.C.; Anderson, L.B.; Bugge, A.;, Physical Activity and Sedentary Time Are
Positively Associated With Academic Performance: A 3-Year Longitudinal Study. Journal of Physical
Activity & Health, 2019. 16(3): p. 177-183.
22. Wickel, E.E., Sedentary Time, Physical Activity, and Executive Function in a Longitudinal Study of
Youth. Journal of Physical Activity & Health, 2017. 14: p. 222-228.
23. Petrigna L, T.E., Brusa J, Rizzo F, Scardina A, Galassi C, Lo Verde D, Caramazza G, Bellafiore M, Does
Learning Through Movement Improve Academic Performance in Primary School children? A
Systematic Review. Frontiers in Pediatrics, 2022.
24. Donnelly, J.E., et al., Physical activity, fitness, cognitive function and academic achievement in
children: A systematic review. Medicine & Science in Sports & Exercise, 2016. 48(6): p. 1197-1222.
25. Downs, S.H. and N. Black, The Feasibility of Creating a Checklist for the Assessment of the
Methodological Quality Both of Randomised and Non-Randomised Studies of Health Care
Interventions. 1998, British Medical Association. p. 377.
26. Giovanni Ottoboni, A.C., Alessia Tessari, The Effect of Structured Exercise on Short-Term Memory
Subsystems: New Insight on Training Activities. International Journal of Environmental Research
and Public Health, 2021. 18(14): p. 1-10.
27. Trudeau, F. and R.J. Shephard, Physical education, school physical activity, school sports and
academic performance. 2008.
28. Florence, M.D., M. Asbridge, and P.J. Veugelers, Diet Quality and Academic Performance. Journal of
School Health, 2008. 78(4): p. 209-215.
29. Duncan, S.J., et al., Healthy Homework: A Physical Activity and Nutrition Intervention for Children
Research Project Full Application (GA210F). 2010, Auckland University of Technology: Auckland:
New Zealand.
30. Duncan, S.J., et al., Efficacy of a compulsory homework programme for increasing physical activity
and healthy eating in children: the healthy homework pilot study. International Journal of Behavioral
and Nutrition and Physical Activity, 2011. 8(127).
31. Duncan, S., et al., Effects of age, walking speed, and body composition on pedometer accuracy in
children. Research quarterly for exercise and sport, 2007. 78(5): p. 420-428.
32. Duncan, E.K., J.S. Duncan, and G. Schofield, Pedometer-determined physical activity and active
transport in girls. International Journal of Behavioral Nutrition & Physical Activity, 2008. 5: p. 1.
33. Gualtieri, T. and L.G. Johnson, Reliability and validity of a computerized neurocognitive test battery,
CNS Vital Signs. Archives of Clinical Neuropsychology, , 2006. 21(7): p. 623-643.

Page 19/21
34. CNS Vital Signs - Clinical Practice Test Domains. 2018 [cited 2018 2018, June 18]; Available from:
http://www.cnsvs.com/ClinicalPractice.html.
35. Ministry of Education. e-asTTle, www.e-asttle.tki.org.nz, New Zealand. 2016.
36. Hattie, J.A.C., et al. Validation Evidence of asTTle Reading Assessment Results: Norms and Criteria.
asTTle Tech. Rep. 22. 2003.
37. Hattie, J.A.C., G.T.L. Brown, and P.J. Keegan, A National Teacher-Managed, Curriculum-Based
Assessment System. International Journal of Learning, 2003. 10: p. 771-778.
38. Lavery, L. and G.T.L. Brown, Overall Summary of Teacher Feedback from the Calibrations and Trials
of the asTTle Reading, Writing, and Mathematics Assessments Technical Report 33, University of
Auckland, 2002: p. 1-7.
39. Dong, Y. and C.Y.J. Peng, Principled missing data methods for researchers.
40. Ziv, G.L., O.; Netz, Y., Selecting an appropriate control group for studying the effects of exercise on
cognitive performance. Psychology of Sport and Exercise, 2024. 72.
41. Hollar, D., et al., Effect of a two-year obesity prevention intervention on percentile changes in body
mass index and academic performance in low-income elementary school children. American
Journal of Public Health, 2010. 100(4): p. 646-653.
42. Liu, S.L., J.C.; Tenenbaum, G., Does Exercise Improve Cognitive Performance: A Conservative
Message from Lord's Paradox. Frontiers in Psychology, 2016. 7.
43. Ciria, L.F.R.-C., R; Vadillo, M.A.; Holgado, D.; Luque-Casado, A.; Perakakis, P.; Sanabria, D., An umbrella
review of randommized control trials on the effects of physical exercise on cognition. Nature Human
Behaviour, 2023. 7(6).
44. Dwyer, T., et al., An investigation of the effects of daily physical activity on the health of primary
school students in South Australia. International Journal of Epidemiology, 1983. 12(3): p. 308-313.

Image 1
Image 1 is available in the Supplementary Files section.

Figures

Page 20/21
Figure 1

Graph of average daily steps at three timepoints in study.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.

Image1.png
Appendix.docx

Page 21/21