What factors relate to three profiles of perception of motor competence in young children?

ABSTRACT The study aims were to 1) examine profiles of perception of motor competence (PMC) in relation to actual motor competence (AMC), i.e. under-estimators (UEs), realistic estimators (REs) and over-estimators (OEs) and 2) investigate associations between the profiles and selected socioecological factors at the individual, family and environmental levels. PMC (Pictorial Scale of Perceived Movement Skill Competence) and AMC (Test of Gross Motor Development-Third Edition) were administered to a representative sample of children from 37 childcare centres in Finland (n=441;6.2±0.6yrs;52% boys). Socioecological factors were investigated using a parental questionnaire. The three profiles were formed based on age- and gender-adjusted PMC and AMC z-scores. Multinomial logistic regression showed that OEs (n=81; p=0.04) tended to be younger than REs (n=306; p=0.04) and UEs (n=54; p=0.03). Parents of OEs reported more child health and developmental issues than parents of REs (p=0.03). Parents of UEs self-reported providing more support for physical activity than parents’ of REs (p=0.04). REs tended to live in denser population areas than UEs (n=54; p=0.03). Whilst PMC profiles revealed some socioecological differences, future research needs to focus on a broader range of potential correlates and untangle methodological analyses challenges to deepen the knowledge about PMC development in children.


Introduction
Perception of motor competence (PMC) reflects a child's own expectations and conviction of having competence in given motor tasks . Developmental theories (Harter, 1999;Stodden et al., 2008) and previous studies (Brian et al., 2018;Lopes et al., 2016) suggest that children less than eight years old do not yet compare themselves to others and tend to have inflated self-perceptions, including PMC. High PMC can motivate children to practice motor skills and be more physically active, as well as predict a healthier lifestyle and body composition (De Meester, Maes et al., 2016;Robinson et al., 2015;Stodden et al., 2008). Therefore, children's tendency to overestimate their actual motor competence (AMC) is considered to be a positive motivator for increasing physical activity (PA) (Stodden et al., 2008;Robinson et al., 2015). Due to the inflated PMC of children, the relationship between PMC and AMC is not expected to correlate (Lopes et al., 2016;Hall et al., 2019;Lopes et al., 2018;True et al., 2017;Barnett et al., 2021). Yet, not all young children have inflated PMC (Robinson, 2011). Therefore, it is crucial to study not only the PMC level but also the discrepancy between PMC and AMC in children (Weiss & Amorose, 2005).
PMC development is closely related to comparison to others and to self-esteem, thus, cognitive maturation is a key element in constructing (accurate) PMC (Harter, 1999;Harter & Pike 1984). According to Harter's competence motivation theory (1978), motivation increases when a child successfully masters a task; thus, if a child has plenty of opportunities to be physically active, it enables also the development of AMC. Stodden et al., (2008) stated that if a child does not have AMC, perceptions of competence will drop as the child gets older, and the cognitive maturation level allows him/her to evaluate his/her AMC more precisely (Goodway & Rudisill, 1997). Consequently, as a function of age, PA may also drop (True et al., 2017;Babic et al., 2014;Carcamo-Oyarzun et al., 2020). Interestingly, however, a recent systematic review and meta-analysis including 69 studies found that the strength of the relationship between AMC and PMC was not moderated by age or developmental status of the person (De Meester et al., 2020), contradicting the conceptual model of Stodden et al. (2008). Hence, more research on this matter is warranted.
One way of examining the accuracy between PMC and AMC and associated factors, is to study different profiles of PMC. Previous studies made with children's PMC profiles have used different statistical analyses to identify PMC profiles. A study by De Meester,  used cluster analysis and reported that a combination of high AMC and PMC was related to a higher levels of PA and lower weight status. In another study (Bardid et al., 2016) cluster analyses identified four AMC-based profiles; two groups were characterized by corresponding levels of AMC and PMC (i.e., low-low and high-high) and two groups were characterized by divergent levels of AMC and PMC (i.e., high-low and low-high). Children in the low-low and high-low group displayed significantly lower levels of autonomous motivation for sports and lower levels of global self-worth than children in the low-high and high-high group. Pesce et al., (2018) used regression techniques to identify three profiles of PMC; under-estimators (UEs), realistic estimators (REs), and over-estimators (OEs). According to Pesce et al., (2018) most girls underestimated their ball skills whereas most boys overestimated theirs. Also, OEs practiced a larger amount of sport than UEs. Finally, in a study by Duncan et al., (2018) children were classified as profiles of low, medium, or high PMC based on tertile analysis. In their study, boys tended to have higher AMC and PMC. In addition, they reported that children characterized as low PMC tended to also have lower AMC (Duncan et al., 2018). All of these aforementioned studies investigated differences between the profiles via individual-related correlates, such as PA, weight status, motivation, sport participation and gender differences. However, as past studies have managed to explain less factors associated with PMC (Crane et al., 2017) rather than AMC (Mota et al., 2020) in young children, we questioned if there is a need to broaden the study beyond the individual factors related to PMC profiles.
According to the socioecological model (Bronfenbrenner, 1994), behaviours can be influenced on many different levels, not solely on the individual level. In the socioecological model, family factors include interacting closely with the child and his/her individual factors (Bronfenbrenner, 1994). In early childhood, parents play a critical role -on one hand, as a behavioural example, and on the other hand, increasing or decreasing the amount of AMC and PA opportunities (e.g., equipment, surveillance, attitudes). Moreover, parental encouragement and the quality of instruction are important for motor development (Donnelly, et al., 2017;Gallahue & Donnelly 2003). One previous study has already shown that a family factor, such as parental educational level , can be associated with child's PMC. Moreover, according to the socioecological model, the importance of environment is related to the possibilities the child has to interact actively with the environment .  showed that less access to electronic devices (an environmental factor) was associated with better perception of locomotor skills in five to seven year olds. Nevertheless, so far, there is less evidence on family and environmental factors associated with PMC or PMC profiles as most available studies consider only individual factors. However, AMC studies have shown that family and environmental factors are important to consider in relation to a child's development Freitas et al., 2013;Iivonen & Sääkslahti, 2014) and therefore, future research should broaden the PMC investigation to consider these aspects. The first study aim was to explore the possibility of three PMC profiles in young children in relation to their AMC: UEs, REs and OEs profiles and secondly, to examine the differences of the profiles related to socioecological variables.

Materials and methods
The Ethics Committee of the University of Jyväskylä, Finland, granted ethical approval for the study. The parents of the participating children provided written assent. Children were informed about the study procedures and their right to opt out of participation without any consequences.
The study participants were part of the Skilled Kids study Laukkanen et al., 2018), which aimed to explore children's AMC, PMC and its covariates in a geographically representative sample of three to seven-yearold children from Finnish childcare centres. In this study, due to more validity evidence of the PMC for children who are over five years of age Venetsanou et al., 2018;, the sample of this study includes only those children who are 5-7 years of age. Cluster-random sampling was done according to the protocol of the international Health Behaviour in School-aged Children (HBSC) (World Health Organization, 2020). The childcare centres were chosen from the Finnish National Registry of Early Educators, which included in total 2600 childcare units. The number of childcare centres involved in a single region was weighted with the population density of the area. Altogether, 37 childcare centres participated, while 10 centres (27%) refused to participate due to lack of space (n = 1), interest (n = 4), time (n = 2) or low numbers of children (n = 3). If a randomly chosen childcare centre declined to participate, the next one on the list from the same area was recruited. In total, the majority of the children (78.5%) gave assent for study participation. Measurements were conducted in the childcare centre settings between November 2015 and September 2016 by two researchers (DN and AS) with two research assistants Laukkanen et al., 2018).
Throughout the data collection process, the protocols of several assessment tools were carefully followed . PMC was measured with the version of the pictorial scale of Perceived Movement Skill Competence (PMSC)  for young children that aligns with the Test of Gross Motor Development-Third Edition (TGMD-3; Ulrich, 2019). The PMSC test contains 13 items subdivided into locomotor skills (six skills, such as run, gallop, hop, skip, horizontal jump and slide, max. 24 points) and ball skills (seven skills, such as two-hand strike, onehand forehand strike, dribble, catch, kick, overhand throw and underhand throw, max. 28 points) Ulrich, 2019). During data collection, the test was done oneto-one with each child in a quiet room. For the PMSC administration, children were shown images of each skill (a good and poor performance) from the gender-specific PMSC booklet. First, the child was asked "Have you tried this skill before?" If the child responded "yes", (s)he was then asked to specify which of the two following picture options most resembled him/her. If a child had never tried the skill before, (s)he was asked to imagine how good (s)he would be in the given task. For example, two images showed a boy/a girl galloping. The child was asked "this child is pretty good at galloping, this child is not that good at galloping, which child is most like you?" If the child chose the more competent child, (s)he was asked to then choose between "really good" (4 points) or "pretty good" (3 points). If the child chose the less competent child, (s)he was asked to choose between "sort of good" (2 points) or "not that good" (1 point). The test per child took on average 10 minutes and it was done before the AMC measurements. In total, the maximum total score for PMSC was 52 points. The higher the child scored, the higher the PMC.
Internationally, the PMSC has demonstrated good reliability and validity for assessing young children's PMC in many countries and cultures Venetsanou et al., 2018;Diao et al. 2018). In the present sample, the test-retest reliability of the PMSC (conducted over 14 days) was tested with 53 children. The intraclass correlation (ICC) estimates and their 95% confident intervals were calculated based on a single measurement, absolute-agreement, 2-way mixed-effects model. The ICC of locomotor skills was 0.75 (95% CI = 0.60-0.85), in ball skills 0.82 (95% CI = 0.72-0.90) and in total PMSC 0.85 (95% CI = 0.76-0.91).
AMC was measured with the TGMD-3 (Ulrich, 2019) through live observation in different childcare centres. In the TGMD-3, each skill and its criteria were evaluated by an experienced observer. Two observers collected the data and both observers had passed Ulrich's (the test developer) official TGMD-3 reliability test performed via video-analysing and also had prior experience with live and video-recorded observation. Every skill had three to five skill criteria scored as either present (1) or absent (0). The AMC skills were aligned with PMSC skills, thus, divided into locomotor (six skills; run, gallop, hop, skip, horizontal jump and slide, max. 46 points) and ball skills (seven skills; two-hand strike, one-hand forehand strike, dribble, catch, kick, overhand throw and underhand throw, max. 54 points). Each skill was demonstrated once by the researcher before the practice trial of each child. The assessment trials were performed twice, as instructed in the manual. The TGMD-3 was completed in groups of three to four children. One group session took approximately 45 to 60 minutes. The TGMD-3 gross motor index (max. 100 points) was the sum of both subscales (locomotor and ball skills), and the raw test scores were used in the analysis. The test has good to excellent intrarater and interrater reliability (Ulrich, 2019) and is valid and reliable internationally (Cools et al., 2009) and nationally (Rintala et al., 2017). In this sample, to determine interrater reliability between the two observers, both coded the same performance for the 167 children. One observer evaluated the skills based on live-observation and the other evaluated the same skills through video-observation. The results were calculated based on a two-way random model of consistency for single measures. Interrater reliability between the observers for the TGMD-3 gross motor index was 0.88 (95% CI = 0.85-0.92) .
BMI was calculated based on weight (Seca 877) and height (Charder HM 200P) assessments as weight/height 2 (kg/m 2 ). To use age-appropriate BMI scores, BMI was later converted to BMI standard deviation scores (BMI SDS) using Finnish national BMI references, which were analysed based on data of Finnish children conducted between 1986 and 2008 (Saari et al., 2011). The raw BMI SDS were used in the analysis.
The parental questionnaire included the following elements of the socioecological model, individual-related factors (age, gender, BMI SDS, age of independent walking, time spent sedentary and outdoors, participation in organised sports, and child health issue) Laukkanen et al., 2018), family-related factors (respondent's gender, parents' mean educational level, respondents' PA frequency and sedentary behaviour, PA parenting, shared family PAs) and finally, environmental factors (electronic devices used and residential density), described in detail in Table 1. The parental questionnaires were based on three internationally well-known parental questionnaires (Cleland et al., 2011;Rodrigues et al., 2005;Telford et al., 2004) modified for the Skilled Kids study. They were administered to families through childcare centre staff. Moreover, for child health, parents were asked about additional health factors that could have influenced their children's development/PA. If the respondent answered "yes", a more specific description was required.
For family factors, the gender of the respondent and his/her partner were collected. Henceforth, they are called mothers and fathers. The term "respondent" refers to the parent who answered the questionnaire, so each respondent was one parent of choice not both parents. Respondents' sedentary behaviour was collected using the International Physical Activity Questionnaire's (IPAQ) short form, which has acceptable reliability and validity (Craig et al., 2003). Parents' educational level was classified into four categories and their percentage distribution within the sample is reported. Finally, respondents' PA frequency was collected via parental questionnaire by asking if the respondent exercises: Not at all (0 points), randomly a few times per month (1 point), approximately once per week (2 points), two to three times per week (3 points) or over four times per week (4 points) Laukkanen et al., 2018) (Table 1).
PA parenting questions (shown to be valid in Finnish children) (Laukkanen et al., 2018) refer to the sum and mean values of three types of practices: Parents' co-participation in PA, direct support of child's PA and encouragement of PA (Cleland et al., 2011). Additionally, shared family PAs were ascertained with the following question: "Please evaluate how often you engage in physical activities, such as cycling, walking, playing outdoors or indoors, hiking and playing games, together as a family so that at least one parent is actively involved." The environmental factors were a) the number of electronic devices used by the child and b) the residential density of the place where the child attended childcare. Residential density was evaluated indirectly using the postal code of the childcare centre attended by the child and the national population density registry for categorisation (Laukkanen et al., 2018;. The significance level was set at p < 0.05. From the Skilled Kids study participants, all those children who were over 59 months old (4.9 yrs.) and had PMSC and TGMD-3 gross motor results were included in the analysis. The children were categorized based on z-scores into the three PMC profiles (UE, RE and OE). First, we tested the z = 1 limit, as in previous studies (Pesce et al., 2018;Schmidt et al., 2013). However, as young children tend to have inflated PMC (Lopes et al., 2016;Stodden et al., 2008), the typically used z = 1 did not allow enough children to be categorized into the negative (UE) and positive (OE) profiles. Our goal was to identify the 10% of children with the highest and the lowest AMC and PMC scores, relative to their age. Those children who had lower PMC than AMC belonged to the UE profile, and those who had higher PMC than AMC belonged to the OE profile. Therefore, we modified the PMC and AMC z-scores as follows based on our data: low z ≤-1.5, middle z = −1.49 to 1.24 and high z = >1.25. Adjustments were made by calculating z-scores for each age group (5, 6 and 7 years) by gender. Subsequently, we unified the groups so that, regardless of age or gender, the three profiles were established. For example, a child with high PMC and low AMC was classified into the positive profile (OE), a child with middle PMC and high AMC was classified into the negative profile (UE) and a child with consistent evaluations was classified into the RE profile. To analyse gender differences between the profiles, a t-test was used.
To examine associations between the PMC-AMC ratio and individual-related, family-related and environmental factors, a multinomial logistic regression was used. The socioecological factors were added to the multinomial logistic regression model simultaneously, and the variables were excluded one by one, when the p-value was less than 0.05. Gender was retained in the final model (Table 4) regardless of its statistical insignificance, as it has been shown to be associated with PMC (Robinson, 2011;Pesce et al., 2018) and AMC Iivonen & Sääkslahti, 2014;Barnett et al., 2013). All other statistically insignificant factors were removed. The statistically insignificant factors removed were as follows: Respondents' sedentary behaviour, parents' mean educational level, respondents' PA frequency, child's time spent on sedentary activities, BMI SDS, child's time spent outdoors, child's age of independent walking, shared family PAs, participation in organised sport, and finally, electronic devices used. The model's goodness of fit showed that it was suitable for the data (Pearson χ2 = 812.70, df = 792, p = 0.30; Deviance χ2 = 656.05, df = 792, p = 1.00). The likelihood ratio tests for comparing the intercept only model and the model with covariates were significant (χ2 = 22.814, df = 10, p = 0.011), so the final model with covariates (gender, age, child health issue, parental support for PA and residential density) was significantly better than Acceptable reliability and validity (Venetsanou et al., 2018) PA parenting possible range 0-7 / 0 = never 0.5 = less than once/week 1.5 = 1-2 times/week 3.5 = 3-4 times/week 5.5 = 5-6 times/week 7 = daily PQ. PA parenting refers to the sum of parents' co-participation in PA, direct support of child's PA and encouragement for PA.
Acceptable reliability and validity (Gallahue & Donnelly, 2003) Shared family PAs possible range 1-5 / 0 = never 1 = less than once/week 2 = 1 to 2 times/week 3 = 3 to 4 times/week 4 = 5 to 6 times/week 5 = every day PQ. "Evaluate how often you engage in PA, such as cycling, walking, playing outdoors or indoors, hiking and playing games, together as a family so that at least one parent is actively involved". the model without any covariates. There was no evidence of multicollinearity between covariates (maximum VIF value = 1.03).
For the PMC-AMC ratio model, the Nagelgerke pseudo R-squared was 0.06. The overall correct classification percentage was 69.4%, indicating that it would be possible to re-classify 69.4% of the children to the profiles in which they were originally classified.
The descriptive data of the UE, RE and OE profiles (mean and SD) are reported in Table 3.
The children in the OEs profile (6.06 yrs.) were likely to be younger than children in REs (6.25 yrs.; p = 0.04) and children in UEs (6.25 yrs.; p = 0.03) profile. Parents of children in the OEs (14.8%) were more likely to report that their children had some health issue compared to REs (7.6%; p = 0.03). The four most common additional factors possibly influencing the development of these children (n = 41/441) were asthma (n = 9; 21.9%), attention deficit hyperactivity disorder (ADHD) (n = 6; 14.6%), verbal difficulties in producing or understanding speech (n = 5; 12.2%) and diabetes (n = 3; 7.3%). UEs received more parental support for PA compared to REs (p = 0.04) as 56.9% of the UEs and 46.2% of the REs were reported to have parental support for PA over three times a week. REs tended to live in denser population areas compared to UEs (p = 0.03) as 17.3% of the REs lived in a metropolitan area compared to 7.4% of the UEs. Also, 12.1% of the REs while 24.1% of the UEs lived in countryside (Table 4).

Discussion
The main purposes of the present study were to divide the sample of children into profiles of PMC in relation to AMC and to discover which factors relate to these profiles based on the socioecological model. To date this was the first study investigating children aged five to seven years by unifying the AMC and PMC scores and examining the possibility for profiles and their differences related to socioecological variables. This knowledge could, according to the socioecological model (Bronfenbrenner, 1994), explain children's PMC and AMC, in addition to target those factors that may be associated with PMC in interventions. Therefore, this study contributes to the growing field of PMC with its large sample size in addition to the use of socioecological model.
One of the main findings of the study was that three different PMC profiles could be identified, despite the children's young age. Although, a challenge in identifying the profiles was that the children's PMC was high -most probably due to young age of the participants. This led to an inability to use statistical methods that have been used previously, such as cluster analyses (Bardid et al., 2016;Weiss & Amorose, 2005), or regression analysis (Pesce et al., 2018). More specifically, the cluster analysis was not suitable due to the homogeneous data sample, which lead to profiles based on AMC scores leaving PMC values aside. The regression analysis was not executed as there was a peak in the maximum value of PMSC assessment tool leading to skewness which could not be logged successfully. Finally, Competence in young children. Statistically significant difference between girls and boys, the level of significance at * p < 0.05, **p < 0.01 ***p < 0.001 Table 3. Descriptive statistics of the three profiles of children with different combinations of perceived and actual motor competence (n = 441).  tertile analysis (Duncan et al., 2018) was not used as we aimed to use both PMC and AMC scores while identifying the profiles. Therefore, these results could be considered as preliminary.

Unit of analysis
The second aim of the study was to investigate if there were some socioecological factors which differentiated the three profiles. According to our hypothesis, children in the OE profile were the youngest while children in the RE profile were the oldest. Past studies suggest that children under eight years of age have inflated PMC (Stodden et al., 2008;Lopes et al., 2016), and, as a function of age, their PMC approximates their AMC more closely (Harter, 1999). This inflated PMC enhances children's willingness to participate in PA or motor tasks (Stodden et al., 2008;Robinson et al., 2015) and stems from young children's cognitive incapacity to make realistic estimations about actual skills (Harter, 1999). The current findings of children's PMC profiles related to age of the children attest these theories. However, a recent systematic review and meta-analysis stated that age is not a moderator of the relationship between AMC and PMC (De Meester et al., 2020). Interestingly, while previous studies have noted that, overall, PMC decreases with age, at the same time, it is important to bear in mind that this assumption was not adequately tested from a person-centred approach using longitudinal data, supported by AMC assessments. Hence, future research with this topic is warranted.
Another individual factor associated with differences between the PMC profiles was "child possessing a health issue". Children who estimated their AMC realistically were less likely to be reported by their parents as having factors influencing their development (n = 23/304, 7.6%), while nearly 15% (n = 12/81) of the children in the OE profile had some kind of a developmental or health issue. These findings should be considered as preliminary because the focus of the current study was not on health or risk factors; as such, the specificity and quantity of information on these factors is insufficient to make any further conclusions. Interestingly, however, it has been found that half of the children with ADHD have problems in motor development (Kaiser et al., 2015), and yet they are likely to overestimate their motor skills (Hoza et al., 2004). As ADHD was the second most frequently mentioned additional developmental factor by parents of OEs, future research should identify any special conditions that may influence young children's motor development and, especially, how they may relate to their PMC, AMC and PA. Also, in a study by Emck et al., (2009) children with emotional, behavioural and pervasive developmental disorders exhibited poorer AMC and unrealistic PMC, with certain indications of disorder-specific characteristics (Emck et al., 2009). As a result, they questioned if the type of developmental issue or disorder may be associated with children's MC and PMC development leading to unrealistic estimations (Emck et al., 2009). In the case of unrealistic evaluations, it has been also questioned if it is a way to protect one's selfimage from failure, a sort of self-defence (Pönkkö, 1999).
The present study also provided novel information on potential correlates at the family level related to differences between the PMC profiles. Children in the UE profile were reported to have more parental support for PA than their REs peers. Interestingly, when looking the profiles closely, it can be seen first of all, that UEs (adults and children), are reported to have less sedentary activities (respondents 5.11 hours daily, children 1.36 hours daily) compared to REs (respondents 5.39 hours daily, children 1.50 hours daily) and OEs (respondents 5.40 hours daily, child 1.48 hours daily) profiles, although these differences are not statistically significant. Also, parents of UEs self-reported more time in engaging shared family PAs. It may be that parents of the UEs engage more in light (physical) activities with their children, but children may not perceive this engagement as support for PA as it is not directly associated to sport-related hobbies, which only by attempting may prompt young children' higher PMC. In general (Korelitz & Garber 2016), discovered in their meta-analysis that parents' reports were more favourable than their children's reports about the parents' behaviours. Thus, it can be inferred that the parents in the current study also overestimated their positive feedback towards their children. However, it remains a question why this was seen only in the profile for UEs. Also, according to the study by Laukkanen et al., (2020) perceptions of high demandingness and high responsiveness in PA parenting, specifically parental expectations and facilitation of PA, were associated with satisfaction of competence need of the child. They stated that it may be possible to identify different types of PA parenting practices that are associated with children's motivation for PA. Therefore, future research focusing on coherence of parental support and children's perceptions of the amount and quality of such support is needed. Under estimation (UE), Realistic estimation (RE), Over estimation (OE), P = p-value, OR = Odds ratio, 95% CI = Confidence interval. * Children in the OE profile tended to be the youngest § Children in RE profile were likely to have less health issues reported by parents than children in OE profile # Children in the UE profile tended to receive more parental support for physical activity than children in RE profile ¤ Children in the RE profile were likely to live in denser areas than those children that belonged to UE profile Finally, one environmental factor was associated with differences between the three profiles of PMC as REs tended to live in denser population areas compared to UEs. A fifth (17.3%) of the REs lived in a metropolitan area compared to 7.4% of the UEs. It can be questioned if children from denser areas are more aware of their skill levels as they may have more possibilities to compare their own AMC level to other children' AMC, as they do not yet go to school at this age. Also, the childcare groups may be smaller in less dense areas of population. In this study, children in a metropolitan area participated in organized sport (56.8 mins/week) significantly more than those in the rest of the country (46.7 mins/week) . The role of participating in hobbies for the PMC of the young child may be important for children' self-belief. Previous studies investigating the correlates of PMC have found that higher sport participation  can be associated with higher PMC. In a study by Pesce et al., (2018) with respect to sport participation, the profile of OEs in locomotor skills practised a larger amount of sport than the UEs profile. Also, by engaging more in sportrelated hobbies, they may have had more feedback from external sources, such as coaches and peers, supporting their realistic perception. Therefore, these preliminary results suggest that residential density can be relevant when investigating PMC profiles.
Previous studies investigating the correlates of PMC profiles, have found that gender differences exist (Duncan et al., 2018;Pesce et al., 2018) in contrast to the current study. In a Duncan et al., (2018) study, boys obtained significantly higher scores than girls for both AMC and PMC scores. In a study by Pesce et al., (2018) most girls underestimated and most boys overestimated their actual ball skills. However, other studies investigating only correlates of PMC, excluding the PMC profiles, have stated that boys seem to have higher overall PMC compared to girls (Venetsanou et al., 2018;Slykerman et al., 2016), even though there are several studies reporting no gender differences (Lopes et al., 2016;Lopes et al., 2018;De Meester et al., 2020). Past studies have tried to explain these gender differences by reflecting on engagement and participation in motor skill activities (Robinson, 2011), as well as differences in content of play and hobbies between the genders (Barnett et al., 2013). Additionally, environmental or socio-cultural factors, such as equal possibilities and enhancement provided for PA (Iivonen & Sääkslahti, 2014;Robinson, 2011) may influence boys and girls motor development. Interestingly, however, a recent systematic review and meta-analysis investigating the strength of association between MC and PMC/physical self-perception, did not find that the relationship between AMC and PMC differed according to gender (De Meester et al., 2020). Authors questioned if this result may be due to methodological issues (such as divergent assessment tools) and recommended that future research should aim to overcome these methodological challenges (De Meester et al., 2020).
Also, previous studies have shown that a lower BMI is a positive correlate for PMC (Duncan et al., 2018;Spessato et al., 2013) as well as lower body weight (Jones et al., 2010;. In contrast, one study has shown that higher BMI  (rather than lower) was positively associated with PMC. As a potential explanation of this finding, Niemistö et al., (2019) stated that in their data sample, there was a lack of obese children, which made it hard to investigate these relationships between PMC and BMI accurately. Indeed, this finding contrasts with the results of the study by Spessato et al., (2013) who reported that the PMC of four to seven year old children was lower in obese children; even though their AMC was similar to children with a normal BMI. Meanwhile, a study of older children (9.50 ± 1.24 yrs) found that a combination of high AMC and PMC was related to higher PA and lower BMI . In the current study, no such differences within BMI SDS between the profiles were found. In general, it appears that, in contrast to our findings, PMC can be lower in obese children. Future research could investigate how perception relates to BMI as a function of age in a sample that reflects children of different BMI status.
Overall, the strengths of our study include (i) a geographically representative sample of children, which enables divergent residential densities to be considered within one country; (ii) a large number of children and families; and (iii) PMC and AMC measurements were matched to align. The major limitation of the study, in line with previous studies using a profile approach to PMC (Duncan et al., 2018;Pesce et al., 2018) is that the results cannot be generalized to other samples. Another sample may under-or over-represent particular profiles if the participants are categorized into the profiles based on the current overall AMC levels. Furthermore, the present study sample mainly consists of high socio-economic status families and has an under-representation of obese participants. This bias could explain why some individual-and family-related factors were the same across the profiles in this sample. Moreover, in the parental questionnaire, some of the test-retest reliability levels were slightly low with high CIs. Additionally, it would be useful for future research to collect more detailed information from parents of young children on their children's early development and possible health factors related to PMC. Lastly, as PA, AMC and PMC interact with each other (Robinson et al., 2015;Stodden et al., 2008), the lack of objectively measured PA levels in children can be seen as a study limitation.
In essence, the current study created three PMC profiles in young children. However, as young children have high PMC, future research should investigate if particular PMC profiles would offer more knowledge on profile differences in PMC and their associated factors. Furthermore, there were few differences between the PMC profiles based on certain socioecological factors. The study consolidated the research by showing that age is crucial element of PMC (profiles). Also, the results imply that a child's health related issues, as well as PA parenting and residential density of the living place may need to be considered while investigating the PMC profiles and their differences. To conclude, future research would benefit from longitudinal studies as well as a broader range of correlates relating to PMC profiles, as children's development occurs in a socioecological context and is driven by many influencing factors. and to Antti Saari for his help with the Finnish BMI SDS. The Finnish Ministry of Education and Culture funded this study (Grant/Award Numbers: OKM/ 48/626/2014 and OKM/59/626/2017).