A systematic review of quantitative studies concerning psychological aspects of early specialisation

Despite the intense and long-standing interest surrounding early sport specialisation, scholars still debate its nature and implications. Previous researchers have also identi ﬁ ed the need for further research relating to the psychological aspects of early specialisation such as lower quality motivation, dropout and burnout. To help guide future research it is important to build upon the quantitative literature concerning such psychological aspects of early specialisation. The speci ﬁ c aims of this paper are to provide an overview of research results of quantitative studies that set out to explore relationships between early specialisation and psychological aspects, and to critically examine the designs of such studies. As such, study design characteristics including participant demographics, the psychological aspects represented, and the research questions and results are explored. Data searches were conducted in PubMed, SportDiscus, and PsychINFO using search terms such as “ early sport speciali* ” . Twenty-one relevant papers met the inclusion criteria. The results highlight that the published papers in this area are broad in some respects (variety of sports, performance levels, and gender), but narrow in others (North American dominance, few psychological aspects explored, and few papers per psychological aspect). Many of the studies are based on cross-sectional and retrospective self-reports. Overall, this paper serves as a foundation on which to design future research studies in this area.

Participation in youth sport is associated with many positive outcomes for health and well-being (e.g., Coakley, 2021).However, outcomes for children who specialise early in one sport are more mixed, with both positive and negative results having been reported (Waldron, DeFreese, Register-Mihalik, et al., 2020).The International Society of Sport Psychology published a position stand stating that sampling several sports in a playful way is superior to early specialisation in terms of continued sport participation and elite CONTACT Charlotte Downing charlotte.downing@gih.seSupplemental data for this article can be accessed online at https://doi.org/10.1080/1612197X.2023.2251977.
performance (Côté et al., 2009).For example, Côté and colleagues (2009) highlighted that children who sample several sports are likely to experience more positive outcomes such as greater intrinsic motivation, reduced risk of dropout, and better social relationships, in comparison to those who specialise early.However, more recent literature reviews have highlighted inconsistent findings regarding the relationships between early specialisation and psychological aspects (e.g., Waldron, DeFreese, Register-Mihalik, et al., 2020).For example, some researchers report no relationship between early specialisation and psychological aspects such as dropout and burnout (Larson et al., 2019).Others highlight unfavourable outcomes of early specialisation, such as increased risk of dropout (Wall & Côté, 2007).The inconsistent findings regarding the reported relationships between early specialisation and psychological aspects serve as a key rationale for conducting this systematic review.
Moreover, in 2014, Côté and Vierimaa revisited the postulated outcomes of early specialisation and suggested that empirical support to underpin a negative relationship between early specialisation and motivation is still underdeveloped.However, they reiterate that the theoretical support underpinning such as a relationship (e.g., research related to the deliberate practice framework and positive youth development) remains strong.A decade later, there are still inconsistent results being reported regarding the theorised negative relationship between early specialisation and motivation (e.g., Downing et al., 2022;Pelletier & Lemoyne, 2020;Russell & Molina, 2018).Therefore, an updated evaluation of the empirical evidence underpinning the theorised relationship between early specialisation and motivation is warranted.
Recent publications have explicitly identified research exploring psychological aspects of early specialisation as a key area for further development (Kliethermes et al., 2021;Waldron, DeFreese, Register-Mihalik, et al., 2020).Against this backdrop, investigating the study designs of empirical studies concerning the psychological aspects of early specialisation is an important step in examining the literature base, and to guide future research in this area.Consequently, in this review, we focus on quantitative literature that has investigated early specialisation in relation to psychological aspects linked to subjective well-being (e.g., life satisfaction and positive/negative affect), as well as psychological and social well-being (e.g., burnout, motivation, anxiety).In doing so, we use the term "psychological aspects".This term was carefully chosen to reflect the various study designs represented within this paper.For instance, using the term psychological "outcomes" can incorrectly imply causality, and using the term psychological "correlates" does not fit non-correlation analyses such as t-tests.

Considerations regarding participant demographics throughout previous literature
In this section we will point out some issues with research into early specialisation that have been, or could be, addressed.One such issue is that sports have different traditions and assumptions which may alter the age at which specialisation typically occurs.Some sports can be considered "late sports"; for example, Noble and Chapman (2018) report that the peak performance age of African marathon runners is 27 years.Yet, in other sports it is possible to be selected into specialist training academies prior to age 12, such as in soccer (Clarke et al., 2018).Furthermore, gymnastics and figure skating are typically referred to as "early sports", where early specialisation has been described as advantageous, or even necessary, to prepare for high-performance demands during late adolescence (e.g., Côté et al., 2009;Kliethermes et al., 2021;LaPrade et al., 2016).Considering that the suggested outcomes of early specialisation differ between so-called "early" and "late" specialisation sports (e.g., Côté et al., 2009;LaPrade et al., 2016), it is valuable to investigate what sports are represented in previous quantitative literature.
Another issue that could be taken into consideration is that the popularity, and competitiveness, of particular sports fluctuates between countries.For example, floorball is popular in Sweden with over 118 000 players in 2020, and almost 33 000 of those being under 10 years old (Svensk Innebandy, 2020).There are 906 floorball clubs nationally and the possibility to compete year-round.Comparatively, floorball is extremely small in other countries, such as Brazil where there are only eight registered clubs, most of which do not offer training for younger children (Floorball Brazil, 2020).Thus, in some countries and sports it might be difficult, perhaps even impossible, to specialise early while popular and well-funded sports sometimes recruit young children into intensive training programmes in the hopes of identifying future talent.These country-related differences is another illustration of why it is important to look into the study designs of earlier research.
The role of gender in early specialisation is also an important consideration.A recent review found that females are underrepresented in all aspects of the talent development literature (Curran et al., 2019).This is problematic as results pertaining to male sports involvement are applied to female athletes without true understanding of their developmental needs (Curran et al., 2019).Therefore, it would be a notable strength if previous research regarding early specialisation comprised an even spread of male and female participants.
It is important to consider how these demographic factors might impact early specialisation.Overall, such considerations underpinned our decision to explore the extent to which athletes with a wide range of demographic characteristics (i.e., sports, countries, genders) are represented in quantitative research exploring the psychological aspects of early specialisation.

Study designs and research questions in previous literature
There are fundamental study design characteristics which indicate scientific quality for quantitative studies, such as clearly defined participant inclusion criteria and using reliable and valid measurement for variables (Joanna Briggs Institute, 2017).As highlighted by previous researchers, early specialisation is inconsistently defined and measured (e.g., DiSanti & Erickson, 2019;Mosher, Fraser-Thomas, et al., 2020, 2022).For this paper, we define early specialisation as training intensively and year-round in a single sport before age 12 (LaPrade et al., 2016).However, we recognise the ongoing discussion around how early specialisation is defined and measured.As such, the measurement of early specialisation in previous literature becomes a component of the results and discussion in this paper.
Over the last nine years, there have been several publications recommending future directions for investigating early specialisation in relation to psychological aspects (e.g., Côté & Vierimaa, 2014;Horn, 2015;Kliethermes et al., 2021).Such recommendations have called for more complex quantitative research designs (e.g., longitudinal designs, mediation analysis).For example, Kliethermes and colleagues (2021) explicitly state that important psychosocial mediators such as parental pressure "remain relatively unexplored" (p.138).By reviewing the study designs and research questions explored in previous quantitative literature, we can reflect on recent developments in the literature and provide valuable guidance for future research with regards to what types of questions have been more, or less, investigated.

Aim
The aims of this paper are to provide an overview of research results of quantitative studies that set out to explore relationships between early specialisation and psychological aspects, and to critically examine the designs of such studies.

Method
This systematic review follows the Preferred Reporting Items for Systemic review and Meta-Analysis (PRISMA: Moher et al., 2009;Page et al., 2021).Our study design includes several key components: (1) clear eligibility criteria outlined by the PICOS model (Liberati et al., 2009), (2) evaluation of scientific quality using the critical appraisal checklist for analytical cross-sectional studies (Joanna Briggs Institute, 2017), and (3) extraction and synthesis of data relevant to our study aims (e.g., reported results, number of participants, sport types, measurement tools used).

Eligibility criteria
Eligibility criteria are deliberately stringent to target a very specific area of research, and follow a modified version of the PICO principles known as PICOS (Liberati et al., 2009).Our categories for inclusion are thus concerned with participants, intervention (also called exposure or phenomena), comparison (between or within-participants design), outcomes (psychological aspects), and study design.Below we outline the eligibility criteria in relation to these five components.

Participants
Our participant requirements are broad; athletes of all ages, levels, gender, sport, and country are included.These broad inclusion criteria for participant demographics are an important part of investigating who is represented in the literature.

Phenomenon
For this systematic review, we are studying the phenomenon of early specialisation.DiSanti and Erickson (2019) highlighted that many papers do not provide a specific definition of early specialisation.Therefore, it is unsuitable to apply a stringent eligibility criterion based on definition and we included all papers which explicitly aimed to measure or capture early specialisation.Specifically, the papers had to outline the approach which they used to measure early specialisation or provide details of a proxy measure.Because the cut-off point for "early" specialisation is contentious, a specific age is not applied (e.g., < 12 years) and papers are excluded if they do not consider specialisation timing.Papers that captured specialisation age and used this as a between groups-variable (e.g., earlier/later specialisation categories) are included.
Deliberate practice (Ericsson et al., 1993) is commonly considered a key component of early specialisation (Mosher, Fraser-Thomas, et al., 2020).Deliberate practice involves intensive training that is specifically focused on skills acquisition that does not provide immediate reward (Ericsson et al., 1993).Only papers which explicitly stated that they captured deliberate practice as a proxy measure for early specialisation were included in this review.

Comparison
We included both between-group and within-group comparisons.Between-group comparisons for those categorised as early specialisers or non-early specialisers are seen as relevant if the comparisons are made for psychological aspects (e.g., are early specialisers different in motivation to non-early specialisers?).We also included papers which correlated indicators of early specialisation to psychological aspects (e.g., are markers of early specialisation related to dropout?).However, studies are only included if the independent variable (e.g., historical training volume) is specified as a proxy measure to capture early specialisation.

Outcomes
For this review, the outcome category of the PICOS model relates to psychological aspects, including motivation and burnout.While well-being is typically poorly and inconsistently defined in sport psychology literature, we have used the Lundqvist and colleagues (2011) paper on outlining conceptual considerations for well-being as a guide.Specifically, they consider wellbeing to include three aspects: subjective well-being (e.g., life satisfaction, sport-related affect), psychological well-being (e.g., self-acceptance, autonomy) and social well-being (e.g., social acceptance, social integration; Lundqvist, 2011).
We included dropout, although it can be conceptualised as both an outcome of motivational factors and a behavioural component.This was because dropout is included in several consensus statements from psychological organisations regarding early specialisation (e.g., Côté et al., 2009;Kliethermes et al., 2021).Due to our primary focus on wellbeing, we removed papers focused on psychological aspects such as decision making and sleep.While psychological aspects are often related to performance, we did not include performance alone as an outcome.

Study design
We only included empirical papers which collected and analysed quantitative primary data.Due to the nature of early specialisation, some aspects of secondary data are permitted, such as historical training logs or competition results.However, these had to be in addition to primary data.
Focusing on quantitative data reflects the notion that early specialisation research is largely quantitative (DiSanti & Erickson, 2019).Qualitative studies are excluded due to the different criteria required to assess quality across qualitative and quantitative studies.Furthermore, the typical research aims of qualitative papers (e.g., subjective experiences, self-reflections) do not match our aim of reviewing literature exploring the relationship between early specialisation and psychological aspects.Mixed-methods research was included, however, if the quantitative data component met the other eligibility criteria (participants, phenomenon, comparison, and outcomes).We had no further eligibility criteria for study design, enabling us to review a broad range of types of quantitative studies.

Information sources
The literature search was conducted in the electronic databases PsycINFO, SPORTDiscus and PubMed.Only published peer-reviewed papers written in English were screened.All sources were articles published between 1st January 1990 and 1st March 2022.However, the reference lists from all selected papers were manually searched by the first author, which included additional types of sources (e.g., book chapters, unpublished work), and studies published beyond our specified publication dates.

Search strategy
We searched for relevant quantitative research using the combination of terms "early speciali*" OR "diversification" AND "sport".We also searched using the key phrases "sport* speciali*", OR "early sport speciali*", OR "developmental model of sport participation".Boolean operators were used to narrow the search field to minimise the number of articles unrelated to sport science.Truncation was included to account for American and English spelling, as well as additional related words (e.g., specialisation, specialised).This search generated 1290 unique articles.Additional relevant records were identified from reference lists (n = 4).All unique articles were scanned for relevance, i.e., those which were peer-reviewed, written in English, and empirically explored psychological aspects of early specialisation using quantitative primary data.

Data management
References, including abstracts, were imported from the respective databases (e.g., PubMed) into EndNote where the screening and sorting process was conducted.As illustrated in Figure 1, the first author initially screened all article titles and journals and removed those that were clearly unrelated to sport science (n = 262; e.g., specialisation in language learning).The next step included screening abstracts where articles which were unrelated to sport specialisation were removed (e.g., diversification in the context of physical education, and diversification in terms of ethnic diversity in sport; n = 128).Next, papers were removed which did not collect empirical data (e.g., reviews, commentaries; n = 172), or were not related to psychological variables (e.g., injury prevalence, performance; n = 141).These initial screening steps were cross-checked by the second author.At this stage articles that were qualitative in design (n = 18) or focused on coach/parent perceptions rather than athlete experiences (n = 13), were removed.
The first author then removed papers that were non-empirical (n = 172) prior to the other two authors independently screening the remaining list of relevant abstracts (n = 245) and comparing results.The third author led discussions regarding any differences and disagreements concerning the included or excluded papers.The remaining articles (n = 73) were read in full.In this final step, articles which did not specifically measure early sport specialisation (n = 26) or did not have a psychological aspect as an outcome variable (n = 22) were removed.Furthermore, four papers which captured early specialisation and at least one psychological outcome variable were removed because their analyses did not empirically explore a relationship between the variables (n = 4; e.g., Bush et al., 2021).The remaining 21 articles were included.

Data extraction and synthesis
To address our research questions, relevant data were manually extracted from the 21 articles, including study aims, designs and methodological approaches, psychological aspects and their measurement, demographic characteristics, and reported results.Data were then compiled into descriptives, including frequencies and means, where appropriate.These data are presented in Tables 1-3.
To provide an overview of study design characteristics and assess the quality of articles, an adapted version of the Joanna Briggs Institute Critical Appraisal Checklist for Analytical and Cross-sectional Studies was used (Joanna Briggs Institute, 2017).The checklist includes evaluation markers for foundational study design characteristics such as clearly defined inclusion criteria, identification of confounding factors, and the appropriateness of statistical analyses.Item three of the checklist (Was the exposure [early specialisation] measured in a valid and reliable way?) was removed because there is no goldstandard measurement for early specialisation.We considered standard criteria for early specialisation measurement to involve direct or proxy measures for the key components of the early specialisation definition.That is, the measurement of early specialisation should include the attributes year-round training, intensive training, single-sport training, exclusion of other sports, and a specific age threshold of "roughly age 12" (LaPrade et al., 2016, p. 1).While the age threshold of prior to age 12 was not an inclusion criterion for eligibility in our review, we believe that using age 12 as the age threshold for "early" is valid on the basis of previous literature (e.g., Mosher, Fraser-Thomas, et al., 2020).

Results
Within this results section, we first present our results regarding how psychological aspects are explored in relation to early specialisation (research question 1; Table 1) and summarise the findings of this previous research (research question 2; Table 1).Thereafter, the demographic characteristics of those represented throughout the early specialisation literature is summarised (research question 3; Table 2).Finally, the study design characteristics are outlined (research question 4; Table 3).

Motivation and enjoyment
Within the identified literature (n = 21), studies exploring the relationship between early specialisation and motivation/enjoyment are the most prevalent (n = 11; see Table 1).The presence of multiple studies investigating similar research questions is a noted, albeit modest, strength of the literature.The use of theory when exploring the relationship between motivation and early specialisation differs between the studies.While some papers draw from one theoretical framework, others use several (e.g., self-determination theory; Deci & Ryan, 2000, and deliberate practice; Ericsson et al., 1993).Consequently, when quantifying the prevalence of particular theories within the literature, some studies are represented more than once.Eight studies (e.g., Downing et al., 2022;Russell, 2014), use self-determination theory for conceptualising and measuring athlete motivation.Ten studies use the deliberate practice framework (Ericsson et al., 1993) and/or the developmental model of sport participation (Côté et al., 2007) to underpin the theorised negative motivational outcomes of early specialisation (e.g., Lima et al., 2020).Deliberate practice is a theoretical concept of expertise development, rather than motivation.As such, deliberate practice in isolation perhaps provides a narrow perspective from which to view motivational correlates of early specialisation.
Even when motivation is underpinned by an established motivation theory, non-dominant questionnaires are often used.For example, Russell and colleagues (2014Russell and colleagues ( , 2015Russell and colleagues ( , 2017Russell and colleagues ( , 2018) ) collected data using versions of the Sports Motivation Scale (Pelletier et al., 1995).While this measure is grounded within self-determination theory, it is surpassed in reliability and validity by more recent measures, such as the Behavioural Regulation in Sport Questionnaire (Lonsdale et al., 2008) which is used in two articles within the present review (Downing et al., 2022;Waldron, DeFreese, Pietrosimone, et al., 2020).
As noted in Table 2 the identified studies report inconsistent results concerning early specialisation and motivation.For example, Waldron, DeFreese, Pietrosimone, and colleagues (2020) reported no between-group differences for early, late, and non-specialisers in terms of intrinsic motivation.In another study, Russell found that specialisers scored higher for introjected motivation (a controlled form of motivation associated with negative outcomes) than non-specialisers (Russell, 2014).However, in the study by Downing and colleagues (2022), participants who reported higher degrees of early specialisation reported lower controlled motivation.

Burnout
Burnout is empirically explored as a correlate of early specialisation in six of the identified studies.Four of these use the same theoretical underpinnings, including the same conceptualisation of burnout (Larson et al., 2019;Russell & Molina, 2018;Strachan et al., 2009;Waldron, DeFreese, Pietrosimone, et al., 2020).These four papers also use the Athlete Burnout Questionnaire (ABQ; Raedeke & Smith, 2001).Two of the more recent papers use study-specific measures of burnout (Meisel et al., 2022;Rugg et al., 2021).Notably, Meisel and colleagues' (2022) measure only captures one aspect of burnout; psychological exhaustion.Rugg and colleagues ( 2021) have yet another approach, whereby athletes who specialised early self-reported burnout as reason for dropout.
There are inconsistent results across the studies exploring burnout.For example, Larson and colleagues (2019) reported no relationship between burnout and any of the explored markers of early specialisation (e.g., age of first training camp, training volume between ages 6 -12).Similarly, Russell and Molina (2018) found no between-group differences for burnout between specialisers and non-specialisers, and Meisel and colleagues (2022) reported no between-group differences for early or late specialisers.The other three studies found that early specialisers reported higher burnout than late or non-specialisers (Rugg et al., 2021;Strachan et al., 2009;Waldron, DeFreese, Pietrosimone, et al., 2020).As such, across the six burnout studies, half report an association between early specialisation and burnout, and the other half do not.

Dropout
Within the limits of our inclusion criteria, five papers have empirically explored early specialisation and dropout, or attrition (Downing et al., 2022;Fraser-Thomas et al., 2008;Larson et al., 2019;Rugg et al., 2021;Wall & Côté, 2007).Four of these included parents, to some degree.Specifically, one study collected data exclusively from parents, who reported their child's training history (Wall & Côté, 2007).Another study collected data from both parents and athletes (Larson et al., 2019), and the remaining two studies collected data from athletes but captured aspects of parental influence (Downing et al., 2022;Fraser-Thomas et al., 2008).
A notable difference between the studies is how dropout was captured.Specifically, three of the studies (Fraser-Thomas et al., 2008;Rugg et al., 2021;Wall & Côté, 2007), recruited athletes who had dropped out prior to data collections, and one study captured dropout intentions among continuing athletes and dancers (Downing et al., 2022).Larson and colleagues (2019) asked parents to confirm if their child was still participating or had dropped out.This prospective study design includes data that was collected before the athletes withdrew from their training.
There are inconsistent results reported across the five studies in terms of early specialisation and dropout (see Table 2).Three papers report no relationship between early specialisation and dropout (Downing et al., 2022: Larson et al., 2019;Rugg et al., 2021) and two papers report that some, but not all, markers of early specialisation are related to dropout (Fraser-Thomas et al., 2008;Wall & Côté, 2007).For example, dropout athletes reported earlier onset for dry-land training (swimming) and off-ice training (hockey) than continuing athletes (Fraser-Thomas et al., 2008;Wall & Côté, 2007).Yet no differences between dropouts and continuers are reported for other variables such as age when commencing competition (Fraser-Thomas et al., 2008) and main sport start age (Wall & Côté, 2007).

Other psychological aspects
Several other psychological aspects are represented within the identified literature including perceived quality of life, mental toughness, mental health, anxiety, self-perceptions, stress, and resilience.As these variables are only empirically examined by one paper each, we do not present the theoretical underpinnings of these studies here.
The majority (8) of these papers found no relationship between early specialisation and psychological aspects.For example, Mosher, Baker, et al. (2020) reported no betweengroup differences for specialisers and samplers for anxiety, and McFadden and colleagues (McFadden et al., 2016) reported no differences between early or late specialisers regarding mental health or mental illness.However, one paper found that those who reported a higher degree of early specialisation also scored lower for attitudes towards physical activity, self-perceptions, and social support (Pelletier & Lemoyne, 2020).
Athletes represented a range of athletic levels (low level to elite), across approximately 24 different sports including soccer and swimming/diving (10 papers each); basketball (8); ice hockey, volleyball and athletics/track/cross country (7 each); tennis and gymnastics (6 each); baseball/softball and wrestling (5 each), American soccer and golf (4 each); cheerleading and dance (3 each); field hockey, lacrosse, martial arts, figure skating, handball, badminton, bowling, boxing, equestrian, and skiing (1 each).This is not a complete list, however: two studies recruited both specialised and non-specialised athletes, but sports are only specified for the specialised athletes (Patel & Jayanthi, 2018;Russell, 2014.Two other studies only reported the most prevalent sports in the sample and not a full list (Russell & Symonds, 2015;Waldron, DeFreese, Pietrosimone, et al., 2020).
Although some studies (n = 9) focused on specific sports such as swimming (Fraser-Thomas et al., 2008: Larson et al., 2019) and ice hockey (McFadden et al., 2016, most studies (n = 13) comprised several sports (e.g., Buhrow et al., 2017;Moesch et al., 2013).Large-scale cross-sectional studies incorporating several sports can provide information on general trends, yet the relative numbers of participants in some of those sports can be very small.While Buhrow and colleagues (2017) list the sports from which recruitment took place, they do not provide numbers of participants for each Representation of psychological aspects is generally diverse and represents several countries.For example, studies exploring the motivational correlates of early specialisation represented all five countries identified, even if studies from the USA are the most common (i.e., Brazil (1), Denmark (1), Sweden (1), Canada (2), USA ( 6)).Studies exploring the relationship between early specialisation and burnout are exclusively from North America (i.e., Canada (2), USA ( 4)).

Study designs and research questions assessment
In this section, the results regarding our assessment of study design fundamentals and the research questions explored throughout the previous literature.

Critical appraisal checklist
An adapted version of the critical appraisal checklist for analytical cross-sectional studies was used to assess fundamental study design characteristics (Joanna Briggs Institute, 2017).Overall, four of the 21 papers fulfilled all seven criteria on the JBI critical appraisal checklist (Downing et al., 2022;Larson et al., 2019;McFadden et al., 2016;Wall & Côté, 2007).
Item 1: Clear Participant Inclusion Criteria.Most of the studies ( 19) included clear participant inclusion criteria, with the exception of two papers where participants were recruited from university-based sport camps but justification for this sample is not provided (Russell & Molina, 2018;Russell et al., 2017).
Item 2: Detailed Description of Participants and Setting.Fourteen of the 21 papers described the participants and setting in detail.The remaining seven did not report key demographic information (e.g., not reporting gender or full list of sports).In the paper by Wall and Côté (2007), while the athletes are described in detail, the data were collected via interviews with parents.As the background information for the parents is described in less detail, this paper is considered "unclear".
Item 3: Objective and Standard Criteria Used to Measure Early Specialisation.Eleven of the 21 papers are considered to have used some objective and standard measurement for early specialisation.However, three studies categorised athletes who specialised after age 12 as early specialisers (Buhrow et al., 2017;Lima et al., 2020;Patel & Jayanthi, 2018).For seven papers, we found the validity and reliability regarding measurement of early specialisation unclear (Patel & Jayanthi, 2018;Russell, 2014;Russell & Limle, 2013;Russell & Molina, 2018;Russell & Symonds, 2015;Russell et al., 2017;Ryder et al., 2021).This is due to the way in which the characteristics of the specialised athletes are described.For example, in one paper athletes in the "specialised" group had a mean age of 15.84 and self-reported specialising around age 9 (M = 9.36; SD = 3.36; Russell et al., 2017).As such it is not possible to identify if, or to what extent, those within the "specialised" group could have specialised after age 12.
Items 5 and 6: Confounding Factors Identified and Controlled for.Thirteen of the 21 papers mentioned possible confounding factors, but only seven included strategies to control for these.For example, Russell and Molina (2018) only recruited female participants and specified this as a way to account for gender as a confounding factor.Due to the mixed-methods design of the paper by Patel and Jayanthi (2018), this paper is labelled as "unclear".Specifically, confounding factors are not statistically controlled for, but they are discussed in relation to the qualitative data.
Item 7: Reliable and Valid Measure for Psychological Aspect.The majority (17) of the papers used valid and reliable measures for the psychological aspects.These measures included validated questionnaires as well as behavioural indicators (i.e., recruiting those who had dropped out; Larson et al., 2019).Two of the papers used a mixture of validated questionnaires and study-specific measures (Downing et al., 2022;Pelletier & Lemoyne, 2020).For example, Downing and colleagues (2022) captured motivation and perceptions of parental involvement using standardised, validated questionnaires, but a studyspecific, non-validated measure was used to capture dropout intentions.The two remaining studies used only study-specific non-validated and single item measures for burnout (Meisel et al., 2022;Rugg et al., 2021).
Item 8: Appropriate Statistical Analysis Used.All the papers are considered to have appropriate statistical analyses.

Study design fundamentals
Nineteen studies used questionnaires to gather data, two studies included retrospective interviews to extract quantitative data (Fraser-Thomas et al., 2008;Wall & Côté, 2007), and one study included a semi-structured interview in addition to questionnaires (Patel & Jayanthi, 2018).Importantly, 20 of the studies employed first-generation research questions (i.e., exploring if x is related to y; Zanna & Fazio, 1982).Only one study used a secondgeneration research question by exploring perceptions of parental involvement as a moderator in the relationship between early specialisation and motivation (Downing et al., 2022).Importantly, all studies recruited participants via non-probability sampling.
While all papers used retrospective data collection methods to some extent, the depth of this retrospective data differed between studies.For instance, in the study by Buhrow and colleagues (2017) athletes simply self-reported the age at which they specialised.In contrast, Moesch and colleagues (2013) analysed self-reported training volume for every year of the athletes' main sport and used volume to explore the relationship between accumulated practice hours up to age 12 and subsequent outcome variables (e.g., volition).
While training is an important part of early specialisation history, most studies do not attempt to capture psychological aspects retrospectively (e.g., motivation as a young athlete, or over time).Instead, they have explored potential relationships between one or more retrospectively recalled variables (i.e., some aspect(s) of early specialisation) and one or more variables as they were experienced in the present (e.g., motivation at the time of the study).A notable exception is the study by Waldron and colleagues (2020) who captured burnout, motivation, stress and social support retrospectively using adapted prompts from validated questionnaires.While reliability indicators remained stable with the retrospective adaptations, full psychometric testing was not undertaken.
As displayed in Table 3, the conceptualisation and measurement of early specialisation differs greatly between the 21 articles.Specifically, 11 captured early specialisation dichotomously (i.e., separating the sample into early vs. non-early specialisers).This aligns with study designs exploring between-group differences.For example, Strachan and colleagues ( 2009) compared specialisers and samplers on various psychological aspects.Those who captured early specialisation, or components of early specialisation, as a continuous variable (e.g., Larson et al., 2019) had research questions related to varying degrees of early specialisation.For example, Moesch and colleagues (2013) correlated the continuous variable of childhood training volume with two motivation-based questionnaires.
In studies where vast amounts of data were collected concerning participants' training history, the individual training components are typically analysed separately.For example, Larson and colleagues (2019) collected data on several markers of early specialisation, including time in competition, age beginning the sport, age of commencing competition and whether training was year-round (≥8 months/year), yet these markers are all explored independently in relation to the outcome variables.In contrast, Downing and colleagues (2022) created a population-specific measurement tool to combine key markers of early specialisation into a continuous index representing degrees of early specialisation.Such differences in how data is analysed impact the overall conclusions that can be drawn.
Being able to answer research questions concerning early specialisation requires a measurement approach that captures early specialisation.Of the 11 papers that measured early specialisation dichotomously, various age thresholds have been applied with regards to what is considered "early" for specialisation.For example, Rugg and colleagues (2021) considered specialisation prior to age 15 to be "early", whereas Mosher, Baker et al. (2020) used 12 years as their age threshold for "early" specialisation.Lima and colleagues (2020) had another approach to capturing "early" specialisation whereby pubertal growth was used as an indicator of physical maturation rather than chronological age.

Discussion
The purpose of the present review is to provide an overview of research results of quantitative studies that set out to explore relationships between early specialisation and psychological aspects, and to critically examine the designs of such studies.
Findings related to psychological aspects of early sport specialisation Our results highlight a narrow scope in terms of theory-driven approaches to study psychological aspects related to early specialisation.While some well-supported theories are incorporated, such as self-determination theory, utilising other theoretical approaches could further elevate knowledge of early specialisation.For instance, the construct of motivational climates within achievement goal theory (Ames, 1992) may provide a basis for exploring how the training environment in which early specialisation takes place might impact present, and possibly future, motivation.
Despite suggestions that early specialisation is related to psychological aspects such as stress, perfectionism, and problematic identity development (e.g., ISSP: Côté et al., 2009;AOSSM: LaPrade et al., 2016), these aspects are sparsely (stress), or not at all (perfectionism and athletic identity), addressed in the empirical literature.As such, we agree with Kliethermes et al. (2021) that there is a need for further exploration of psychosocial aspects of early specialisation, such as self-confidence, self-esteem, depression, and anxiety.Athletic identity would be another particularly interesting area for future research as high-achieving athletes, which might include early specialising youth athletes, typically report a strong athletic identity (Lochbaum et al., 2022).
Inconsistent results regarding possible relationships between early specialisation and psychological aspects are identified in this systematic review, just as in previous reviews (e.g., Waldron, DeFreese, Register-Mihalik, et al., 2020).The present review builds upon the findings of previous reviews, however, by providing a unique perspective on the possible reasons underpinning the inconsistencies throughout the literature base (e.g., demographic considerations, research design).

Demographic considerations
Our results indicate that previous research does not capture the breadth of youth sport participant contexts due to unequal distributions regarding different sports, gender, ages, or countries; nor does it consider possible contextual differences between them (e.g., popularity of the sport within a specific country).For instance, some demographic characteristics are over-or under-represented within the previous literature.In some cases, key demographic characteristics are not comprehensively reported.Overall, we think it is appropriate to raise questions about whether the spread and variety of demographic characteristics is substantial enough to underpin to underpin generalised applied recommendations.As such, this is clearly an area in need of more nuanced research.
Variety and prevalence of specific sports One clear limitation of the current literature is the way in which sport populations are explored, namely in samples which are mixed in terms of both sports and performance level.While it may be suitable to group similar sports (e.g., ball sports), mixing categories such as aesthetic and non-aesthetic sports arguably leads to a heterogeneity that makes it harder to draw conclusions.
Importantly, "early sports" are included as part of mixed athlete samples and in some cases in very small numbers (e.g., just 9% of specialised participants are cheerleaders, gymnasts and dancers in Russell's, 2014 paper).This is problematic as differences concerning the outcomes of early specialisation between "early sports" and other sports has been widely suggested.For example, several position statements have stated that early specialisation may be advantageous in sports such as gymnastics and figure skating (Kliethermes et al., 2021;LaPrade et al., 2016).We argue that support for the notion of early specialisation being advantageous in aesthetic sports ought to include studies that either compared aesthetic and non-aesthetic sports, or recruited exclusively from within aesthetic sports.

Gender representation
Our results highlight that females are approximately equally as prevalent as males within papers included in this review, which is a welcome difference from sport talent development research generally which often underrepresents females (Curran et al., 2019).However, the identified missing data with reference to gender serves as an important message for future researchers to be diligent in reporting key demographic information about study participants.To be clear, we are not suggesting a rigid reporting of binary genders; rather, that researchers should provide sufficient description of participant gender (including non-binary genders where necessary/appropriate).

Age representation
The age of the sampled participants is also an important factor to consider in regard to the previous literature.For instance, Wiersma (2000) highlighted a potential trend towards specialisation becoming increasingly early and intensive.Some initial support for this statement can be seen in research examples such as Downing and colleagues (2022) who reported a positive relationship between early specialisation and age, and Rugg and colleagues (2021) who reported increasing prevalence of early specialisation among student athletes in recent decades.Therefore, it is possible that more recent studies on younger athletes will record higher prevalence, and/or a greater degree, of early specialisation.While the age ranges within some of the identified studies are narrow and young (e.g., 11-12 years; Mosher, Baher, et al., 2020), there are examples of older samples (e.g., 18-23 years in Waldron, DeFreese, Pietrosimone, et al., 2020) and wider spreads (e.g., 12-20 years in Downing et al., 2022).We suggest that future research should consider, and clearly justify, the age range of recruited participants.This may be particularly important for studies which recruit from multiple sports, where homogeneity is perhaps already compromised.
Another age-based consideration is the temporal subjectivity of how psychological aspects may change over time.Within the identified articles participants appear to represent similar age brackets, namely adolescence and early adulthood.Collecting data from younger participants, for whom specialisation is ongoing or more recent, is a logical advantage.However, retrospective reports from older or even dropout/retired athletes might provide insight into longer-term impacts of early specialisation.Research should continue to explore early specialisation from several perspectives across multiple age groups (i.e., current experiences as well as retrospective reflections).Additionally, future research would benefit from recruiting within a narrow age range in order to avoid multiple timebased perspectives within the same data, or perhaps using age as a covariate in analyses.
Importantly, the age threshold for "early" specialisation is most commonly stated as 12 years old (Côté et al., 2007;LaPrade et al., 2016;Mosher, Fraser-Thomas, Baker, et al., 2020).However, those who specialised after age 12 are routinely included as part of early specialised athlete groups in current literature.Future research should either use age 12 as the threshold for what is considered "early" specialisation or provide a clear rationale for using a different age threshold.

Country representation
It has been 10 years since Storm and colleagues (2012) suggested that early specialisation may be more culturally bound than we recognise, yet little research has explored this notion further.Indeed, there is a clear knowledge gap regarding the psychological aspects relating to early specialisation beyond Western cultures.Such a finding is congruent with reports of research participants being predominantly from Western, educated, industrialised, rich and democratic (WEIRD) societies (Henrich et al., 2010).Yet the fundamental structures and organisation of sports, and therefore specialisation trajectories, has the potential to be vastly different between countries.For example, Storm and colleagues (2012) highlighted how compulsory physical education in Denmark introduced a degree of sampling to all athletes that might differ substantially from the sports systems in, for example, the USA and Canada where replacing school-based physical activity with more main-sport training may be a more readily available option for school-aged children.Future research would benefit from further investigation of between-country differences regarding early specialisation.
As well as the structure of physical education and competitive sport within an area, the availability of training venues and access to high-quality coaches could also impact on specialisation pathways (Storm et al., 2012).Interestingly, Bell et al. (2016) found that young people in smaller schools in the USA are more likely to self-report being low-specialised or a multisport athlete than those in larger schools.This hints that perceptions, as well as differences of how specialised training is conducted, could differ based on an athlete's location.While no research has explored such phenomena, heightened competition among larger schools/clubs, or for those in more densely populated areas, may impact on whether early specialisation occurs or not, and what the psychological implications might be.

Study designs and research questions in previous research
The study designs and research questions used to explore psychological aspects of early specialisation are arguably simplistic.First generation research questions (i.e., is x related to y) perhaps feed the reductionistic nature of early specialisation research, because they are fundamentally built upon the assumption that early specialisation is similar for all participants, and that it results in either positive or negative outcomes.Exploring the whys underpinning any relationships between early specialisation and psychological aspects is an essential next step.
Researchers are often inferring causality based on cross-sectional research designs, or retrospectively collected training history data.While longitudinal research whereby data concerning early specialisation are collected from entry into sport until age 12 (and longer to research longer-term implications) would be beneficial, it is perhaps not feasible.However, it is certainly not impossible to collect longitudinal data relating to children's participation in sport over several years.For example, Thedin Jakobsson and colleagues (2018) recorded nine years of longitudinal data concerning sport participation relating to adherence and dropout in a study of 241 young people.

Limitations and future research directions
This review specified English language articles as a criterion and reported a high prevalence of publications from countries with English as an official language.There is undoubtedly relevant high-quality research in other languages, making this an important limitation of our review, and indeed most systematic reviews in sport psychology.Despite searching for articles published between 1January 1990 and 1March 2022, most of the articles in this review were published in the last five years (i.e., since 2017).This fastmoving pace of the literature may justify updated reviews to be conducted in line with future developments surrounding early specialisation.Future research, especially reviews which cover sociocultural aspects, may wish to include literature in other global languages.
Only papers which specifically targeted early specialisation are included within our analysis, yet we recognise that its conceptualisation is a challenge.Therefore, future researchers may wish to review research published prior to the widespread adoption of the term "early specialisation".Such a review may be better equipped to discuss the merging of literatures concerning early specialisation and deliberate practice, where there is extensive cross-referencing.We recognise that the deliberate practice literature, or perhaps research into other topics such as training volume or sport scheduling, can have relevance for the early specialisation discourse.However, mixing of terminologies and measurement approaches is adding to the inconsistency within the early specialisation literature.Once greater clarity and consistency is achieved, drawing upon multiple theories and terminologies can perhaps become more fruitful.
There is a notable increase in qualitative research in sport psychology (McGannon et al., 2021) and important qualitative research into early specialisation has recently emerged (e.g., Clarke et al., 2018;Morrice & Andronikos, 2021;Wall et al., 2020;Wixey et al., 2020).While quality checklists do exist for different types of research (e.g., JBI, 2020), these checklists do not always provide a good foundation for comparing research quality across the different types of research and we were therefore not able to include both qualitative and quantitative research into this review.
Finally, this paper does not include socioeconomic status as part of demographic characteristics.Socioeconomic status is not routinely reported in sport psychology literature but has the potential to impact early specialisation.In their discussion paper, Baker and colleagues (2020) specifically state socioeconomic status as a possible alternative explanation to some of the outcomes of early specialisation.As such, socioeconomic status, and other situational considerations such as the training environment, are important areas for future research.

Conclusions
This systematic review highlights inconsistencies regarding the reported relationship between early specialisation and various psychological aspects.Importantly, some psychological aspects have received very little, if any, research attention to date.Future studies that investigate psychological aspects such as stress, perfectionism, and athletic identity are particularly warranted.
Studies that empirically explore the psychological aspects of early specialisation largely rely on cross-sectional, retrospective study designs with non-probability sampling and it is possible that such design limitations underpin the inconsistent results found.Study designs and research questions should be further developed to ensure a more nuanced understanding of early specialisation.Particularly, we question whether cross-sectional research, regardless of whether some variables are retrospective, is suitable to discuss the causal relationships which are often assumed between early specialisation and psychological aspects.Future research should also pay greater attention to how demographic and cultural factors may influence the relationship between early specialisation and psychological aspects.
If researchers are interested in analysing the implications of early specialisation, longitudinal or prospective research is an essential next step to provide a stronger foundation upon which to discuss causal relationships.Of equal importance is examining the mechanisms behind those implications.Such research would benefit from qualitative inquiry to explore what aspects of early specialisation are perceived as negative, or indeed positive, by the athletes themselves.
The specific research questions are: (1) in what ways are psychological aspects explored in relation to early specialisation?(2)whatare the findings of this previous research?(3)whatdemographic characteristics are represented throughout the early specialisation literature (including sports, genders and countries represented)?and(4)what fundamental study design characteristics are present within the early specialisation literature (including research questions and methods)?

Table 1 .
Psychological aspects, psychological measures, and results.
Dropouts did fewer sports Dropouts did less unstructured swimming play Dropouts began dry land training earlier Dropouts attended first training camp earlier Dropouts reached "top in club" status earlier Dropouts took less time off competitive swimming No between-group differences for start age in competitive swimming Larson et al. (2019) Burnout Athlete Burnout Questionnaire DMSP No relationships and no between-group differences for burnout or dropout
DP, DMSP Dropouts started off-ice training earlier No between-group differences for non-hockey organised training volume, active play activities, hockey start age, organised hockey start age, skating lesson start age, deliberate play activities

Table 2 .
Demographic characteristics of participants.Mean age is given to 1dp in some instances where 2dp was not reported in the original publication.Percentage of female participants has been manually calculated in cases where percentages were not reported in the original publication.Mixed sports refers to samples comprising 5 or more sports.*Sports only reported for those who had specialised.**Sports only reported for the four most prevent sports in the sample, not a complete list.

Table 3 .
Measurement and conceptualisation of early specialisation.
*This study compared elite and near-elite athletes but specified that elite athletes began training in their main sport earlier (a key marker of early specialisation).