Skip to main content
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2026 Apr 10;44(3):724–739. doi: 10.1111/bjdp.70047

Children in Multicultural Malaysia Prefer Their Ingroup Over an Outgroup but Imitate Indiscriminately

Rachel Y Chin 1,2,✉, Daniel B M Haun 2, Pei Jun Woo 1, Frankie T K Fong 1,2,3,4,✉
PMCID: PMC13452000  PMID: 41960720

Abstract

This study scrutinised whether children's imitative tendency varied depending on the model's ethnicity in a multicultural nation, Malaysia. 123 Malaysian Chinese and Malay children aged four to six were shown how to complete two goal‐oriented, tool‐use tasks using either an inefficient but normative method or an efficient alternative. One‐third of the children watched a demonstration by an ingroup model (Malaysian Chinese/Malay), another one‐third watched a more‐familiar outgroup model (a Westerner/White individual) and the remaining watched a less‐familiar outgroup model (African). Children were also asked to rate their preference for each model and the models' knowledge of Malaysia. Children were equivalently likely to copy the inefficient tool use regardless of the model type, despite having a stronger preference and perceived model knowledge for their ingroup model. We also observed unexpected imitative patterns in a quarter of our sample. Our findings offered insights into the importance of considering cultural contexts when studying children's social learning.

Keywords: children, imitation, ingroup, Malaysia, multicultural, social learning


Statement of Contribution.

What is already known on this subject?

  • Young children display preferential copying of an ingroup over an outgroup model based on various social cues.

  • Such ingroup bias has also been established in the prosocial development literature using different experimental tasks.

What the present study adds?

  • We adapted a simple tool‐use paradigm to evaluate ingroup bias in Malaysian children.

  • Different measures may reveal different patterns of in‐group bias in preschool children.

  • Unexpected imitative patterns were observed in a quarter of our sample, suggesting that presentation mode may play a role in how children perceive the demonstration shown.

INTRODUCTION

By preschool age, children develop a strong propensity for faithful imitation – replicating modelled behaviours even when doing so leads to an inefficient approach (Nielsen, 2018). However, they do not simply imitate anyone indiscriminately but use social cues to guide who they imitate and whose information to follow (Meltzoff & Marshall, 2018). It is well established in the literature that children display preferential copying of an ingroup model over an outgroup model based on various social markers (e.g., linguistic cues, Andrási et al., 2024; Buttelmann et al., 2013; gender, Wang et al., 2021; also see Fong et al., 2023).

We are interested in whether such an ingroup bias is generalizable to a multicultural society and thus focused on Malaysia, where children grow up with constant exposure to multiple prominent cultural groups. As a multicultural society, Malaysia consists of three major ethnicities (Malay, Chinese and Indian) living in a shared community since European colonialism in the 16th century (Reddy & Selvanathan, 2020; Yew et al., 2021). As of 2023, there is a distribution of 70.1% Malay, 22.6% Chinese, 6.6% Indians and 0.7% of other numerous Indigenous groups and migrants (Statistics Malaysia, 2023a). Malaysian government's various past initiatives (e.g., Alliance Party to unite the three ethnics), together with the advocacy of ‘One Malaysia’ introduced in 2009 being deep‐rooted among contemporary Malaysians, have built mutual respect and acceptance for one another co‐living in a multicultural nation (Saad, 2012). This melting pot of culture allowed Malaysian children to be accustomed to daily exposure to other ethnicities in their living environment. Furthermore, Malaysian children have extensive first‐hand experience interacting and forming friendships with different ethnicities in kindergartens and schools, enriched by teachers with skills and resources for multicultural education (e.g., embracing the uniqueness of other cultures and languages; Yusof et al., 2014; advocating racial harmony; Omar & Dan, 2017).

With such rich exposure to multiculturalism, it is unclear whether Malaysian children will perceive normative information from ingroup members preferentially and thus display a cultural ingroup bias in the context of faithful imitation. To examine this question, we adapted a previously established paradigm (Fong, Imuta, et al., 2021a) wherein children were required to complete two tasks either using an inefficient method verbally labelled as the normative one (‘everybody uses’) or a more efficient method verbally labelled as the non‐normative one (‘nobody uses’). The experimental tasks were demonstrated by either an ingroup (Malay or Chinese), more‐familiar outgroup (a Westerner/White individual), or less‐familiar outgroup (African) model. The reason for such manipulation is due to the difference in levels of exposure for Malaysian children, as Malaysian children are more exposed to Westerners/White people compared to Africans. Studies found Malaysian Chinese children to recognise faces of their own race (Chinese) and ‘experienced other‐race’ (Westerner/White people) better than faces of ‘less experienced other‐race’ (African), presumed to be influenced by higher exposure to Chinese and Westerners/White people (Tan et al., 2012; Tham et al., 2017). Additionally, we included a parental cultural exposure report in this study and our findings also supported this. We hypothesised that children in the ingroup or more‐familiar outgroup condition would be more likely to select the socially preferred inefficient tool than the less‐familiar outgroup condition. Additionally, children were asked how much they liked each model and how much they thought each model knew about Malaysia. We predicted that children would prefer ingroup models and expect them to be more knowledgeable about Malaysia than both outgroup models.

METHOD

This study is pre‐registered. All study materials and de‐identified data can be found on the study's OSF page: https://osf.io/xzfcg/overview?view_only=26caf8c6b90c44dc87c83cb89fa3d783. All recruitment and procedural aspects of this study were approved by the Sunway University School of Psychology's Ethics Committee (Approval Code: 202309077 and 202309049). Kindergarten gatekeeper and parental consent were obtained before data collection. All the children received a small gift for their participation.

Participants

A total of 136 typically developing preschool children were recruited and tested via kindergartens located in Kuala Lumpur1 and Selangor,2 Malaysia. 13 participants were excluded due to an experimental session being interrupted by other children or teachers (n = 3), failing to understand and complete the tasks (n = 6) and loss of interest/unresponsiveness (n = 4). Our final sample comprised 123 (61 males, 62 females) 4‐ to 6‐year‐old children (M age = 5.07 years, SD = 0.78). 57 of them were identified by their parents as Malaysian Malay and 66 as Malaysian Chinese.

Two parents (1.6%) reported their highest level of education as early childhood education; one (0.8%) as primary education; 13 (10.6%) as secondary education; 10 (8.1%) as pre‐university education (diploma or certificate equivalent in other countries); and 92 (74.8%) reported having completed tertiary education. We also collected reports of household income using Malaysian income classifications: the B40 low‐income group (less than RM4,849), the M40 middle‐income group (between RM4,850 and RM10,959) and the T20 high‐income group (more than RM10,960). In our study, 22 households (17.9%) reported their income levels as the low‐income group; 45 households (36.6%) as the middle‐income group; 26 households (21.1%) as the high‐income group; and 30 parents (24.4%) preferred not to reveal their income levels. The majority of our participants were from the middle‐income group, while Malaysia's latest average household income in 2022 was revealed to be part of the high‐income group, with RM13,325 for Kuala Lumpur and RM12,233 for Selangor (Department of Statistics Malaysia, 2023). Taken together, children in this study were considered to come from families with a middle to high socio‐economic status.

Brief cultural exposure reports

To our knowledge, there is no existing data, publication, or measure on Malaysian children's exposure to other ethnicities. We thus included a brief measure to chart participating children's exposure to other ethic groups. This provides a better understanding and representation of the current sample's cultural exposure to their ingroups and outgroups as members of a multicultural society. We first asked about children's knowledge of and interaction with other ethnic groups (parents were allowed to choose multiple groups if applicable), ‘Does your child know or interact with any of the following groups?’. Malaysian Malay parents reported that more than half of their children (n = 39; 68.4%) had knowledge and interacted with other ethnic groups, with approximately one‐third (n = 18; 31.6%) who did not know or interact with any other ethnic groups. 30 Malaysian Malay participants (52.6%) had knowledge of and interaction with Malaysian Chinese, followed by 24 participants (42.1%) for Malaysian Indian, 12 participants (21.1%) for Western/White foreigners; and only four participants (7.0%) for non‐White/Western foreigners. Similarly, among Malaysian Chinese children, 49 participants (74.2%) knew and interacted with other ethnic groups, with only 17 participants (25.8%) who did not know or interact with other ethnic groups. 45 Malaysian Chinese participants (68.2%) had knowledge of and interaction with Malaysian Indian, 13 participants (19.7%) for Western/White foreigners; and 14 participants (21.1%) for non‐Western/White foreigners. We failed to collect Malaysian Chinese participants' exposure to Malaysian Malay due to experimental error. However, we assumed such a level of exposure to be similar to that of Malaysian Malay to Malaysian Chinese as participants were recruited from Kuala Lumpur and Selangor with similar levels of exposure to other ethnicities. See Figure 1 for visual plot.

FIGURE 1.

FIGURE 1

Children's knowledge and interaction with other ethnicities.

We also gauged children's exposure to foreign shows through parental reports, ‘Does your child watch any foreign shows or programs? If yes, can you name a few?’ Almost all parents (n = 118; 95.9%) reported that their children watch foreign shows. Some popular shows reported by the parents included Blippi, Diana and Roma, Paw Patrol and Cocomelon, which are commonly watched through online streaming platforms such as YouTube, Netflix and Disney Hotstar.

Experimental design and materials

Children were randomly assigned to three between‐subject experimental conditions with a balanced distribution of one‐third participants in each condition: ingroup (n = 42), more‐familiar outgroup (n = 40) and less‐familiar outgroup (n = 41), manipulated by the ethnicity of the model who demonstrated the experimental tasks. Children were shown how to complete two goal‐oriented, tool‐use tasks using either an efficient or an inefficient tool. The box task involved opening a box to retrieve a ball, and the pom task involved moving five pom‐pom balls from one container to another. The demonstrated inefficient methods were identified as the ones that ‘everybody uses’ (normative) and the efficient alternatives as the ones ‘nobody uses’ (non‐normative). Following previous research (Fusaro & Harris, 2008), the models used headshakes and nods to indicate which option was socially preferred.

Box task

The box task involved a ball placed inside a wooden box (20 cm × 18.3 cm × 10.5 cm) to be retrieved. Two tools were available to complete the task: (1) efficient tool, which is the long stick (29.5 cm in length), or (2) inefficient tool, which is the five short sticks with Velcro (6 cm in length each) to be assembled. See Figure 2.

FIGURE 2.

FIGURE 2

(a) Screenshot images of models. (b) Apparatus and key steps. (c) Step‐by‐step demonstration.

Pom task

The pom task involved five pom‐poms (diameter of approximately 2 cm) in one container to be moved to another identical container (7 cm in diameter; 6.5 cm in height). Two tools were available to complete the task: (1) an efficient tool, which is the big scoop (11.5 cm in length with a scoop of 4 cm in diameter) that allows transferring of all the pom‐poms at once, or (2) an inefficient tool, which is the small scoop (10.5 cm in length with a scoop of 2.5 cm in diameter) that only allows transferring of one pom‐pom at a time. See Figure 2.

Demonstration videos

The two experimental tasks were demonstrated using pre‐recorded videos presented on an iPad. We presented our models digitally on an iPad while normative verbal cues were provided in real life to control for potential bias towards models' accents. A total of eight videos were recorded with two videos from each model demonstrating the two tasks: one Malaysian Malay and one Malaysian Chinese model for the ingroup condition (for Malaysian Malay and Malaysian Chinese children, respectively); a Western/White model for the more‐familiar outgroup condition, and an African model for the less‐familiar outgroup condition. All models were middle‐aged females. See Figure 2 for images of models.

To control for the perceived attractiveness of the model (as a potential factor that may influence imitative learning), we asked 29 adults (M age = 22.24; SD age = 6.68) to rate using a five‐point Likert scale on ‘How attractive do you think this model is’. A total of 30 participants were originally recruited, with one excluded due to extreme scores. Our results confirmed that all models were perceived as equivalently attractive. Models' facial expressions, actions' pace and demonstration duration were consistent across all videos.

Box task

As illustrated in the images in Figure 2, the video began with the model displaying a neutral facial expression. The model opened the box with the long stick (efficient tool) and shook her head while the experimenter commented on the normativity of the behaviour during the experimental session. The model then put the long stick down and returned to the starting position with hands on the table. The model picked up one of the wooden fragments (inefficient tool) and nodded her head while the experimenter, once again, commented on the normativity of the behaviour during the experimental session. Subsequently, the model joined the five wooden fragments one by one to form a long tool and opened the box again. Finally, the model picked up the ball with a smile.

Pom task

As illustrated in the images in Figure 2, the video began with the model displaying a neutral facial expression. The model scooped all five pom‐poms with the big scoop (efficient tool) and shook her head while the experimenter commented on the normativity of the behaviour during the experimental session. The model then put the big scoop down and returned to the starting position with hands on the table. The model picked up the small scoop (inefficient tool) and nodded her head while the experimenter, once again, commented on the normativity of the behaviour during the experimental session. Subsequently, the model started transferring the pom‐poms one by one from one container to another using the small scoop. Finally, she returned to the starting position with a smile (6).3

Procedures

The experimenter sat across from the child in a quiet room while a second experimenter sat behind the experimenter to observe, code and assist.

Familiarisation phase

The experimenter started the session by familiarising the child with a Child‐Adapted Likert Scale for later use in the rating tasks: a five‐point Likert scale with the smallest circle indicating the least and the largest circle indicating the most (see Figure 3). The use of circles was due to existing literature revealing the more friendly alternative of the smiley face Likert scale to be ineffective for children to communicate judgements (Hall et al., 2016). We used items/options common to children of our sample (e.g., chilli, doctor) to practice rating how much they liked something and how much someone knew about something. We started with a demonstration by the experimenter before the child's attempt.

FIGURE 3.

FIGURE 3

Child‐Adapted Likert Scale.

The experimenter first explained how to use the circles to indicate their preference. The experimenter briefed, ‘If you like the item, point to the largest circle. If you do not like the item, point to the smallest circle. If you like the item a little bit or do not like it a little bit, you can point to one of these middle circles’. Subsequently, the experimenter showed flashcards of three examples (i.e., dolphin, lemon and apple). The experimenter first asked the child to identify the item, ‘What is this?’, and then provided an example by expressing how much they liked each item using verbal cues and non‐verbal cues (pointing to the respective circle), ‘Yes! It is a dolphin. I like dolphins a lot, so I point to the largest circle’. This was repeated for the other two examples; lemon as the unpreferred choice was expressed as ‘I don't like lemons, so I point to the smallest circle.’; apple as a less preferred choice compared to dolphins were expressed as ‘I like apples, but not as much as I like dolphins. So, I point to the middle circle.’

To practice further, the experimenter presented flashcards of two additional items, ice cream and chilli, to ask how much the child liked the items. Similar to the explanation phase, the child was first asked to identify the items, then asked, ‘How much do you like ice cream/chilli? Can you tell me by pointing to the circles?’ The experimenter presented two more flashcards of two human characters, a doctor and a chef. This time, they were asked, ‘How much do you think a doctor/chef knows about cooking? Can you tell me by pointing to the circles?’ See https://osf.io/xzfcg/overview?view_only=26caf8c6b90c44dc87c83cb89fa3d783 for the stimuli and full procedure.

Experimental phase

After familiarisation, the experimenter began with the experimental tasks: ‘Okay, now we are going to play a game’. Both apparatus and iPad were kept next to the experimenter, out of sight and reach of the child. The experimenter introduced one of the apparatuses depending on the counterbalancing: ‘Look here. We are going to take something out of the box (for box task)/We are going to move the pom‐poms from this container to another (for pom task). Let's watch a video before you try’. Subsequently, the experimenter moved the apparatus away and brought out the iPad for video demonstration. Their respective videos were played depending on the experimental condition (ingroup, more‐familiar outgroup, or less‐familiar outgroup). The experimenter played the video and narrated: ‘Let's watch the video. This person is from Malaysia/Florovia/Delmaria’. Malaysia was used for the ingroup condition, Florovia for the more‐familiar outgroup condition, and Delmaria for the less‐familiar outgroup condition. Malaysia, Florovia and Delmaria were interchanged depending on the condition. Florovia and Delmaria were made‐up names that children were unfamiliar with in order to manipulate the outgroupness and control for children's prior knowledge/experience (with existing foreign countries) as a potential confounding factor.

The experimenter explicitly commented on the normative nature of each tool option during the demonstration. Specifically, while the model used the efficient task and shook their head, the experimenter made the non‐normative narration: ‘You can use this, but nobody in Malaysia/Florovia/Delmaria uses this’; while the model picked up the inefficient task and nodded their head, the experimenter made the normative narration: ‘instead, everybody in Malaysia/Florovia/Delmaria uses this’. The reason for specifying the models' country is to explicitly prescribe the norm to be group‐specific. The demonstration was repeated once, ‘Let's watch the video one more time’.

After watching the demonstration videos, the experimenter offered the child a turn to complete the same task: ‘Remember, there is no right or wrong answer. You can do whatever you like. Your turn! Ready? 1, 2, 3 … go!’ The same procedure was repeated for the alternative second task.

Model rating phase

After completing the two tasks, the experimenter smoothly transitioned to the final task: ‘Almost done! I just have a few more questions to ask’. The experimenter briefed participants again on how to use the Child‐Adapted Likert Scale: ‘Remember, the largest circle means a lot. The smallest circle means not at all. The circles in the middle mean a little bit’. After clarifying that children were clear with the instructions (by children's simple verbal affirmation or non‐verbal head nodding), the experimenter asked children about their preference for all three models (ingroup, familiar outgroup and unfamiliar outgroup) and how much they thought each model knew about Malaysia. The child provided their answers by pointing to the circles on the Child‐Adapted Likert Scale. The presentation order of the model was dependent on counterbalancing. The experimenter presented a flashcard of the first model and asked, ‘How much do you like her?’ After receiving the response, the experimenter asked the second question, ‘How much do you think she knows about Malaysia?’ The experimenter then showed the child the rest of the models, one after another and asked the same two questions. Finally, the experimenter gifted each participant a toothbrush as a token of appreciation before they left.

Coding

The use of the efficient tool was coded as ‘0’, and the inefficient tool was coded as ‘1’. Additionally, we discovered a portion of our participants imitated the exact actions demonstrated by the model with some even imitating nodding and shaking their heads, which should only serve as non‐verbal communication of social norms. We referred to this as high‐fidelity copying and categorised these participants into three groups based on their imitative fidelity. Specifically, those who used both tools but did not re‐enact any non‐verbal cues were coded as ‘1’; those who also copied either one of the non‐verbal cues (either nodding or shaking head) were coded as ‘2’; and children who used both tools and re‐enacted both non‐verbal cues were coded as ‘3’.

The Child‐Adapted Likert Scale was coded with 1 to 5 (1 representing the least and 5 representing the most). All available recordings (N = 126) were coded by a second coder with 100% reliability.

RESULTS

Unexpectedly, approximately one‐quarter of our participants copied even the non‐verbal cues from the models in the video (nodding and shaking head) instead of making an active decision to select one tool to complete the task. In this paper, we label this form of blanket copying ‘high‐fidelity imitation’ (see Descriptives subsection below for more details). Considering these participants may have perceived the purpose of the task differently, these participants (n = 31) were excluded from the main analysis of the imitation data and only the remaining 92 children were included.

Imitation rates

Binomial tests revealed that children's replication of the normative, inefficient tool was above chance in both box and pom tasks across all three experimental conditions (ps < .001).

Comparison between conditions was analysed using a Generalised Linear Mixed Model (GLMM) with a binomial distribution and a logit link function, treating the outcome variable (tool choice) as a dichotomous variable. A random participant intercept was included as a random effect to account for within‐subject correlations on the two experimental tasks. An initial null model with participants' ethnicity, gender and age entered as fixed effects revealed no effect of any of these predictors. In a second model, experimental conditions (group membership of models) and task, alongside participants' ethnicity, gender and age were entered as fixed effects. Model comparisons revealed no significant difference between the null model and full model, Χ 2(3) = 0.812, p = .847, indicating that children's imitative tendency did not vary across the three experimental conditions (see Figure 4).

FIGURE 4.

FIGURE 4

Proportion of normative tool use across conditions and tasks. Error bars indicate standard errors.

Model ratings

Model ratings were analysed using repeated measures ANOVAs. Greenhouse–Geisser corrections were applied to any violations of sphericity (tested using Mauchly's Test). The main effects were followed up with Bonferroni‐corrected post‐hoc tests.

Children's preference towards models was analysed using a 2 (Participants' ethnicity: Malaysian Malay and Malaysian Chinese) by 3 (Condition: ingroup, more‐familiar outgroup, less‐familiar outgroup) mixed‐design model. The main effect of participants' ethnicity, F(1, 110) = 11.80, p = <.001, ηp2 = .09, revealed that Malaysian Malay rated preference for each model significantly higher (M = 3.55; SE = 0.14) than Malaysian Chinese (M = 2.93; SE = 0.13). There was also a main effect for condition, F(2, 220) = 23.10, p = <.001, ηp2 = .163, with a large effect size. Mean scores revealed that preference towards the ingroup model (M = 3.90, SE = 0.13) was rated significantly higher than either the more‐familiar outgroup (M = 2.75, SE = 0.14) or less‐familiar outgroup model (M = 3.06, SE = 0.15), p < .001. However, no interaction effect was reported between participants' ethnicity and condition, F(2, 220) = 0.12, p = .888, indicating both Malaysian Malay and Malaysian Chinese rated preference towards all three models similarly (see Figure 5a).

FIGURE 5.

FIGURE 5

Mean ratings for Malaysian Malay and Chinese children in both rating tasks. Error bars indicate standard errors. Dots indicate observed data.

Children's perceived model's knowledge about Malaysia was also analysed similarly. There was no main effect of participants' ethnicity, F(1, 109) = 0.33, p = .570, indicating both Malaysian Malay and Malaysian Chinese rated the model's knowledge about Malaysia similarly. There was a main effect for condition, F(2, 218) = 7.38, p = <.001, ηp2 = .05, with a small to medium effect size. Mean scores revealed that the ingroup model (M = 3.84, SE = 0.13) was perceived to know Malaysia significantly more than either the more‐familiar outgroup (M = 3.29, SE = 0.14) or less‐familiar outgroup model (M = 3.36, SE = 0.14). However, no interaction effects were reported between participants' ethnicity and condition, F(2, 218) = 1.77, p = .172, indicating both Malaysian Malay and Malaysian Chinese rated knowledge of all three models similarly (see Figure 5b).

Descriptives: different types of high‐fidelity imitators

Of the 31 ‘high‐fidelity imitators’ (who were excluded from the main imitation analysis), about 20–25% of Chinese children displayed the highest level of fidelity across conditions, but none of the Malay children did so. The majority of Chinese children tended to include one non‐verbal cue (either nodding or shaking head) across all conditions (75% in ingroup; 50% in more‐familiar outgroup; 57% in less‐familiar outgroup). Whereas only approximately 27% of Malay children did so in the ingroup condition and 48% in the less‐familiar outgroup condition (the same proportion was observed for the more‐familiar outgroup condition). Taken together, Malaysian Chinese children were generally showing a stronger tendency for ‘blanket copying’ compared to Malay children, especially in the ingroup condition. Interestingly, Malay children's imitative fidelity in the more‐familiar outgroup condition was seemingly lower than either the ingroup or the less‐familiar outgroup condition (See Figure 6).

FIGURE 6.

FIGURE 6

Proportion of three sub‐types of high‐fidelity imitations across condition and participant ethnicity.

DISCUSSION

We adapted a previous paradigm (Fong et al., 2021b) to evaluate if Malaysian children were more likely to faithfully imitate an ingroup than an outgroup model. In our study, Malaysian Chinese and Malaysian Malay preschool children watched either a local ethnic ingroup model (Chinese/Malay), a more‐familiar outgroup model (a Westerner/White individual), or a less‐familiar outgroup model (African) demonstrating how to complete two simple goal‐oriented instrumental tasks using either an efficient but non‐normative tool or an inefficient but normative tool. Regardless of whether the model was an ingroup or outgroup member, for both Malaysian Malay and Malaysian Chinese children, the majority of children copied the inefficient, normative options to complete both tasks.

We propose that the reduced ingroup bias effect may be due to children's proclivity to follow norms (even when endorsed by an outgroup member) due to their customary beliefs living in a multicultural and collectivistic society. As indicated by our parental reports and in light of current literature, contemporary Malaysian children have been incorporating themselves into a multicultural environment that strongly advocates unity among ethnicities (Yusof et al., 2014). This allowed all Malaysian children to share common norms (e.g., only using the right hand to give and receive; impolite to point at others with the index finger), despite their differences in ethnicity, dietary habits, traditional customs and religious beliefs in daily life (Ahmad & Yusof, 2010). Children are encouraged to adhere not only to norms specific to their ethnic ingroup but also to norms of a bigger, shared group – the national identity. Additionally, Malaysian children were shown to be widely exposed to other ethnicities and media. Such exposure in a multicultural society may influence cultural nuances that differ from monocultural societies. We note that our brief cultural exposure parental report did not capture other forms of exposure to outgroups, such as exposure via parents' activities (e.g., parents' exposure to media with the presence of their children), parents' interpersonal relationships with outgroups and parents' orientation to other cultural practice which can be fairly different between parents who received their tertiary education abroad versus locally. Future investigations should consider including passive forms of children's exposure to other cultural groups via their parents' daily life and background. Alternatively, likewise in many other Asian countries, most Malaysian parents tend to adopt an authoritarian parenting style (Keshavarz & Baharudin, 2009; Masiran, 2022), involving high levels of discipline with less regard to providing warmth and acceptance for children (Bi et al., 2018). Mofrad and Uba (2014) reported authoritarian parenting to be the most common parenting style in the city center of Malaysia, with Malay parents being the most demanding and Chinese parents being the most unresponsive (unsupportive of child's individuality and endeavours; Baumrind, 1978). Since such parental style is usually associated with high levels of conformity and obedience towards adults' instructions (Hellmer et al., 2021). In line with this premise, another potential explanation for our finding is that children might have treated their own imitative behaviour as a form of obedience. Some hints from research about children's authority bias learning suggest that children are inclined towards learning from a dominant figure or a figure with higher status (Danovitch & Shenouda, 2018; Lucas et al., 2017; Ma et al., 2025; McGuigan, 2013), presumably because they are perceived as a credible source of knowledge (Fong et al., 2023). Even so in Malaysia – a society where obedience is highly valued (Mu et al., 2024), replicating normative behaviours, regardless of whom they have learned from, may be treated as meeting an authority figure's expectations to be a good student or child. This form of authority‐driven learning may also be observed in peer‐to‐peer learning, where more capable peers may be viewed as an authority figure with relevant expertise as well. For example, Barr and Hayne (2003) found that older siblings were influential in younger siblings' imitative learning. Children may treat older‐siblings, more capable peers and adults as reliable ‘cultural experts’ to learn from (Fong et al., 2023; for a recent review about cultural learning via peer interactions, see Lew‐Levy & Amir, 2024). This kind of motivation for imitation remains understudied (Fong & Haun, 2025).

We did not find an ingroup bias effect in children's imitative responses, however, we replicated an ingroup bias in the rating tasks (Wilks et al., 2019). First, both Malay and Chinese children indicated that the ingroup model was more knowledgeable about Malaysia than either of the outgroup models. This was unsurprising and confirmed our manipulation of group‐specific knowledge status about the models. When asked how much they like each model, children's rating for the ingroup model was higher than either of the outgroup models. Interestingly, this ingroup bias was stronger in Malay than Chinese children. Potentially, this may be due to religious or cultural teachings unique to our Muslim Malay sample to express positive feelings more strongly, however this needs further validation.

Such mismatch in ‘ingroup bias’ patterns across tasks highlight the value of employing multiple theoretically related measures within a study, where findings may complement each other but also reveal more about the multidimensionality of social learning in children. The imitation task captured children's tendency for norm compliance, and the liking rating task measured how much a child liked a model (higher rating for ingroup model compared to either outgroup model indicated a preference towards ingroup model). It is possible that thinking about norm compliance behaviours primes children to identify with a common identity (at a societal level; common in‐group identity model, see Dovidio et al., 2007), whereas evaluating how much an individual likes an ingroup vs. outgroup model may prompt them to think more about their individual social identity (at a personal level). This dynamic form of social identification may be more apparent in multicultural societies as individuals switch between social identities flexibly depending on social contexts.

Focusing on the procedural aspect of our study, one may argue that having an ingroup experimenter to comment on the demonstration videos (providing narration of the normative cues to control for bias towards models' accents) may have led children to interpret the use of an inefficient but normative tool to be one that is agreed by their ingroup, inducing a normative response in all conditions. We think this is unlikely because the experimenter's narration included clear reference to the countries of all models (‘nobody/everybody in Malaysia, Florovia and Delmaria uses this’), making the narration culturally specific. The inclusion of the models' countries should be sufficient to highlight that the actions are not a norm of an ingroup but of either an ingroup, familiar outgroup, or less‐familiar outgroup.

Surprisingly, approximately one‐quarter of our sample re‐enacted the model's demonstration by using both tool options to complete the tasks (in the same order), with more than half even including the models' non‐verbal cues. This might be due to the experimenter's instructions not being explicit enough that the participants should choose only one tool to complete the task, however, their inclusion of non‐verbal cues was noteworthy. Thus, this proportion of children might have perceived the purpose of the task or learning context differently compared to the majority. Thus far, this has not been observed nor reported in extant studies using similar paradigms and procedures (Corriveau et al., 2017; DiYanni et al., 2015; Fong et al., 2021a), including those conducted in East Malaysia (Fong et al., 2021). We thus speculate that such form of ‘blanket copying’ may be a kind of digital screen effect – when contemporary children may now process information conveyed on screens as socially credible and copy all behaviours seen with high‐fidelity given the rapid change in digital landscapes in this digital era (first proposed based on findings using a very similar paradigm, Fong et al., 2021b; also see Nielsen et al., 2021). Almost all Malaysian children are exposed to TV shows and interactive devices with a broad range of digital media including livestreaming media such as YouTube, Instagram and Tiktok, which rapidly have become a vital part of their daily learning of the social world (Baboo, 2013). It is also worth noting that we observed a difference in imitative tendency between our samples, with Chinese children generally showing stronger tendency than Malay children. This difference may be a reflection of cultural nuances in children's social learning approaches guided by different culturally appropriate socialisation priorities between Malay and Chinese families in Malaysia (Masiran, 2022). We did not have enough power to conduct a statistical analysis, however the descriptive pattern charted is clear and warrants further follow‐up investigation.

The generalizability of the current findings should be considered cautiously. It is unclear whether we will replicate the same pattern of results if the outgroup models are local Malaysians (e.g., showing Malay children a demonstration from an Indian or Chinese model), as this study only tested the effect with foreign outgroup models. Similarly, we are also uncertain if we would replicate this pattern in a minimal group paradigm (where children are assigned to arbitrary groups typically using colours as group markers), as recent research suggests that children perceive minimal groups similarly to naturalistic groups (Yang et al., 2022). Further, future investigations could also employ other ways of evaluating ingroup bias, such as a resource distribution task commonly used in prosociality and racial bias research (Spence & Imuta, 2020).

This study addresses recent calls to extend developmental psychology research to include under‐represented or understudied samples (Draper et al., 2022; Nielsen et al., 2017) and include researchers from under‐represented populations (Chuan‐Peng et al., 2025; Singh, 2024). It raises questions about result generalizability, but more importantly the role cultural and environmental factors may play in children's socio‐cognitive development (Fong et al., 2022). Ongoing investigation into the commonalities and differences in children's social learning approaches between monocultural and multicultural ecologies promises to shed light on children's capacities for adaptive social learning across diverse cultures.

AUTHOR CONTRIBUTIONS

Rachel Y. Chin: Conceptualization; data curation; project administration; formal analysis; writing – original draft; writing – review and editing; investigation; visualization. Daniel B. M. Haun: Conceptualization; writing – review and editing; resources; funding acquisition. Pei Jun Woo: Conceptualization; methodology; writing – review and editing; supervision; resources; funding acquisition. Frankie T. K. Fong: Writing – original draft; writing – review and editing; conceptualization; methodology; funding acquisition; supervision; formal analysis; visualization.

CONFLICT OF INTEREST STATEMENT

All authors declare no conflict of interest.

ACKNOWLEDGEMENTS

This work was partly supported by the Max Planck Society. We thank Karran Kumar, Tan Sze Yi, Chan Zhi Jing, Nur Fadhilah Noor Azmi, collaborating kindergartens and participating families for their assistance with data collection. Open access publishing facilitated by Victoria University of Wellington, as part of the Wiley ‐ Victoria University of Wellington agreement via the Council of Australasian University Librarians

Footnotes

1

Kuala Lumpur is the capital city located in center of Peninsular Malaysia known for its urban landscape with a population of approximately 2 million people (Statistics Malaysia, 2023b).

2

Selangor is Malaysia's most populated state (more than 7 million people) consisting of both urban and rural areas that hosts the biggest portion of Malaysia's financial activities (Statistics Malaysia, 2023c).

3

For both tasks, the non‐normative efficient tool was demonstrated first, only showing ‘how it works’ but not completing the task. This was followed immediately by the demonstration of the inefficient method and subsequently task completion (goal attainment). This format of demonstration was intended to show a clear preference of method, using only one of the two options to complete each task with a naturalistic flow.

Contributor Information

Rachel Y. Chin, Email: rachelcyy0706@gmail.com.

Frankie T. K. Fong, Email: frankie.fong@vuw.ac.nz.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are openly available in OSF at https://osf.io/xzfcg/overview?view_only=26caf8c6b90c44dc87c83cb89fa3d783.

REFERENCES

  1. Ahmad, Y. , & Yusof, N. M. (2010). Ethnic boundary among students in Malaysian primary schools and social interaction: A conceptual framework. Procedia – Social and Behavioral Sciences, 7, 82–91. 10.1016/j.sbspro.2010.10.013 [DOI] [Google Scholar]
  2. Andrási, K. , Oláh, K. , Zsoldos, R. A. , & Király, I. (2024). Preschoolers retain more details from event sequences 1 week following an ingroup demonstration. Journal of Experimental Child Psychology, 238, 105781. 10.1016/j.jecp.2023.105781 [DOI] [PubMed] [Google Scholar]
  3. Baboo, S. B. (2013). Media literacy in the lifeworlds of Malaysian children. Global Studies of Childhood, 3(1), 72–85. 10.2304/gsch.2013.3.1.72 [DOI] [Google Scholar]
  4. Barr, R. , & Hayne, H. (2003). It's not what you know, it's who you know: Older siblings facilitate imitation during infancy. International Journal of Early Years Education, 11(1), 7–21. 10.1080/09669760304714 [DOI] [Google Scholar]
  5. Baumrind, D. (1978). Parental disciplinary patterns and social competence in children. Youth & Society, 9(3), 239–276. 10.1177/0044118X7800900302 [DOI] [Google Scholar]
  6. Bi, X. , Yang, Y. , Li, H. , Wang, M. , Zhang, W. , & Deater‐Deckard, K. (2018). Parenting styles and parent–adolescent relationships: The mediating roles of behavioral autonomy and parental authority. Frontiers in Psychology, 9, 2187. 10.3389/fpsyg.2018.02187 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Buttelmann, D. , Zmyj, N. , Daum, M. , & Carpenter, M. (2013). Selective imitation of in‐group over out‐group members in 14‐month‐old infants. Child Development, 84(2), 422–428. 10.1111/j.1467-8624.2012.01860.x [DOI] [PubMed] [Google Scholar]
  8. Chuan‐Peng, H. , Xu, Z. , Lazić, A. , Bhattacharya, P. , Seda, L. , Hossain, S. , Jeftić, A. , Özdoğru, A. A. , Amaral, O. B. , Miljković, N. , Ilchovska, Z. G. , Lazarevic, L. B. , Bao, H. W. S. , Ghodke, N. , Moreau, D. , Elsherif, M. C. C. , Ghai, S. , Carneiro, C. F. D. , & Azevedo, F. (2025). Open science in the developing world: A collection of practical guides for researchers in developing countries. Advances in Methods and Practices in Psychological Science, 8(3), 25152459251357565. [Google Scholar]
  9. Corriveau, K. H. , DiYanni, C. J. , Clegg, J. M. , Min, G. , Chin, J. , & Nasrini, J. (2017). Cultural differences in the imitation and transmission of inefficient actions. Journal of Experimental Child Psychology, 161, 1–18. 10.1016/j.jecp.2017.03.002 [DOI] [PubMed] [Google Scholar]
  10. Danovitch, J. H. , & Shenouda, C. K. (2018). Adults' and children's understanding of how expertise influences learning. Experimental Psychology, 65(1), 1–12. 10.1027/1618-3169/a000387 [DOI] [PubMed] [Google Scholar]
  11. Department of Statistics Malaysia . (2023). Household Income Survey Report 2022 . https://open.dosm.gov.my/dashboard/household‐income‐expenditure
  12. DiYanni, C. J. , Corriveau, K. H. , Kurkul, K. , Nasrini, J. , & Nini, D. (2015). The role of consensus and culture in children's imitation of inefficient actions. Journal of Experimental Child Psychology, 137, 99–110. 10.1016/j.jecp.2015.04.004 [DOI] [PubMed] [Google Scholar]
  13. Dovidio, J. F. , Gaertner, S. L. , & Saguy, T. (2007). Another view of “we”: Majority and minority group perspectives on a common ingroup identity. European Review of Social Psychology, 18(1), 296–330. 10.1080/10463280701726132 [DOI] [Google Scholar]
  14. Draper, C. E. , Barnett, L. M. , Cook, C. J. , Cuartas, J. A. , Howard, S. J. , McCoy, D. C. , Merkley, R. , Molano, A. , Maldonado‐Carreño, C. , Obradović, J. , Scerif, G. , Valentini, N. C. , Venetsanou, F. , & Yousafzai, A. K. (2022). Publishing child development research from around the world: An unfair playing field resulting in most of the world's child population under‐represented in research. Infant and Child Development, 32(6), e2375. 10.1002/icd.2375 [DOI] [Google Scholar]
  15. Fong, F. T. K. , & Haun, D. B. (2025). Two gaps in studying high‐fidelity imitation across diverse childhood ecologies. Child Development Perspectives, 19(4), 189–197. [Google Scholar]
  16. Fong, F. T. K. , Imuta, K. , Redshaw, J. , & Nielsen, M. (2021a). When efficiency attenuates imitation in preschool children. British Journal of Developmental Psychology, 39(2), 330–337. 10.1111/bjdp.12366 [DOI] [PubMed] [Google Scholar]
  17. Fong, F. T. K. , Imuta, K. , Redshaw, J. , & Nielsen, M. (2021b). The digital social partner: Preschool children display stronger imitative tendency in screen‐based than live learning. Human Behavior and Emerging Technologies, 3(4), 585–594. 10.1002/hbe2.280 [DOI] [Google Scholar]
  18. Fong, F. T. K. , Nielsen, M. , & Corriveau, K. H. (2023). I copy you as I believe you know about our culture: Combining imitation and selective trust literatures. Infant and Child Development, 32(5), e2453. [Google Scholar]
  19. Fong, F. T. K. , Nielsen, M. , & Legare, C. H. (2022). Revisiting an extant framework: Concerns about culture and task generalization. [DOI] [PubMed]
  20. Fong, F. T. K. , Redshaw, J. , & Nielsen, M. (2021). A cross‐cultural comparison: Imitation of inefficient tool use in Australian and Malaysian preschool children. Journal of Cultural Cognitive Science, 5, 37–47. 10.1007/s41809-021-00077-6 [DOI] [Google Scholar]
  21. Fusaro, M. , & Harris, P. L. (2008). Children assess informant reliability using bystanders' non‐verbal cues. Developmental Science, 11(5), 771–777. 10.1111/j.1467-7687.2008.00728.x [DOI] [PubMed] [Google Scholar]
  22. Hall, L. , Hume, C. , & Tazzyman, S. (2016). Five degrees of happiness: Effective smiley face likert scales for evaluating with children. Proceedings of the 15th International Conference on Interaction Design and Children (311–321). 10.1145/2930674.2930719 [DOI]
  23. Hellmer, K. , Stenberg, G. , & Fawcett, C. (2021). How does preschoolers' conformity relate to parental style, anonymous sharing, and obedience? International Journal of Developmental Science, 15(3–4), 49–59. 10.3233/DEV-210313 [DOI] [Google Scholar]
  24. Keshavarz, S. , & Baharudin, R. (2009). Parenting style in a collectivist culture of Malaysia. European Journal of Social Sciences, 10(1), 66–73. [Google Scholar]
  25. Lew‐Levy, S. , & Amir, D. (2024). Children as agents of cultural adaptation. Behavioral and Brain Sciences, 1‐68, 1–68. 10.1017/S0140525X24001377 [DOI] [PubMed] [Google Scholar]
  26. Lucas, A. J. , Burdett, E. R. R. , Burgess, V. , Wood, L. A. , McGuigan, N. , Harris, P. L. , & Whiten, A. (2017). The development of selective copying: Children's learning from an expert versus their mother. Child Development, 88(6), 2026–2042. 10.1111/cdev.12711 [DOI] [PubMed] [Google Scholar]
  27. Ma, S. , Cui, Y. K. , Suárez, S. , Chen, E. E. , & Corriveau, K. H. (2025). Are dominant figures more trustworthy? Examining the relation between parental authoritarianism and children's trust preferences in the U.S. and China. Infant and Child Development, 34(3), e70023. 10.1002/icd.70023 [DOI] [Google Scholar]
  28. Masiran, R. (2022). A review of parenting in a multicultural country: The Malaysian experience. IntechOpen. 10.5772/intechopen.101173 [DOI] [Google Scholar]
  29. McGuigan, N. (2013). The influence of model status on the tendency of young children to over‐imitate. Journal of Experimental Child Psychology, 116(4), 962–969. 10.1016/j.jecp.2013.05.004 [DOI] [PubMed] [Google Scholar]
  30. Meltzoff, A. N. , & Marshall, P. J. (2018). Human infant imitation as a social survival circuit. Current Opinion in Behavioral Sciences, 24, 130–136. 10.1016/j.cobeha.2018.09.006 [DOI] [Google Scholar]
  31. Mofrad, S. , & Uba, I. (2014). Parenting style preference in Malaysia. The European Conference on Psychology & The Behavioral Sciences .
  32. Mu, Y. , Pei‐Yin, S. , Chi, L. K. , & Saeid, M. (2024). Comparative analysis of parenting styles and their impact on children's well‐being: A cross‐cultural perspective in Malaysia, China, and Taiwan. International Journal of Child Health and Nutrition, 13(1), 31–46. 10.6000/1929-4247.2024.13.01.4 [DOI] [Google Scholar]
  33. Nielsen, M. (2018). The social glue of cumulative culture and ritual behavior. Child Development Perspectives, 12(4), 264–268. 10.1111/cdep.12297 [DOI] [Google Scholar]
  34. Nielsen, M. , Fong, F. T. K. , & Whiten, A. (2021). Chapter ten – Social learning from media: The need for a culturally diachronic developmental psychology. In Advances in child development and behavior (Vol. 61, pp. 317–334). Academic Press. 10.1016/bs.acdb.2021.04.001 [DOI] [PubMed] [Google Scholar]
  35. Nielsen, M. , Haun, D. , Kärtner, J. , & Legare, C. H. (2017). The persistent sampling bias in developmental psychology: A call to action. Journal of Experimental Child Psychology, 162, 31–38. 10.1016/j.jecp.2017.04.017 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Omar, N. , & Dan, W. C. (2017). Multiculturalism and Malaysian children's literature in English. The English Teacher, 36, 1–19. [Google Scholar]
  37. Reddy, G. , & Selvanathan, H. P. (2020). Multiracial in Malaysia: Categories, classification, and campur in contemporary everyday life. In Rocha Z. L. & Aspinall P. J. (Eds.), The palgrave international handbook of mixed racial and ethnic classification (pp. 649–668). Springer International Publishing. 10.1007/978-3-030-22874-3_34 [DOI] [Google Scholar]
  38. Saad, S. (2012). Re‐building the concept of nation building in Malaysia. Asian Social Science, 8(4), 115–123. 10.5539/ass.v8n4p115 [DOI] [Google Scholar]
  39. Singh, L. (2024). A vision for a diverse, inclusive, equitable, and representative developmental science. Developmental Science, 27(6), e13548. [DOI] [PubMed] [Google Scholar]
  40. Spence, J. L. , & Imuta, K. (2020). Age‐related changes in children's accent‐based resource distribution. Journal of Experimental Child Psychology, 193, 104807. 10.1016/j.jecp.2020.104807 [DOI] [PubMed] [Google Scholar]
  41. Statistics Malaysia . (2023a). Share of population in Malaysia from 2019 to 2023, by ethnicity [Graph]. In Statista. https://www.statista.com/statistics/1017372/malaysia‐breakdown‐of‐population‐by‐ethnicity/
  42. Statistics Malaysia . (2023b). Number of population in Malaysia as of July 2023, by state (in millions) [Graph]. In Statista. https://www.statista.com/statistics/1040670/malaysia‐population‐distribution‐by‐state/
  43. Statistics Malaysia . (2023c). Share of the gross domestic product (GDP) of Malaysia in 2022, by state [Graph]. In Statista. https://www.statista.com/statistics/1181820/malaysia‐gross‐domestic‐product‐contribution‐by‐state/
  44. Tan, C. B. Y. , Stephen, I. D. , Whitehead, R. , & Sheppard, E. (2012). You look familiar: How Malaysian Chinese recognize faces. PLoS One, 7(1), e29714. 10.1371/journal.pone.0029714 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Tham, D. S. Y. , Bremner, J. G. , & Hay, D. (2017). The other‐race effect in children from a multiracial population: A cross‐cultural comparison. Journal of Experimental Child Psychology, 155, 128–137. 10.1016/j.jecp.2016.11.006 [DOI] [PubMed] [Google Scholar]
  46. Wang, Z. , Fong, F. T. K. , & Meltzoff, A. N. (2021). Enhancing same‐gender imitation by highlighting gender norms in Chinese pre‐school children. British Journal of Developmental Psychology, 39, 133–152. 10.1111/bjdp.12356 [DOI] [PubMed] [Google Scholar]
  47. Wilks, M. , Kirby, J. , & Nielsen, M. (2019). Developmental changes in young children's willingness to copy the antisocial actions of ingroup members in a minimal group context. Developmental Psychology, 55(4), 709–721. 10.1037/dev0000667 [DOI] [PubMed] [Google Scholar]
  48. Yang, X. , Yang, F. , Guo, C. , & Dunham, Y. (2022). Which group matters more: The relative strength of minimal vs. gender and race group memberships in children's intergroup thinking. Acta Psychologica, 229, 103685. 10.1016/j.actpsy.2022.103685 [DOI] [PubMed] [Google Scholar]
  49. Yew, W. C. , Awang, A. H. , Selvadurai, S. , Mohd Noor, M. , & Chang, P. K. (2021). A comparative study of Islam and Buddhism: A multicultural society perspective. Religion, 12(12), 1098. 10.3390/rel12121098 [DOI] [Google Scholar]
  50. Yusof, N. M. , Abdullah, A. C. , & Ahmad, N. (2014). Multicultural education practices in Malaysian preschools with multiethnic or monoethnic environment. International Journal of Multicultural and Multireligious Understanding, 1(1), 12–23. 10.18415/ijmmu.v1i1.7 [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The data that support the findings of this study are openly available in OSF at https://osf.io/xzfcg/overview?view_only=26caf8c6b90c44dc87c83cb89fa3d783.


Articles from The British Journal of Developmental Psychology are provided here courtesy of Wiley

RESOURCES