Hostname: page-component-78c5997874-s2hrs Total loading time: 0 Render date: 2024-11-10T05:54:22.910Z Has data issue: false hasContentIssue false

A comparison of WPPSI-IV performance between Finland-Swedish minority children and the Scandinavian test norms: Findings from The FinSwed Study

Published online by Cambridge University Press:  22 November 2023

Jannika Salonen
Affiliation:
Department of Psychology and Logopedics, Faculty of Medicine, University of Helsinki, Helsinki, Finland Department of Child Neurology, Helsinki University Hospital and University of Helsinki, Helsinki, Finland
Susanna Slama
Affiliation:
Department of Psychology and Logopedics, Faculty of Medicine, University of Helsinki, Helsinki, Finland Terveystalo, Helsinki, Finland
Anu Haavisto*
Affiliation:
Department of Psychology and Logopedics, Faculty of Medicine, University of Helsinki, Helsinki, Finland Department of Women’s and Children’s Health, Karolinska Institutet, Stockholm, Sweden
Johanna Rosenqvist
Affiliation:
Department of Psychology and Logopedics, Faculty of Medicine, University of Helsinki, Helsinki, Finland Division of Neuropsychology, HUS Neurocenter, Helsinki University Hospital and University of Helsinki, Helsinki, Finland Mehiläinen Therapy Clinic, Helsinki, Finland
*
Corresponding author: Anu Haavisto, email: anu.haavisto@ki.se
Rights & Permissions [Opens in a new window]

Abstract

Objective:

The Swedish Wechsler Preschool and Primary Scale of Intelligence (WPPSI-IV) is commonly used for assessing young children belonging to the Swedish-speaking minority in Finland (Finland-Swedes), but there is no information about the generalizability of this test and its norms to this minority. Cross-cultural comparisons of WPPSI-IV are also scarce. We compared the performance of Finland-Swedish children to the Scandinavian norms of the Swedish WPPSI-IV and explored the relationship between sociodemographic factors (age, sex, parental education level, bilingualism) and the performance.

Method:

The Swedish WPPSI-IV was administered to 79 typically developing 5–6-year-old Finland-Swedish children assessed for The FinSwed Study. Their performance was compared to the Scandinavian norms using MANOVA, t-test, and confidence interval comparisons. Associations with sociodemographic variables were explored using regression analyses.

Results:

Finland-Swedish children performed, on average, 1/3 SD higher than the Scandinavian norms, a difference which was statistically significant with medium-sized effects. However, individual subtests and indexes did not differ significantly from the norms. Significant associations with sociodemographic factors were found for some but not all index scores.

Conclusions:

This study provides clinically important information for using the Swedish WPPSI-IV with the Finland-Swedish minority and demonstrates aspects that clinicians working with this minority should take into account. The results are presumably partly explained by characteristics of the present sample, and partly by cultural and linguistic differences between the Finland-Swedish population and the Scandinavian countries. The findings also illustrate that cross-cultural differences in cognitive performance may be present even between similar cultures with the same language.

Type
Research Article
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited.
Copyright
Copyright © INS. Published by Cambridge University Press 2023

Cognitive assessments with preschoolers can be of great importance for early detection of different developmental delays and disorders (Baron & Anderson, Reference Baron and Anderson2012). This is an age of fast cognitive development and brain maturation (Korkman et al., Reference Korkman, Kemp and Kirk2001; Tsujimoto, Reference Tsujimoto2008), which also makes it an ideal time for interventions and treatments (Baron & Anderson, Reference Baron and Anderson2012). Cognitive development is not only dependent on biological factors, but also on the child’s environment. Studies have found differences in cognitive test performance between different countries and cultures (e.g., Duggan et al., Reference Duggan, Awakon, Loaiza and Garcia-Barrera2019; Pérez-García et al., Reference Pérez-García, Luna, Torres-Espínola, Martínez-Zaldívar, Anjos, Steenweg-de Graaff, Weber, Grote, Gruszfeld, Verduci, Poncelet, Escribano, Tiemeier, Koletzko and Campoy2019; Roivainen, Reference Roivainen2010, Reference Roivainen2019; Rosenqvist et al., Reference Rosenqvist, Lahti-Nuuttila, Urgesi, Holdnack, Kemp and Laasonen2017). Assessing individuals belonging to linguistic and cultural minorities is complicated, since tests are rarely developed for such groups. Although cross-cultural neuropsychology has been a growing field of research in the last decades (Olson & Jacobson, Reference Olson and Jacobson2015), studies regarding cross-cultural child neuropsychology are still limited (for a review, see Byrd et al., Reference Byrd, Arentoft, Scheiner, Westerveld and Baron2008).

When conducting cognitive assessments with children aged 2–7 years, clinicians can use the Wechsler Preschool and Primary Scale of Intelligence (WPPSI; Wechsler, Reference Wechsler2014). The WPPSI is part of the Wechsler scales; some of the most used tests among psychologists worldwide (Benson et al., Reference Benson, Floyd, Kranzler, Eckert, Fefer and Morgan2019; Piotrowski, Reference Piotrowski2017; Rabin et al., Reference Rabin, Paolillo and Barr2016). This study investigated the generalizability of the Swedish version of WPPSI-IV (Wechsler, Reference Wechsler2014) to 5–6-year-old children belonging to the Swedish-speaking minority in Finland.

Cross-cultural aspects of the Wechsler scales for children

The Wechsler scales for children have been adapted to many different languages and norms have been gathered in several countries. Some adaptations have also been made for certain language minorities. Some minorities have separate published tests (e.g., WPPSI-IV CDN-F for French-speaking Canadian children, although relying on norms for English-speaking Canadians; Wechsler, Reference Wechsler2013) and for some minorities, smaller modifications have been made to improve reliability (e.g., special instructions in the German WISC-IV for German-speaking Swiss children; Grob et al., Reference Grob, Petermann, Lipsius, Costan-Dorigon, Petermann and Daseking2008). However, overall, tests, norms, or adaptations are seldom published for minority groups.

One minority group with no local Wechsler scales or norms is the Swedish-speaking minority in Finland (Finland-Swedes). Swedish is a national language in Finland and is the official first language of 5.2% of the Finnish population (Official Statistics of Finland, 2021). Approximately 40% of the Finland-Swedes are bilingual (Saarela, Reference Saarela2021). A recent survey for psychologists working with Finland-Swedish children showed that there is great variability in the usage of tests and norms (Rosenqvist et al., Reference Rosenqvist, Slama and Haavisto2022). When assessing young children, the Swedish WPPSI-IV with Scandinavian norms was the most common test choice and using it with Finland-Swedish children may seem suitable because of the shared language. However, the Swedish spoken in Finland differs somewhat from the Swedish spoken in Sweden, with different words and expressions, as well as with influences from the Finnish language. Accordingly, some psychologists considered especially verbal subtests to be problematic when used with this population (Rosenqvist et al., Reference Rosenqvist, Slama and Haavisto2022). There are also cultural differences, such as differences regarding education systems and demographic variability between Finland and Scandinavia, which are factors known to influence cognitive test performance (e.g., Babcock et al., Reference Babcock, Miller, Saklofske and Zhu2018). Therefore, information is needed about how this test and its norms are generalizable to Finland-Swedish children.

Cross-cultural comparisons of WPPSI are still scarce. One recent study compared the performance of the South Korean WPPSI-IV normative sample when scored with Korean versus US norms (Park & Lee, Reference Park and Lee2016). Korean 4–7-year-olds performed lower on most verbal subtests when scored with US norms, which was suggested to be partly due to the cultural loading of some items, since many original questions from the US version were kept in the Korean standardization. They also found that Korean children performed higher on most nonverbal subtests.

Reasons for cross-cultural differences

Several factors explain the differences found in test performance between cultures (for a review, see e.g., Ardila, Reference Ardila2020; Olson & Jacobson, Reference Olson and Jacobson2015), one of which is language. Especially verbal tests have been considered culturally loaded and in need of adaptations across different cultures (International Test Commission, 2019). Cross-cultural comparisons of performance in verbal tests are also difficult, as developing a language test of the exact same difficulty can be considered impossible (Roivainen, Reference Roivainen2013). The same words may not exist in different languages or may have different meanings, which also makes the translation of tests difficult (Olson & Jacobson, Reference Olson and Jacobson2015). Further, linguistic differences may affect performance in, for example, working memory tests, due to aspects such as word length (e.g., Georgas et al., Reference Georgas, Van de Vijver, Weiss, Saklofske and Georgas2003; Roivainen, Reference Roivainen2019). The number of bilinguals in a population is also one factor that can relate to test performance between cultures (e.g., Garratt & Kelly, Reference Garratt and Kelly2007; Lauchlan et al., Reference Lauchlan, Parisi and Fadda2012; Rosselli et al., Reference Rosselli, Ardila, Navarrete and Matute2010). For instance, there seems to be a monolingual advantage in vocabulary (Bialystok et al., Reference Bialystok, Luk, Peets and Yang2010), for young children especially in expressive vocabulary (Korkman et al., Reference Korkman, Stenroos, Mickos, Westman, Ekholm and Byring2012; Korpinen et al., Reference Korpinen, Slama, Rosenqvist and Haavisto2023; Thordardottir, Reference Thordardottir2011). However, linguistic reasons cannot be the only explanation for cross-cultural differences, since differences have been found also on nonverbal tests (Roivainen, Reference Roivainen2019; Rosselli & Ardila, Reference Rosselli and Ardila2003) as well as between people from different countries who speak the same language (Babcock et al., Reference Babcock, Miller, Saklofske and Zhu2018; Duggan et al., Reference Duggan, Awakon, Loaiza and Garcia-Barrera2019; Miller et al., Reference Miller, Weiss, Beal, Saklofske, Zhu and Holdnack2015).

Sociodemographic factors, such as education level, are also known to be associated with cross-cultural differences in cognitive test performance (Olson & Jacobson, Reference Olson and Jacobson2015). For children, parental education level explains a significant proportion of the variance in cognitive test scores, including WPPSI-IV (Eriksen et al., Reference Eriksen, Kesmodel, Underbjerg, Kilburn, Bertrand and Mortensen2013; Park & Lee, Reference Park and Lee2016; Raiford et al., Reference Raiford, Coalson, Engi, Raiford and Coalson2014; Wechsler, Reference Wechsler2014). Differences in the education system the child belongs to can also have an effect, which is shown, for example, in international comparisons of educational attainment (Leino et al., Reference Leino, Ahonen, Hienonen, Hiltunen, Lintuvuori, Lähteinen, Lämsä, Nissinen, Nissinen, Puhakka, Pulkkinen, Rautopuro, Sirén, Vainikainen and Vettenranta2019; Schleicher, Reference Schleicher2019). Lastly, age (Korkman et al., Reference Korkman, Lahti-Nuuttila, Laasonen, Kemp and Holdnack2013; Rosenqvist et al., Reference Rosenqvist, Lahti-Nuuttila, Urgesi, Holdnack, Kemp and Laasonen2017) and sex (Giofrè et al., Reference Giofrè, Allen, Toffalini and Caviola2022; Palejwala & Fine, Reference Palejwala and Fine2015; Walter et al., Reference Walter, Daseking and Pauls2020) are associated with cognitive test performance, and these associations may be mediated by cultural factors. For example, developmental curves have been shown to differ between different cultures (Rosenqvist et al., Reference Rosenqvist, Lahti-Nuuttila, Urgesi, Holdnack, Kemp and Laasonen2017).

More subtle cultural differences may also contribute to the discrepancies in test performance between cultural groups. Ardila (Reference Ardila2005) presented cultural factors that may affect cognitive testing, including the social elements of the testing situation, the expectancy to do one’s best, as well as different culture-dependent elements (e.g., pictures) used in the tests. For example, Georgas et al. (Reference Georgas, Van de Vijver, Weiss, Saklofske and Georgas2003) found that South Korean children performed significantly better in some processing speed tests than children from many other countries, which was thought to be a result of cultural values about educational achievement. Differences in processing speed tests have also been hypothesized to be related to familiarity and values regarding speeded performance (Agranovich et al., Reference Agranovich, Panter, Puente and Touradji2011; Ardila, Reference Ardila2005; Eizaguirre et al., Reference Eizaguirre, Vanotti, Aguayo Arelis, Rabago Barajas, Cores, Macías, Benedict and Cáceres2020). Presumably, there are also cultural factors affecting cognitive test performance yet to be identified (Olson & Jacobson, Reference Olson and Jacobson2015).

Some differences found in cross-cultural comparisons could also relate to aspects of the standardization process. Differences regarding sample size and exclusion criteria could lead to some variability in standardized scores in the normative samples (e.g., Roivainen, Reference Roivainen2013). Nevertheless, the normative samples are meant to represent the general population and its normal variability and hence reflect the culture and demographics of that population.

Aims

The main aim of the present study was to compare the performance of Finland-Swedish 5–6-year-old children on the Swedish WPPSI-IV to the Scandinavian test norms. Since previous studies have shown differences in test performance also between different cultures with the same language (e.g., Babcock et al., Reference Babcock, Miller, Saklofske and Zhu2018), we hypothesized that some differences would be found between the Finland-Swedish children’s performance and the Scandinavian test norms. Further, this study aimed to explore how age, sex, bilingualism, and parental education relate to the WPPSI-IV performance of Finland-Swedish children. Based on previous results (e.g., Giofrè et al., Reference Giofrè, Allen, Toffalini and Caviola2022; Korpinen et al., Reference Korpinen, Slama, Rosenqvist and Haavisto2023; Raiford et al., Reference Raiford, Coalson, Engi, Raiford and Coalson2014; Rosenqvist et al., Reference Rosenqvist, Lahti-Nuuttila, Urgesi, Holdnack, Kemp and Laasonen2017), we hypothesized that there would be some associations between sociodemographic variables and WPPSI-IV performance. These results are important to psychologists working with this minority, but also contribute to the growing field of cross-cultural child neuropsychology.

Methods

Participants

The sample in the present study consisted of 79 Finland-Swedish children aged 5:0–6:11 years assessed for The FinSwed Study, a large study examining how Finland-Swedish children perform on Swedish cognitive tests. Only monolingual Swedish-speaking and bilingual Swedish-Finnish-speaking children attending Swedish-speaking daycare or pre-primary school were included. Bilingualism was reported by parents. All bilinguals had been exposed to both languages before the age of one (as described in Korpinen et al., Reference Korpinen, Slama, Rosenqvist and Haavisto2023). The exclusion criteria were based on the exclusion criteria for the standardization of the Scandinavian WPPSI-IV (Wechsler, Reference Wechsler2014). The sample thus included typically developing children with no developmental, neurological, or psychiatric diagnosis, or individual support. Information about the exclusion process is provided in Figure 1. Demographics of the sample are presented in Table 1.

Figure 1. Flow chart of the exclusion process. Note. 1Four families withdrew after initial consent due to the COVID-19 pandemic.

Table 1. Demographic characteristics of the sample (N = 79)

Note. Parental education was grouped into three levels: Level 1: Upper secondary education, vocational education, or lower; Level 2: Bachelor’s degree or University of applied sciences degree; Level 3: Master’s or Doctoral degree. Ages reported in the table are the ages during the first assessment session.

Procedure

The study was completed in accordance with the Helsinki Declaration. Ethical approval was granted by the University of Helsinki Ethical Review Board in the Humanities and Social and Behavioural Sciences 6/2019. The Heads of Early Childhood Education of each municipality, heads of each participating daycare and pre-primary school, and parents of the participating children gave their written consent before participation.

The sampling procedure was carried out to represent the Swedish-speaking population in mainland Finland. All bilingual regions in Finland were included in the study. Municipalities were chosen within these regions to match the percentage of Finland-Swedish children living in urban, semi-urban, and rural areas, as well as the education level of the region. For practical reasons, clinical psychologists assisting with the data collection chose which daycares and pre-primary schools to include. Randomized sampling was then used within the chosen daycares and pre-primary schools. The final sample was representative of the Finland-Swedish population census (Official Statistics of Finland, 2017) regarding home region and type of region (urban, semi-urban, rural). However, parental education level for the present study sample (see Table 1) was significantly higher than the education level of Finland-Swedish 30–40-year-olds from bilingual regions in general, of which 25.6% have a higher university degree, χ2 (1, n = 156) = 20.03, p < .001 (Official Statistics of Finland, 2017).

The cognitive assessments were administered by clinical psychologists or trained research assistants during 10/2019–2/2021. All protocols were examined and, if needed, corrected when the data was transferred to electronical form. Most of the assessments were conducted before the outbreak of the COVID-19 pandemic (n = 56). Assessments were typically divided into three sessions (range = 1–6), which included breaks if needed. Due to the pandemic and societal restrictions, for three children, the interval between sessions was longer than 2 months. In these cases, test scores for the latter sessions were calculated using a new test age. During three assessments, the administrator wore a face mask, following the recommendations of mask use in Finland. No children wore masks during the assessments.

Measures

WPPSI-IV

WPPSI-IV (Wechsler, Reference Wechsler2014) is a test used by psychologists to measure general cognitive ability in children aged 2:6–7:7 years. For children aged 4:0–7:7 years, the Swedish test version consists of 15 subtests (M = 10, SD = 3), which produce scores for five primary indexes (M = 100, SD = 15): Verbal Comprehension Index (VCI), Visual Spatial Index (VSI), Fluid Reasoning Index (FRI), Working Memory Index (WMI), and Processing Speed Index (PSI); four ancillary indexes, of which only Vocabulary Acquisition Index (VAI) was used in this study; as well as the Full Scale IQ (FSIQ, M = 100, SD = 15). These are presented in Table 2. The Swedish test version has Scandinavian norms gathered in Sweden, Norway, and Denmark (Wechsler, Reference Wechsler2014).

Table 2. Means, SD, and ranges for the subtests and indexes in the Finland-Swedish sample (N = 79)

Note. The bolded subtests compose the FSIQ. The Verbal Comprehension Index is composed by the two bolded subtests. The processing speed subtest Animal Coding was not administered in this study.

The appropriateness of the language used in the test was evaluated by the researchers in The FinSwed Study, with assistance from a linguist. The language was generally regarded as suitable for the Finland-Swedish population. However, some minor phrasing modifications were made. In three country-specific items of the Information subtest, questions regarding Finland were asked in addition to the questions about Sweden.Footnote 1 The scaled scores were calculated based on the questions regarding Finland.

Background information

Parents of the participating children completed a background questionnaire with questions regarding the child’s home language, daycare background, possible developmental problems, and medical conditions, as well as regarding the parents’ language and education. The parents’ level of education was categorized into three levels.

Data analyses

Due to missing data, mainly caused by administration errors, some items were imputed using Expectation Maximation estimation. There were missing data in six subtests, five on item and one on subtest level, for 1–2 participants each. Data was imputed for seven children, in 1–2 subtests each. All final imputed values were in the expected range.

Based on graphical inspection, assumptions for normal distribution were met. The variance was compared to the expected SD of 3 for subtests and 15 for indexes, using one-sample chi-square test, with Bonferroni correction. The analyses were first done for the five primary indexes and the VAI together with FSIQ and then for the 13 subtest scores using the EnvStats-package (Millard, Reference Millard2013). The variance of the Information subtest (χ2 (78) = 128.49, p = .008), and the VCI (χ2 (78) = 132.97, p = .001), were significantly larger in the present sample than in the normative sample (see Table 2).

In the main analysis, a MANOVA was used to compare the performance of Finland-Swedish children on the 14 subtests against the normative means of the Scandinavian WPPSI-IV norms. Further, a t-test was undertaken to compare performance on the FSIQ. The differences in individual subtests and indexes were further analyzed using confidence intervals based on nonparametric bootstrapping with 5000 resamples using MANOVA.RM (Friedrich et al., Reference Friedrich, Konietschke and Pauly2019). Paired samples t-test was used to compare the results in the Information subtest when Finland-specific or Sweden-specific items were used. Multiple linear regression analyses were performed to analyze the influence of sociodemographic factors (age, sex, bilingualism, and parental education) on the WPPSI-IV indexes and the FSIQ. To further examine the effect of age, generalized additive model (GAM) using cubic regression splines (O’sullivan et al., Reference O’sullivan, Yandell and Raynor1986; Wood & Augustin, Reference Wood and Augustin2002) was used with the mgcv-package (Wood, Reference Wood2017). Finally, to explore the effect of the sample’s overrepresentation of highly educated parents, another t-test was conducted in a subsample matched by maternal education level to the Scandinavian normative sample. Significance was set at two-tailed p ≤ .05. Eta-squared (η2), Cohen’s d, and R 2 were used as measures of effect size. Statistical analyses were completed using IBM SPSS Statistics 28.0 and R software version 4.05.

Results

Performance on the WPPSI-IV compared to Scandinavian norms

Based on MANOVA, there was a significant main effect in the WPPSI-IV, Wilk’s lambda = 0.39, F(14, 65) = 7.55, p < .001, η 2 = .07, with Finland-Swedish children performing significantly higher compared to the Scandinavian population norms, with a medium effect size. The FSIQ was also significantly higher in Finland-Swedish children compared to the Scandinavian population norms, t(78) = 4.30, p < .001, d = 0.48. When using bootstrapped confidence interval comparisons, no significant differences were found in individual subtests or indexes, nor in the FSIQ (Figures 2 and 3).

Figure 2. Means and bootstrapped confidence intervals (standard errors presented in parenthesis) of WPPSI-IV subtest scores compared to the Scandinavian norms (Mean = 10).

Figure 3. Means and bootstrapped confidence intervals (standard errors presented in parenthesis) of WPPSI-IV FSIQ and index scores compared to the Scandinavian norms (Mean = 100). Note. FSIQ: Full Scale IQ; VCI: Verbal Comprehension Index; VSI: Visual Spatial Index; FRI: Fluid Reasoning Index; WMI: Working Memory Index; PSI: Processing Speed Index; VAI: Vocabulary Acquisition Index.

Country-specific verbal items

The mean standard score for the Information subtest was significantly higher when the Finland-specific questions were used (M = 9.90) instead of the Sweden-specific ones (M = 9.48), t (76) = −2.97, p = .004, d = 0.11 (n = 77, data missing for two participants). The largest differences in standard scores due to these items were found in children with high total raw scores (range of difference in standard scores: 0–6).

Relationship with age, sex, bilingualism, and parental education

Multiple linear regression analyses (Table 3) presented significant associations in the WMI, PSI, and VSI with age. However, visual inspection showed some outliers in these indexes (VSI n = 6, WMI n = 2, PSI n = 4), which were then excluded. In follow-up regression analyses, only the PSI age effect remained significant (p = .002), with standardized test scores decreasing with increasing age. The PSI was also the only primary index showing a significant effect of age when using GAM (Figure 4), which showed that especially the youngest children performed significantly higher than the Scandinavian mean, whereas older children performed similarly to the norms. The nonlinearity in the middle of the curve is a result of the outliers mentioned previously.

Figure 4. Generalized additive model showing age effects on the PSI (N = 79).

Table 3. Multiple linear regression analyses exploring the relationships between sociodemographic variables and performance on WPPSI-IV FSIQ and indexes (N = 79)

Note. PED: Parental education; FSIQ: Full Scale IQ; VCI: Verbal Comprehension Index; VSI: Visual Spatial Index; FRI: Fluid Reasoning Index; WMI: Working Memory Index; PSI: Processing Speed Index; VAI: Vocabulary Acquisition Index.

10 = Girl, 1 = Boy, 20 = Monolingual, 1 = Bilingual, 30 = PED Level 1 and Level 3, 1 = PED Level 2, 40 = PED Level 1 and Level 2, 1 = PED Level 3.

There was also a significant difference in the VCI depending on parental education level: Children with parents with a higher university degree (Level 3) performed significantly better than children with parents with upper secondary or vocational education (Level 1). Bilingual children performed significantly lower on the VSI and VAI compared to monolingual children. Girls scored significantly higher on the PSI than boys.

The effect of parental education level was further explored by inspecting the FSIQ and index scores for all levels of education (Figure 5). Children of parents with lower education levels (upper secondary or vocational education) performed similarly as the normative mean or higher in all indexes except the VCI and the VAI.

Figure 5. Means and SD in the FSIQ and indexes according to parental education level (N = 79). Note. PED: Parental education level; Level 1: Upper secondary education, vocational education, or lower; Level 2: Bachelor’s degree or University of applied sciences degree; Level 3: Master’s or Doctoral degree. FSIQ: Full Scale IQ; VCI: Verbal Comprehension Index; VSI: Visual Spatial Index; FRI: Fluid Reasoning Index; WMI: Working Memory Index; PSI: Processing Speed Index; VAI: Vocabulary Acquisition Index.

To examine how much the overrepresentation of highly educated parents affected the results, a follow-up t-test was done with a sample that matched the Scandinavian normative sample regarding maternal education level. Children with mothers with a Bachelor’s or University of applied sciences degree (Level 2) and with a Master’s or Doctoral degree (Level 3) were randomly excluded until the proportions of the three education levels matched the Scandinavian normative group. The results showed that the FSIQ was still significantly higher (n = 40, M = 105.00, SD = 13.61) for the Finland-Swedish children than the normative mean, t (39) = 2.32, p = .025, d = 0.37.

Discussion

The main aim of the present study was to compare Finland-Swedish 5–6-year-old children’s performance on the WPPSI-IV to the Scandinavian test norms. Assessments during this age are important, but there is a risk for inaccuracy when using norms gathered in one culture with individuals from another culture (e.g., Babcock et al., Reference Babcock, Miller, Saklofske and Zhu2018). As the Swedish WPPSI-IV is the most commonly used cognitive test for assessing young Finland-Swedish children (Rosenqvist et al., Reference Rosenqvist, Slama and Haavisto2022), information about how this test and its Scandinavian norms are generalizable to this minority is necessary. This study showed that Finland-Swedish children, in general, scored approximately 1/3 SD higher on the WPPSI-IV than the Scandinavian normative group. Different sociodemographic factors (age, sex, bilingualism, and parental education) were also found to relate to WPPSI-IV performance, which was expected based on previous research with other tests (e.g., Giofrè et al., Reference Giofrè, Allen, Toffalini and Caviola2022; Korkman et al., Reference Korkman, Stenroos, Mickos, Westman, Ekholm and Byring2012; Raiford et al., Reference Raiford, Coalson, Engi, Raiford and Coalson2014; Rosenqvist et al., Reference Rosenqvist, Lahti-Nuuttila, Urgesi, Holdnack, Kemp and Laasonen2017).

In general, Finland-Swedish children performed higher on the Swedish WPPSI-IV than the Scandinavian norms with medium effect sizes. The mean FSIQ of Finland-Swedish children was almost 7 IQ scores higher than in the norms, and 5 IQ scores higher when the education levels were matched to the normative group. However, when using confidence interval comparisons, no individual subtest or index differed significantly from the norms. This may be due to the present sample size being modest. Further, it is noteworthy that confidence interval comparisons are a stricter method than pairwise t-tests. Previous studies using t-tests with large samples have found significant results with group differences smaller than in the present study (e.g., Babcock et al., Reference Babcock, Miller, Saklofske and Zhu2018; Park & Lee, Reference Park and Lee2016).

There are some possible explanations as to why this sample of Finland-Swedish children scored overall higher than the normative Scandinavian sample of the WPPSI-IV. One reason could be differences in the exclusion criteria of the samples. Although the exclusion criteria for the present study followed the criteria in the WPPSI-IV manual (Wechsler, Reference Wechsler2014), there are differences between the countries (e.g., regarding educational support and diagnostics) that may have led to more strict exclusions from the sample. This may also have inflated the parental education. In the present sample, 18.6% of the selected children were excluded due to language and 27.3% due to exclusion criteria. However, a comparison of the exclusions could not be made since this information is not provided in the WPPSI-IV manual. Further, the performance of the present sample was normally distributed with a similar variance as in the normative data in all but one subtest and one index, which suggests that the exclusion criteria have not been too strict.

Another factor that may explain the generally higher scores for Finland-Swedish children is parental education level. Finland-Swedes are, in general, a quite highly educated population, also compared to the Scandinavian countries (Eurostat, 2022; Official Statistics of Finland, 2017). The sample for this study, however, had an overrepresentation of parents with higher university education (44.3% of the mothers), both compared to the population census for Finland-Swedes (25.6%) and the Scandinavian WPPSI-IV normative sample (22.2%). Furthermore, the present study had an underrepresentation of parents with lower education levels, which was also the case in the Scandinavian normative sample (Wechsler, Reference Wechsler2014). Since parental education level is a well-known predictor of cognitive test performance (Eilertsen et al., Reference Eilertsen, Thorsen, Holm, Bøe, Sørensen and Lundervold2016; Eriksen et al., Reference Eriksen, Kesmodel, Underbjerg, Kilburn, Bertrand and Mortensen2013), this is presumably one reason for the high scores of the present sample. Nonetheless, children with parents in the lowest education level (upper secondary or vocational education) still performed similarly as the normative mean or higher in all indexes except the VCI and the VAI. Additionally, when matching the present sample with the Scandinavian normative sample regarding the maternal education levels, the mean FSIQ was still significantly higher by 5 IQ scores compared to the Scandinavian norms. The maternal education level of the follow-up subsample also represented the population census for 30–40-year-old Finland-Swedes (Official Statistics of Finland, 2017). The sample size in this follow-up analysis was, however, quite small (n = 40), which increases the likelihood of a type 2 error. Nevertheless, it seems that the higher scores in the Finland-Swedish sample can only partly be explained by the overrepresentation of highly educated parents, whereas the rest of the difference represents some other characteristics in the Finland-Swedish population or variations between the samples.

Moreover, the difference in the FSIQ between children with parents in the lowest education level group (upper secondary or vocational degree) and children with parents in the highest education level group (higher university degree) was in this study smaller (Mean difference = 8.2) than in the WPPSI-IV US normative sample (Mean difference = 12.5) (Raiford et al., Reference Raiford, Coalson, Engi, Raiford and Coalson2014). This, together with the fact that parental education level did not significantly affect performance on the FSIQ when other sociodemographic factors were taken into account, is consistent with previous studies discussing that the association between parental socioeconomic status and cognitive performance may be smaller in Finland and the Scandinavian countries compared to other countries (for a summary, see Eilertsen et al., Reference Eilertsen, Thorsen, Holm, Bøe, Sørensen and Lundervold2016).

Although Finland and the Scandinavian countries have similar cultures, there are still cultural and linguistic factors that could explain some of the differences in the WPPSI-IV scores. For example, the education system in Finland differs in some regard from the education system in Scandinavian countries, and there are differences in how the countries perform on international comparisons of educational attainment, in which Finland usually has had somewhat higher scores (Leino et al., Reference Leino, Ahonen, Hienonen, Hiltunen, Lintuvuori, Lähteinen, Lämsä, Nissinen, Nissinen, Puhakka, Pulkkinen, Rautopuro, Sirén, Vainikainen and Vettenranta2019; Schleicher, Reference Schleicher2019). Presumably, there are also systemic and cultural differences in the early childhood education systems between Finland and the Scandinavian countries, which could lead to some differences in WPPSI-IV scores. In addition, linguistic differences and bilingualism may also affect results, especially in the verbal subtests, which is further discussed below.

Differences in verbal subtests and indexes

The difference in mean scores between Finland-Swedish children and the Scandinavian norms was in general smaller for verbal subtests than for nonverbal ones. Four out of six verbal subtests were close to the Scandinavian mean, with Similarities and Comprehension being exceptions. The VCI and the VAI were also the indexes closest to the Scandinavian normative mean. This is somewhat surprising since in general, it is thought that verbal subtests are more culturally dependent than nonverbal ones (Wechsler, Reference Wechsler2014). Additionally, out of all index scores, performance on the VCI was most strongly predicted by parental education level. This is in line with previous studies showing that parental education level seems to have a larger effect on verbal than nonverbal subtests (e.g., Cianci et al., Reference Cianci, Orsini, Hulbert and Pezzuti2013; Eilertsen et al., Reference Eilertsen, Thorsen, Holm, Bøe, Sørensen and Lundervold2016; Rindermann et al., Reference Rindermann, Flores-Mendoza and Mansur-Alves2010). Although the present sample had an overrepresentation of highly educated parents, Finland-Swedish children performed generally better in nonverbal than verbal tests. Thus, it seems that the verbal subtests were slightly more difficult for this population. Likewise, Finland-Swedish clinicians have expressed concerns regarding the verbal subtests (Rosenqvist et al., Reference Rosenqvist, Slama and Haavisto2022). Similar results have been found for South Korean children, who performed lower in the verbal subtests than in the nonverbal subtests of WPPSI-IV when scored with American norms (Park & Lee, Reference Park and Lee2016). Regarding the VAI, which measures basic receptive and expressive language skills (Wechsler, Reference Wechsler2012), the Finland-Swedish children’s performance may also reflect the nature of the subtests in this index. After reaching the average level of performance, there are only a few items left for 5–6-year-olds and these items seem to have a relatively high level of difficulty.

Another reason for the relatively lower verbal scores may be that approximately 40% of the children in the sample (and in the population census, Saarela, Reference Saarela2021) were bilingual. Bilingualism predicted significantly lower scores on the VAI. A similar trend was also found for the VCI, which measures verbal reasoning, verbal concept formulation, and acquired knowledge (Wechsler, Reference Wechsler2012). Previously, the bilinguals in this study have been shown to perform significantly lower on Similarities, Vocabulary, and Picture Naming than monolinguals (Korpinen et al., Reference Korpinen, Slama, Rosenqvist and Haavisto2023). Similar bilingual disadvantages in expressive vocabulary tasks have also been found previously in young Finland-Swedish children (Korkman et al., Reference Korkman, Stenroos, Mickos, Westman, Ekholm and Byring2012; Westman et al., Reference Westman, Korkman, Mickos and Byring2008) and internationally (e.g., Thordardottir, Reference Thordardottir2011).

The Similarities and Comprehension subtests had two of the largest mean differences compared to the Scandinavian norms. In addition to verbal skills, these subtests require abstract reasoning and the performance on these subtests was, thus, in line with the Finland-Swedish children’s performance on nonverbal abstract reasoning tasks. Similarities also measures associative and categorical thinking (Wechsler, Reference Wechsler2012), and perhaps these kinds of tasks are practiced more in Finland-Swedish early childhood education. Likewise, Finland-Swedish school-aged children have scored significantly higher in the Similarities subtest of WISC-V when compared to the Scandinavian norms (Rosenqvist et al., Reference Rosenqvist, Slama and Haavisto2023).

There were also some differences in the country-specific items in the Information subtest. Some children were better at answering the questions regarding Sweden, which can presumably be explained by the fact that Finland-Swedish children often consume Swedish media and Swedish child literature. However, overall, Finland-Swedish children scored higher and the standard score for Information was significantly higher when Finland-specific questions were used. This highlights the importance of using items suitable for the cultural group being assessed, but also shows that culture and media can overlap between countries, and therefore developing and adapting a subtest like this is far from straightforward.

Differences in nonverbal subtests and indexes

Finland-Swedish children performed, on average, 6 index scores higher than the mean on the nonverbal indexes VSI, FRI, WMI, and PSI. The mean scaled score for the individual nonverbal subtests was approximately 11. The performance on the nonverbal subtests was more leveled compared to the verbal subtests, without distinct high or low results.

Some relationships between sociodemographic factors and the nonverbal indexes were found. Of particular interest was the significant age effect on the PSI, in which increasing age led to lower scaled scores. The youngest children in the sample (approximately 5–5.5-year-olds) performed significantly higher than the Scandinavian mean, whereas slightly older children (6–7-year-olds) performed closer to the Scandinavian mean. Previous research has shown that the developmental curves and spurts may differ between cultures (Byrd et al., Reference Byrd, Arentoft, Scheiner, Westerveld and Baron2008; Rosenqvist et al., Reference Rosenqvist, Lahti-Nuuttila, Urgesi, Holdnack, Kemp and Laasonen2017). It is possible that the skills measured by the PSI are practiced earlier in Finnish early childhood education, since there is no difference in performance for the oldest children in this sample, nor for school-aged children using WISC-V (Rosenqvist et al., Reference Rosenqvist, Slama and Haavisto2023).

The bilinguals in this sample also scored significantly lower on the VSI than monolinguals. This is quite surprising since there were no differences between mono- and bilingual Finland-Swedish children in nonverbal subtests of the WPPSI-R (Korkman et al., Reference Korkman, Stenroos, Mickos, Westman, Ekholm and Byring2012). Korpinen et al. (Reference Korpinen, Slama, Rosenqvist and Haavisto2023, based on the same data as the present study) have further discussed this result and the possibility of it being a chance finding. There was also a significant relationship between sex and the PSI, with girls scoring significantly higher. Similarly, a recent meta-analysis concluded that females have been shown to perform significantly better than males on WISC processing speed tasks and that this finding has been stable over the years (Giofrè et al., Reference Giofrè, Allen, Toffalini and Caviola2022).

Conclusions and clinical implications

The present study aimed to assess the generalizability of the Swedish WPPSI-IV and its Scandinavian norms to the Finland-Swedish population and to explore the association between WPPSI-IV performance and sociodemographic variables. The results showed significant differences between Finland-Swedish children and the Scandinavian norms and indicate that using the Swedish version with the Finland-Swedish minority imposes some challenges on clinicians. The results also showed that many sociodemographic variables were related to cognitive performance. This further demonstrates that it is complicated to use cognitive tests with individuals from a population with different demographics than the normative group.

This study contributes to a field where information has been lacking and provides support for clinicians to use the Swedish WPPSI-IV with Finland-Swedish children, albeit with caution. Performance on the WPPSI-IV was, in general, significantly higher in the Finland-Swedish minority compared to the Scandinavian norms, with medium-sized effects. These results can partly be explained by the overrepresentation of highly educated parents in the sample, but presumably also by some characteristics in the Finland-Swedish population. The FSIQ was significantly higher than the Scandinavian mean, also when matching parental education to the normative sample. This is essential to remember in clinical settings, as test scores are used to make important clinical decisions. Using the Scandinavian norms directly may lead to milder learning disabilities not being detected. The nonverbal subtests seem to produce scores that are all somewhat higher than the norms, whereas the verbal subtests seem to produce more variable test scores. Performance on verbal subtests requiring abstract reasoning was in line with the performance on nonverbal subtests, while verbal subtests assessing vocabulary or factual knowledge produced the lowest relative scores. Age and bilingualism are also factors important to take into account when interpreting results from WPPSI-IV assessments. In particular, the Scandinavian norms for the PSI seem to be more suitable for Finland-Swedish 6-year-olds than 5-year-olds, whereas bilingualism seems to relate to especially verbal subtests requiring basic vocabulary.

In sum, this study shows that for young children, there are differences in cognitive test performance between cultural groups, despite having the same language and quite a similar culture. Thus, clinicians need to be careful with their interpretations when using norms and tests from other cultures.

Acknowledgements

We thank all the families and clinicians who participated in the data collection and Lecturer of Psychometrics Jari Lipsanen for conducting the main statistical analyses for the study.

Funding statement

The study is supported by a personal grant from Victoriastiftelsen (J.S.) and by grants given to The FinSwed Study by The Swedish Cultural Foundation in Finland, The association Svenska folkskolans vänner, and the foundations Stiftelsen Brita Maria Renlunds minne sr and Oskar Öflunds stiftelse.

Competing interests

None.

Footnotes

In the WPPSI-IV assessments, minor culture-related changes were made in the wording of some verbal questions. All changes were made according to the Statement of Work No. 296412-2 to Master License Agreement No. LSR–111089 with NCS Pearson.

*

Anu Haavisto and Johanna Rosenqvist are the senior authors and contributed equally.

1 All changes were made in accordance with an agreement with NCS Pearson, Statement of Work No. 296412-2 to Master Licence Agreement No. LSR – 111089.

References

Agranovich, A. V., Panter, A. T., Puente, A. E., & Touradji, P. (2011). The culture of time in neuropsychological assessment: Exploring the effects of culture-specific time attitudes on timed test performance in Russian and American samples. Journal of the International Neuropsychological Society, 17 (4), 692701. DOI: https://doi.org/10.1017/S1355617711000592 CrossRefGoogle ScholarPubMed
Ardila, A. (2005). Cultural values underlying psychometric cognitive testing. Neuropsychology Review, 15 (4), 185195. DOI: https://doi.org/10.1007/s11065-005-9180-y CrossRefGoogle ScholarPubMed
Ardila, A. (2020). Cross-cultural neuropsychology: History and prospects. RUDN Journal of Psychology and Pedagogics, 17 (1), 6478. DOI: https://doi.org/10.22363/2313-1683-2020-17-1-64-78 CrossRefGoogle Scholar
Babcock, S. E., Miller, J. L., Saklofske, D. H., & Zhu, J. (2018). WISC-V Canadian norms: Relevance and use in the assessment of Canadian children. Canadian Journal of Behavioural Science/Revue canadienne des sciences du comportement, 50 (2), 97104. DOI: https://doi.org/10.1037/cbs0000096 CrossRefGoogle Scholar
Baron, I. S., & Anderson, P. J. (2012). Neuropsychological assessment of preschoolers. Neuropsychology Review, 22 (4), 311312. DOI: https://doi.org/10.1007/s11065-012-9221-2 CrossRefGoogle ScholarPubMed
Benson, N. F., Floyd, R. G., Kranzler, J. H., Eckert, T. L., Fefer, S. A., & Morgan, G. B. (2019). Test use and assessment practices of school psychologists in the United States: Findings from the 2017 National Survey. Journal of School Psychology, 72, 2948. DOI: https://doi.org/10.1016/j.jsp.2018.12.004 CrossRefGoogle ScholarPubMed
Bialystok, E., Luk, G., Peets, K. F., & Yang, S. (2010). Receptive vocabulary differences in monolingual and bilingual children. Bilingualism: Language and Cognition, 13 (4), 525531. DOI: https://doi.org/10.1017/S1366728909990423 CrossRefGoogle ScholarPubMed
Byrd, D., Arentoft, A., Scheiner, D., Westerveld, M., & Baron, I. S. (2008). State of multicultural neuropsychological assessment in children: Current research issues. Neuropsychology Review, 18 (3), 214222. DOI: https://doi.org/10.1007/s11065-008-9065-y CrossRefGoogle ScholarPubMed
Cianci, L., Orsini, A., Hulbert, S., & Pezzuti, L. (2013). The influence of parents' education in the Italian standardization sample of the WISC-III. Learning and Individual Differences, 28, 4753. DOI: https://doi.org/10.1016/j.lindif.2013.09.009 CrossRefGoogle Scholar
Duggan, E. C., Awakon, L. M., Loaiza, C. C., & Garcia-Barrera, M. A. (2019). Contributing towards a cultural neuropsychology assessment decision-making framework: Comparison of WAIS-IV norms from Colombia, Chile, Mexico, Spain, United States, and Canada. Archives of Clinical Neuro-psychology, 34 (5), 657681. DOI: https://doi.org/10.1093/arclin/acy074 CrossRefGoogle ScholarPubMed
Eilertsen, T., Thorsen, A. L., Holm, S. E. H., Bøe, T., Sørensen, L., & Lundervold, A. J. (2016). Parental socioeconomic status and child intellectual functioning in a Norwegian sample. Scandinavian Journal of Psychology, 57 (5), 399405. DOI: https://doi.org/10.1111/sjop.12324 CrossRefGoogle Scholar
Eizaguirre, M. B., Vanotti, S. I., Aguayo Arelis, A., Rabago Barajas, B., Cores, E. V., Macías, M. A., Benedict, R. H. B., & Cáceres, F. (2020). Symbol digit modalities test-oral version: An analysis of culture influence on a processing speed test in Argentina, Mexico, and the USA. Developmental Neuropsychology, 45 (3), 129138. DOI: https://doi.org/10.1080/87565641.2020.1737699 CrossRefGoogle ScholarPubMed
Eriksen, H. L. F., Kesmodel, U. S., Underbjerg, M., Kilburn, T. R., Bertrand, J., & Mortensen, E. L. (2013). Predictors of intelligence at the age of 5: Family, pregnancy and birth characteristics, postnatal influences, and postnatal growth. PLoS One, 8 (11), 18. DOI: https://doi.org/10.1371/journal.pone.0079200 CrossRefGoogle ScholarPubMed
Eurostat (2022). Population by educational attainment level, sex and age (%) - main indicators [Dataset]. Retrieved from https://ec.europa.eu/eurostat/databrowser/view/EDAT_LFSE_03__custom_3991061/default/table Google Scholar
Friedrich, S., Konietschke, F., & Pauly, M. (2019). Resampling-based analysis of multivariate data and repeated measures designs with the R package MANOVA.RM. The R Journal, 11 (2), 380. DOI: https://doi.org/10.32614/RJ-2019-051 CrossRefGoogle Scholar
Garratt, L. C., & Kelly, T. P. (2007). To what extent does bilingualism affect children’s performance on the NEPSY? Child Neuropsychology, 14 (1), 7181. DOI: https://doi.org/10.1080/09297040701218405 CrossRefGoogle Scholar
Georgas, J., Van de Vijver, F. J. R., Weiss, L. G., & Saklofske, D. H. (2003). A cross-cultural analysis of the WISC-III. In Georgas, J. (Ed.), Culture and children’s intelligence: Cross-cultural analysis of the WISC-III (pp. 277313). Academic Press.CrossRefGoogle Scholar
Giofrè, D., Allen, K., Toffalini, E., & Caviola, S. (2022). The impasse on gender differences in intelligence: A meta-analysis on WISC batteries. Educational Psychology Review, 34 (4), 25432568. DOI: https://doi.org/10.1007/s10648-022-09705-1 CrossRefGoogle Scholar
Grob, A., Petermann, F., Lipsius, M., Costan-Dorigon, J., Petermann, U., & Daseking, M. (2008). Differences in Swiss and German children’s intelligence as measured by the HAWIK-IV. Swiss Journal of Psychology, 67 (2), 113118. DOI: https://doi.org/10.1024/1421-0185.67.2.113 Google Scholar
International Test Commission (2019). The ITC guidelines for translating and adapting tests (2nd ed.). Retrieved from www.InTestCom.org Google Scholar
Korkman, M., Kemp, S., & Kirk, U. (2001). Effects of age on neurocognitive measures of children ages 5 to 12: A cross-sectional study on 800 children from the United States. Developmental Neuropsychology, 20 (1), 331354. DOI: https://doi.org/10.1207/S15326942DN2001_2 CrossRefGoogle ScholarPubMed
Korkman, M., Lahti-Nuuttila, P., Laasonen, M., Kemp, S. L., & Holdnack, J. (2013). Neurocognitive development in 5- to 16-year-old North American children: A cross-sectional study. Child Neuropsychology, 19 (5), 516539. DOI: https://doi.org/10.1080/09297049.2012.705822 CrossRefGoogle ScholarPubMed
Korkman, M., Stenroos, M., Mickos, A., Westman, M., Ekholm, P., & Byring, R. (2012). Does simultaneous bilingualism aggravate children’s specific language problems? Acta Paediatrica, 101 (9), 946952. DOI: https://doi.org/10.1111/j.1651-2227.2012.02733.x CrossRefGoogle ScholarPubMed
Korpinen, E., Slama, S., Rosenqvist, J., & Haavisto, A. (2023). WPPSI-IV and NEPSY-II performance in mono- and bilingual 5-6-year-old children: Findings from The FinSwed Study. Scandinavian Journal of Psychology. DOI: https://doi.org/10.1111/sjop.12895 CrossRefGoogle ScholarPubMed
Lauchlan, F., Parisi, M., & Fadda, R. (2012). Bilingualism in Sardinia and Scotland: Exploring the cognitive benefits of speaking a ‘minority’ language. International Journal of Bilingualism, 17 (1), 4356. DOI: https://doi.org/10.1177/1367006911429622 CrossRefGoogle Scholar
Leino, K., Ahonen, A. K., Hienonen, N., Hiltunen, J., Lintuvuori, M., Lähteinen, S., Lämsä, J., Nissinen, K., Nissinen, V., Puhakka, E., Pulkkinen, J., Rautopuro, J., Sirén, M., Vainikainen, M.-P., & Vettenranta, J. (2019). PISA 18: ensituloksia. Suomi parhaiden joukossa [PISA 18: primary results. Finland among the best]. Retrieved from http://urn.fi/URN:ISBN:978-952-263-678-2 Google Scholar
Millard, S. P. (2013). EnvStats: An R package for environmental statistics (2nd ed.). Springer.CrossRefGoogle Scholar
Miller, J. L., Weiss, L. G., Beal, A. L., Saklofske, D. H., Zhu, J., & Holdnack, J. A. (2015). Intelligent use of intelligence tests: Empirical and clinical support for Canadian WAIS-IV norms. Journal of Psychoeducational Assessment, 33 (4), 312328. DOI: https://doi.org/10.1177/0734282915578577 CrossRefGoogle Scholar
O’sullivan, F., Yandell, B. S., & Raynor, W. J. (1986). Automatic smoothing of regression functions in generalized linear models. Journal of the American Statistical Association, 81 (393), 96103.CrossRefGoogle Scholar
Official Statistics of Finland (2017). Educational structure of population [e-publication]. Retrieved from http://www.stat.fi/til/vkour/index_en.html Google Scholar
Official Statistics of Finland (2021). Key figures on population by region, 1990-2021 [e-publication]. Retrieved from https://pxdata.stat.fi/PxWeb/pxweb/en/StatFin/StatFin__vaerak/statfin_vaerak_pxt_11ra.px/table/tableViewLayout1/ Google Scholar
Olson, K., & Jacobson, K. (2015). Cross-cultural considerations in pediatric neuropsychology: A review and call to attention. Applied Neuropsychology: Child, 4 (3), 166177. DOI: https://doi.org/10.1080/21622965.2013.830258 CrossRefGoogle ScholarPubMed
Palejwala, M. H., & Fine, J. G. (2015). Gender differences in latent cognitive abilities in children aged 2 to 7. Intelligence, 48, 96108. DOI: https://doi.org/10.1016/j.intell.2014.11.004 CrossRefGoogle Scholar
Park, H., & Lee, K. (2016). Intellectual characteristics of Korean children: Analyses of K-WPPSI-IV performance. Korean Journal of Child Studies, 37 (6), 157168. DOI: https://doi.org/10.5723/kjcs.2016.37.6.157 CrossRefGoogle Scholar
Pérez-García, M., Luna, J. D. D., Torres-Espínola, F. J., Martínez-Zaldívar, C., Anjos, T., Steenweg-de Graaff, J., Weber, M., Grote, V., Gruszfeld, D., Verduci, E., Poncelet, P., Escribano, J. D., Tiemeier, H., Koletzko, & B., Campoy, C. (2019). Cultural effects on neurodevelopmental testing in children from six European countries: An analysis of NUTRIMENTHE Global Database. British Journal of Nutrition, 122 (S1), S59S67. DOI: https://doi.org/10.1017/S0007114517000824 CrossRefGoogle ScholarPubMed
Piotrowski, C. (2017). Neuropsychological testing in professional psychology specialties: Summary findings of 36 studies (1990-2016) in applied settings. Journal of the Indian Academy of Applied Psychology, 43 (1), 134144.Google Scholar
Rabin, L. A., Paolillo, E., & Barr, W. B. (2016). Stability in test-usage practices of clinical neuropsychologists in the United States and Canada over a 10-year period: A follow-up survey of INS and NAN members. Archives of Clinical Neuropsychology, 31 (3), 206230. DOI: https://doi.org/10.1093/arclin/acw007 CrossRefGoogle Scholar
Raiford, S. E., Coalson, D. L., & Engi, M. D. (2014). WPPSI-IV score differences across demographic groups. In Raiford, S. E., & Coalson, D. L. (Eds.), Essentials of WPPSI-IV assessment (pp. 215236). John Wiley & Sons, Inc.Google Scholar
Rindermann, H., Flores-Mendoza, C., & Mansur-Alves, M. (2010). Reciprocal effects between fluid and crystallized intelligence and their dependence on parents' socioeconomic status and education. Learning and Individual Differences, 20 (5), 544548. DOI: https://doi.org/10.1016/j.lindif.2010.07.002 CrossRefGoogle Scholar
Roivainen, E. (2010). European and American WAIS III norms: Cross-national differences in performance subtest scores. Intelligence, 38 (1), 187192. DOI: https://doi.org/10.1016/j.intell.2009.10.001 CrossRefGoogle Scholar
Roivainen, E. (2013). Are cross-national differences in IQ profiles stable? A comparison of Finnish and U.S. WAIS norms. International Journal of Testing, 13 (2), 140151. DOI: https://doi.org/10.1080/15305058.2012.657315 CrossRefGoogle Scholar
Roivainen, E. (2019). European and American WAIS IV norms: Cross-national differences in perceptual reasoning, processing speed and working memory subtest scores. Scandinavian Journal of Psychology, 60 (6), 513519. DOI: https://doi.org/10.1111/sjop.12581 CrossRefGoogle ScholarPubMed
Rosenqvist, J., Lahti-Nuuttila, P., Urgesi, C., Holdnack, J., Kemp, S. L., & Laasonen, M. (2017). Neurocognitive functions in 3- to 15-year-old children: An international comparison. Journal of the International Neuropsychological Society, 23 (4), 367380. DOI: https://doi.org/10.1017/S1355617716001193 CrossRefGoogle ScholarPubMed
Rosenqvist, J., Slama, S., & Haavisto, A. (2022). Användningen av kognitiva test vid psykologiska utredningar av barn och unga på svenska i Finland – en översikt [Using cognitive tests for psychological evaluations of children in Swedish in Finland - an overview]. NMI Bulletin, Svenskspråkigt specialnummer, 32, 7493.Google Scholar
Rosenqvist, J., Slama, S., & Haavisto, A. (2023). Generalizability of the Swedish WISC-V to the Finland-Swedish minority – The FinSwed Study [Manuscript submitted for publication].Google Scholar
Rosselli, M., & Ardila, A. (2003). The impact of culture and education on non-verbal neuropsychological measurements: A critical review. Brain and Cognition, 52 (3), 326333. DOI: https://doi.org/10.1016/S0278-2626(03)00170-2 CrossRefGoogle ScholarPubMed
Rosselli, M., Ardila, A., Navarrete, M. G., & Matute, E. (2010). Performance of Spanish/English bilingual children on a spanish-language neuro-psychological battery: Preliminary normative data. Archives of Clinical Neuropsychology, 25 (3), 218235. DOI: https://doi.org/10.1093/arclin/acq012 CrossRefGoogle Scholar
Saarela, J. (2021). Finlandssvenskarna 2021 – en statistisk rapport [The Finland-Swedes 2021 - a statistical report]. Retrieved from https://folktinget.fi/Site/Data/1597/Files/Finlandssvenskarna%202021_statistisk%20rapport_Folktinget_KLAR.pdf Google Scholar
Schleicher, A. (2019). PISA 2018: Insights and interpretations. Retrieved from https://www.oecd.org/pisa/PISA%202018%20Insights%20and%20Interpretations%20FINAL%20PDF.pdf Google Scholar
Thordardottir, E. (2011). The relationship between bilingual exposure and vocabulary development. International Journal of Bilingualism, 15 (4), 426445. DOI: https://doi.org/10.1177/1367006911403202 CrossRefGoogle Scholar
Tsujimoto, S. (2008). The prefrontal cortex: Functional neural development during early childhood. The Neuroscientist, 14 (4), 345358. DOI: https://doi.org/10.1177/1073858408316002 CrossRefGoogle ScholarPubMed
Walter, F., Daseking, M., & Pauls, F. (2020). Sex differences in intelligence in children aged 2: 6-7: 7: Analysis of the factor structure and measurement invariance of the German Wechsler Primary and Preschool Scale of Intelligence-Fourth Edition. Journal of Psychoeducational Assessment, 39 (4), 395421. DOI: https://doi.org/10.1177/0734282920981398 CrossRefGoogle Scholar
Wechsler, D. (2012). WPPSI-IV: Technical and interpretive manual. NCS Pearson, Incorporated.Google Scholar
Wechsler, D. (2013). L'échelle d'intelligence de Wechsler pour la période préscolaire et primaire—Quatrième édition—Version pour francophones du Canada [WPPSI-IV for French-speaking Canadians]. Pearson France-ECPA.Google Scholar
Wechsler, D. (2014). Wechsler Preschool and Primary Scale of Intelligence – Fourth Edition. Manual Del 1. Svensk version. [Manual Part 1. Swedish version]. NCS Pearson, Incorporated.Google Scholar
Westman, M., Korkman, M., Mickos, A., & Byring, R. (2008). Language profiles of monolingual and bilingual Finnish preschool children at risk for language impairment. International Journal of Language & Communication Disorders, 43 (6), 699711. DOI: https://doi.org/10.1080/13682820701839200 CrossRefGoogle ScholarPubMed
Wood, S. N. (2017). Generalized additive models: An introduction with R. CRC.CrossRefGoogle Scholar
Wood, S. N., & Augustin, N. H. (2002). GAMs with integrated model selection using penalized regression splines and applications to environmental modelling. Ecological modelling, 157 (2-3), 157177. DOI: https://doi.org/10.1016/S0304-3800(02)00193-X CrossRefGoogle Scholar
Figure 0

Figure 1. Flow chart of the exclusion process. Note.1Four families withdrew after initial consent due to the COVID-19 pandemic.

Figure 1

Table 1. Demographic characteristics of the sample (N = 79)

Figure 2

Table 2. Means, SD, and ranges for the subtests and indexes in the Finland-Swedish sample (N = 79)

Figure 3

Figure 2. Means and bootstrapped confidence intervals (standard errors presented in parenthesis) of WPPSI-IV subtest scores compared to the Scandinavian norms (Mean = 10).

Figure 4

Figure 3. Means and bootstrapped confidence intervals (standard errors presented in parenthesis) of WPPSI-IV FSIQ and index scores compared to the Scandinavian norms (Mean = 100). Note. FSIQ: Full Scale IQ; VCI: Verbal Comprehension Index; VSI: Visual Spatial Index; FRI: Fluid Reasoning Index; WMI: Working Memory Index; PSI: Processing Speed Index; VAI: Vocabulary Acquisition Index.

Figure 5

Figure 4. Generalized additive model showing age effects on the PSI (N = 79).

Figure 6

Table 3. Multiple linear regression analyses exploring the relationships between sociodemographic variables and performance on WPPSI-IV FSIQ and indexes (N = 79)

Figure 7

Figure 5. Means and SD in the FSIQ and indexes according to parental education level (N = 79). Note. PED: Parental education level; Level 1: Upper secondary education, vocational education, or lower; Level 2: Bachelor’s degree or University of applied sciences degree; Level 3: Master’s or Doctoral degree. FSIQ: Full Scale IQ; VCI: Verbal Comprehension Index; VSI: Visual Spatial Index; FRI: Fluid Reasoning Index; WMI: Working Memory Index; PSI: Processing Speed Index; VAI: Vocabulary Acquisition Index.