The aim of this study was to validate a teacher rating scale (TRS-EN) for assessing early numeracy skills and for identifying mathematically low-performing 3- and 4-year-old children. Fifty-six preschool teachers assessed children's (3-year-olds = 127; 4-year-olds = 145) numerical relational, counting, and basic arithmetic skills. Construct validity evidence through confirmatory factor analysis supported both unidimensional and three-factor (i.e. numerical relational, counting and basic arithmetic skills) models. Also, TRS-EN demonstrated excellent internal consistency in both age groups. The teacher assessment scores corresponded well with standardized direct early numeracy measures, and predicted children's early numeracy skills six months later. The results clearly demonstrate that the TRS-EN can be used to assist teachers in assessing early numeracy skills, and help with identifying mathematically low-performing children.
Keywords: Assessment; early childhood education; early numeracy; low-performing; psychometrics; teacher rating scale
Early numeracy (EN) skills are crucial for later academic performance. Individual differences in EN skills are visible already during the early years and these differences grow even larger throughout the years in education (Zhang et al. [
Core EN skills refer to a group of numerical skills that are considered to be foundational for later mathematical development (Aunio and Räsänen [
Symbolic and non-symbolic number sense refers to the ability to recognize differences in numerosities (i.e. approximate number system; vanMarle [
Counting skills emerge around the age of 2, when children start to produce small number words and connect them into number word sequences (i.e. verbal counting), first by mixing the order (Gelman and Gallistel [
Children must also understand how different numerals and quantities are related to one another (Wynn [
The knowledge of counting and numerical relations builds up into the knowledge of basic arithmetical skills (Starkey, Klein, and Wakeley [
All aforementioned EN skills are known to predict later mathematics learning and furthermore, precede the understanding of more complex mathematical skills (Chu et al. [
Direct assessment (i.e. standardized interview-based tests) requires thorough delving into the material, practicing and time for administering and scoring (Reid et al. [
Previous research on teacher rating scales on K-12 children's academic skills (i.e. language and mathematics) has demonstrated small to moderate correlations between direct assessments (i.e. mean correlation.62, range -.03 to.80; Südkamp, Kaiser, and Möller [
Prior studies have had several limitations such as investigating the validity of the teacher rating scale on a relatively narrow population (e.g. low-income sample; Furnari et al. [
By validating a teacher rating scale for assessing preschoolers' EN skills (TRS-EN), this study aimed to address the need for a valid and reliable but also an easy-to-use assessment tool for teachers to evaluate preschooler's EN skills and to identify low-performing children. Items for TRS-EN were developed in a group of specialists in special education and educational assessment from one Finnish university. Prior research on numeracy assessment tools (VanLuit, Van de Rijt, and Aunio [
Previous teacher rating scales have included a variety of early mathematical skills (see Appendix 1). As TRS-EN was also aimed to identify low-performing children, the items should reflect skills that best predict later mathematical performance (i.e. core EN skills). As the framework of Aunio and Räsänen ([
Table 1. Existing Direct EN Measures Covering Age Groups 3–4.
Measure Acronym Authors Target ages Measured early mathematical skills Research-Based Early Maths Assessment REMA Clements, Sarama, and Liu 3- to 7-year-olds Comparing, ordering, counting, arithmetic, number recognition, subitizing, composing number, geometry, identifying shapes, measuring and matterning Early Arithmetic, Reading and Learning Indicators EARLI DiPerna, Morgan, and Lei 3- to 4-year-olds Counting aloud, counting objects, measurement, numbers and shapes, grouping and pattern recognition Preschool Numeracy Indicators PNI Floyd, Hojnoski, and Key 3- to 6-year-olds One-to-one correspondence, oral counting, number naming, quantity comparison Test of Early Mathematics Ability 3rd Edition TEMA Ginsburg and Baroody 3- to 8-year-olds Counting, number facts and arithmetic Preschool Early Numeracy Scales PENS Purpura 3- to 5-year-olds Counting, numerical relations and arithmetic The Child Math Assessment CMA Starkey, Klein, and Wakeley 3- to 5-year-olds Numerical knowledge, arithmetic, space/geometry and patterning/logical reasoning The Early Numeracy Test ENT VanLuit, Van de Rijt, and Aunio 3- to 7-year-olds Counting, understanding of numerical relations
Counting skills, understanding of numerical relations, basic arithmetic skills seemed to form the most important EN skills in 3- to 4-year-olds. This finding was contrasted to the results of longitudinal studies investigating which EN skills are the strongest predictors of later mathematical performance in 3- to 4-year-old children. The reviewed longitudinal studies that covered both 3- and 4-year-olds suggested that counting (Chu et al. [
Based on these findings, 22 items concerning counting (13 items), numerical relational (7 items) and basic arithmetic skills (2 items) were formed. Before collecting the main data, two preschool teachers were asked to give feedback on TRS-EN. The two preschool teachers were asked whether the scale included all necessary aspects of 3- to 4-year-old's EN development and whether they thought that the items were comprehensible and could be relatively easily assessed if the child was familiar to the teacher. As these preschool teachers did not point out any details that would have to be changed, the original TRS-EN was used for the current sample.
The aim of this study was to validate a teacher rating scale for assessing EN skills and identifying low-performing 3- to 4-year-old children. Specifically, this study aimed to (
Individual differences in early numeracy start emerging already at preschool age and only widen throughout the school years. To provide support as early as possible, it is crucial to identify these low-performing children. This study contributes to the understanding of whether TRS-EN is a valid and reliable tool for teachers to assess children's EN skills and identify low-performing children. Furthermore, this study adds novel information concerning the ability of teacher rating scales to predict later EN performance. Also, this study adds knowledge of the capacity of TRS-EN to differentiate specific EN sub-skills among 3- and 4-year-old children. Results are examined independently for both age groups as preschool-aged children develop rapidly and present large individual differences (Litkowski et al. [
This study was a part of the Active Early Numeracy project, investigating the associations between preschooler's motor skills, physical activity and academic skills. In total, 16 preschools participated in the project in 2020. A research permit was obtained from one large municipality, preschools as well as from the ethical board of the university. The municipality suggested preschools from diverse demographic areas in Southern-Finland. The participants were then recruited from the preschools that permitted the research. Permission forms were sent to children's parents specifying the project's aims and procedure. Parents were asked to discuss with their children whether they were willing to participate. Children were able to withdraw from participating at any time if they or their parents expressed such desire. Direct assessments were not conducted if the child seemed too tired or could not concentrate during that time.
Preschool teachers who were involved in the project were asked whether they thought that they were able to fill in teacher rating scales at the time being. Thus, participants in this study were 56 preschool teachers and 273 children (3-year-olds: n = 127; M
Table 2. Socioeconomic Background of the Participants.
Characteristic 3-year-olds 4-year-olds Children Boys 60 69 Girls 67 76 Total 127 145 Age 3-year-olds 4-year-olds Total 127 146 Home language Finnish 83 98 Other 12 13 Total 95 111 Parents' educational level Graduate degree 55 58 Undergraduate degree 19 24 Upper secondary or vocational school 19 27 Comprehensive school or none 2 3 Total 95 112 Parents' employment status Full-time 86 101 Part-time 4 6 Student 4 1 Unemployed 0 0 Total 94 108 Mean net income (€) per year Above 80 000 22 24 60 000–79 999 25 20 40 000–59 999 22 29 20 000–39 999 9 27 Below 19 999 13 8 Total 91 108
TRS-EN included 22 items measuring counting (13 items) for example 'is able to count forward from 1 to 10 correctly', numerical relational (7 items) for example 'is able to order objects according to a given criteria (e.g. size: from the biggest to the smallest)', and basic arithmetical skills (2 items) for example 'is able to add and subtract with numbers below 10 using objects (e.g. When playing, notices that there are six toy animals in total when there are 3 cats and 3 dogs)'. Counting and numerical relational skills were emphasized in test development (see Table 1). Two items concerning basic arithmetic skills were included to cover for large individual differences in EN skills and to avoid ceiling effects. Every item was assessed on a 3-point scale with the options being: never, often and always. The teacher's were also asked to fill in the preschool's name, playgroup, teacher's name, date and how long the teacher had worked with the child.
The Early Numeracy Test (ENT; VanLuit, Van de Rijt, and Aunio [
Table 3. Descriptive Statistics and Internal Consistency for the Assessments.
Concurrent Subsequent Variable 3-year-olds 4-year-olds 3-year-olds 4-year-olds Mean SD Mean SD Mean SD Mean SD TRS-EN 102 17.51 7.78 .915 114 26.72 10.19 .954 37 28.38 8.03 .915 32 33.97 6.12 .954 TRS-EN Counting 106 10.96 5.06 .871 124 16.72 6.26 .927 40 17.05 5.03 .871 38 29.47 3.73 .937 TRS-EN Relational 117 6.01 2.58 .784 133 8.78 3.18 .885 44 9.86 2.23 .784 43 10.86 1.71 .885 TRS-EN Arithmetic 119 0.55 0.92 .714 133 1.68 1.34 .699 49 1.45 1.24 .714 40 2.25 1.33 .699 ENT 131 14.15 7.36 .891 61 14.18 6.41 .816 63 20.44 7.38 .891 ENT Counting 126 4.25 3.73 .827 61 3.74 3.25 .712 63 7.22 4.36 .827 ENT Relational 109 6.09 3.34 .743 129 10.14 4.31 .829 61 10.44 3.99 .743 63 13.22 3.90 .829
Participating children's parents were asked to report their employment status (i.e. unemployed, student, part-time and full-time), educational level (i.e. comprehensive school or none, upper secondary or vocational school, bachelor's degree and graduate degree), home language (i.e. Finnish or other) and income level (i.e. below 19 999€/year, 20 000–39 999 €/year; 40 000–59 999€/year; 60 000–79 999€/year; above 80 000 €/year) on a form that was sent to children's homes. Family educational level and employment status was determined by using the higher education and employment status if two distinct levels of statuses were reported. Parent's higher educational level and employment status was coded to indicate family educational level and employment status.
The scales were organized into envelopes according to playgroups. Two members of the research group delivered and collected the envelopes in June 2020 as it was not possible to visit preschools from March to July 2020 due to the COVID-19 outbreak. Children's names were written on the scales beforehand. A cover letter including instructions for how to fill in the scale was also handed out to the teachers. The teachers were asked to fill in the scales based on their experience of working with the child. Completing TRS-EN took approximately 10 to 15 minutes per child. Teachers were encouraged to contact the research group if they had any questions concerning the scale.
Direct assessments were administered individually in a quiet location in the children's own preschools during regular preschool hours. All of the assessments were administered by trained research assistants from November 2019 to March 2020 and for predictive validity from November 2020 to June 2021.
First, the distribution and outliers of the data were examined and it was confirmed that there were no significant outliers in the data. To examine the construct validity of TRS-EN, confirmatory factor analysis (CFA) was conducted by using the weighted least square mean and variance adjusted (WLSMV) estimator in Mplus version 8, to consider the non-normally distributed items in the data. Cluster sampling due to teacher evaluation was considered by using the type = complex command, which takes into account the non-independence of observations (Muthén and Muthén BO [
Criterion-related validity was measured in terms of concurrent and predictive validity. Concurrent validity was measured using correlation analyses between TRS-EN and direct assessments. Correlation coefficients below.40 are considered small, above.40 moderate and above.60 strong (Evans [
In order to test whether TRS-EN would identify the same low-performing children as the direct EN measure, a configural frequency analysis (von Eye [
By means of multivariate regression, it was tested whether teacher ratings were affected by the socioeconomic status (SES) of the family (i.e. parents' education and employment status, first language and income level; see Table 2). Maximum likelihood estimation with robust standard errors (MLR) in Mplus version 8 was applied due to skewed SES variables. Family SES variables and children's direct EN measure scores were set as independent variables and TRS-EN scores as dependent variables. Independent variables were allowed to correlate by using with-statements.
Descriptive statistics as well as Cronbach's alphas for TRS-EN and direct EN measure concurrently and subsequently are presented in Table 3. As expected, 4-year-olds presented higher mean scores in the TRS-EN overall, as well as in all sub-skills, than 3-year-olds.
Items that were developed for the TRS-EN were hypothesized to represent a three-factor model (i.e. counting, numerical relational, and basic arithmetic skills) which fitted the data well (see Table 4). Factor loadings were significant (p <.001) and above.561 among the 3-year-olds and above.669 among the 4-year-olds' (see Appendix 3). However, in both age groups, the latent factors were highly correlated and, therefore, a unidimensional model was also tested, which resulted in a good model fit (see Table 4). In the one-factor model, the factor loadings were significant (p <.001) and above.535 for 3-year-olds and above.664 for 4-year-olds (Appendix 3). Thus, both the one-factor and three-factor models presented an acceptable model fit. The significance of the differences between the two models was tested with the DIFFTEST-option in Mplus. Results revealed that the three-factor model fitted the data significantly better compared to the one-factor model in both 3-year-olds, DIFFTEST (
Table 4. Model Fit Indices for One- and Three-factor Models for Both Age Groups.
χ2 CFI TLI RMSEA 3-year-olds Three-factors 257.856 167 .946 .939 .065 <.001 One-factor 276.981 170 .937 .929 .070 <.001 4-year-olds Three-factors 228.106 167 .990 .989 .050 .0012 One-factor 238.893 170 .989 .987 .053 <.001
In both models, one item (item 6), measuring backwards rote counting from 20 to 1 was deleted due to its difficulty (M = 0.16, SD = 0.456) and one item (item 14) measuring children's ability to classify objects according to different principles due to its easiness (M = 1.73, SD = 0.460).
Cronbach's alphas for both factor solutions of the TRS-EN are presented in Table 3. Excellent internal consistency was obtained for the TRS-EN in both age groups. Counting skills factor presented the highest internal consistency and basic arithmetical skills the lowest.
Correlation coefficients between TRS-EN and the direct EN measure are presented in Appendix 2. TRS-EN in total exhibited moderate to strong correlations with the direct EN measure. Moderate correlations were also obtained for 4-year-olds between the TRS-EN counting and numerical relational skills factor and the corresponding factors in the direct EN measure. Also, 3-year-olds' TRS-EN numerical relational skills factor correlated moderately with the corresponding factor of the direct EN measure.
To test whether the TRS-EN would identify the same low-performing children as the direct EN measure, the children were grouped into low-performing and typically performing children in both tests using the 20th percentile as a cutoff. This grouping resulted in four configurations (typical/typical; typical/low; low/typical; and low/low). By means of configural frequency analysis, the observed frequencies were compared to the expected frequencies (see Table 5).
Table 5. Change and Stability in Performance Groups Across TRS-EN and Direct EN Measure.
Configuration (TRS-EN – ENT) Observed frequency Expected frequency z-score 3-year-olds Typical – Typical 60 40.54 3.71 .0002 Type Typical – Low 6 9.35 −1.14 .2547 Low – Typical 18 18.43 −0.11 .9148 Low – Low 12 4.25 3.82 .0001 Type 4-year-olds Typical – Typical 71 46.60 4.33 .0001 Type Typical – Low 10 10.54 −0.17 .8627 Low – Typical 13 12.66 0.10 .9198 Low – Low 9 2.86 3.66 .0002 Type
1 Note: ENT = Early Numeracy Test. For 3-year-olds, numerical relational skills sub-scale was used to categorize children into low-performing and typically performing groups.
For 3-year-olds, the configural frequency analysis identified two stable configurations and two non-significant configurations. Of the 18 children who were identified as low-performing with TRS-EN, 12 (66%) were also identified as low-performing with the direct EN measure. Also, of the 78 children who were identified as typically performing with TRS-EN, 60 (77%) were also identified as typically performing with the direct EN measure. These configurations were occurring more than expected by chance, indicating stability in the classification of low-performing and typically performing status.
For 4-year-olds the configural frequency analysis identified two stable configurations and two non-significant configurations. Of the 19 children who were identified as low-performing with TRS-EN, 9 (47%) were also identified as low-performing with the direct EN measure. Also, of the 84 children who were identified as typically performing with TRS-EN, 71 (85%) were also identified as typically performing with the direct EN measure.
Descriptive statistics for predictive validity are presented in Table 3. Repeated measures ANOVA indicated that TRS-EN scores increased from the first measurement to the second among 3-year-olds, F(
Also, among 4-year-olds teacher ratings in the first time point predicted significantly both direct EN measure scores (β = 625, p <.001) and teacher ratings (β = 715, p <.001) in the second time point. Teacher ratings in the first time point explained a significant amount of variance in the second time point direct EN measure scores (R
To check whether teacher ratings may be affected by family SES, a multivariate regression analysis was applied, where both family SES and direct EN skills were included as independent variables. Both models were saturated, as the number of observed variables was equal to the number of the estimated parameters. To examine model fit indices, we ran the model for both age groups with only numerical relational skills (3-year-olds) or direct EN measure scores (4-year-olds) predicting teacher ratings. Model fit indices were excellent for both, 3-year-olds, CFI = 1.00, TLI = 1.00 and RMSEA <.001 and 4-year-olds, CFI =.977, TLI =.913 and RMSEA =.075. Results confirmed (see Table 6) that the direct EN measure was the only significant predictor of teacher ratings at the first time point, indicating that teacher ratings were affected by the child's actual EN skills measured by the direct measure, rather than the family SES indicators. The model explained 36% of the variance in 3-year-olds and 44% of 4-year-olds teacher ratings. ENT correlated significantly with all other SES variables than parents' employment status.
Table 6. Summary of ENT and SES variables predicting teacher ratings.
R2 β 3-year-olds .357 .002 Numerical relational skills .549 .074 <.001 First language -.124 .096 .198 Parents' employment .080 .090 .371 Parents' education .088 .111 .426 Income level -.151 .122 .217 4-year-olds .436 <.001 ENT .568 .072 <.001 First language .174 .101 .084 Parents' employment -.002 .106 .984 Parents' education .178 .105 .089 Income level .121 .095 .204
2 Note: ENT = Early Numeracy Test.
The aim of this study was to validate a teacher rating scale for assessing 3- and 4-year-old children's EN skills and to be able to identify mathematically low-performing children. The results indicated that the TRS-EN can be used to measure EN skills both as a composite, and also, as three separate constructs measuring counting, numerical relational, and basic arithmetic skills. Although the three-factor solution fitted the data slightly better for both 3- and 4-year-old children, the TRS-EN may well be used to measure children's EN skills as a composite as well, as it allows preschool teachers to efficiently evaluate children's overall skills contrasted to other children. However, by considering the three factors separately, the TRS-EN offers teachers the possibility to examine sub-skills of EN in greater detail to target individualized support at a specific sub-skill area. As prior studies have not examined EN teacher assessments separately for different age groups, the results of this study add novel information on how teacher rating scales can be appropriately used for assessing 3- and 4-year-olds EN skills.
The TRS-EN presented excellent internal consistency parallel to prior findings (Baker et al. [
Further, the results also revealed that teacher ratings corresponded well with the direct EN scores and predicted children's later EN skills, confirming that the TRS-EN scores reflect the ones of the widely used EN measure (see AERA, APA, and NCME [
Prior studies from the US have suggested that teachers may have challenges in identifying individual differences in early numeracy (Kilday et al. [
Despite the promising results, our study has some limitations. One limitation concerning concurrent validity is that teachers assessed children's academic skills a few months after the direct assessments were administered. This may have led to small differences between the results as children develop rapidly at this age (Litkowski et al. [
Future studies should consider the possibility of using teacher rating scales as assessment tools for assessing intervention effects. For example, in single case studies where multiple baseline assessments are recommended, teacher rating scales could provide a cost- and time-efficient instrument for monitoring children's progress and intervention effects.
As concluding remarks, the results of this study suggest that the TRS-EN is a viable tool for teachers to evaluate children's early numeracy skills and identifying low-performing children. As individual differences in early numeracy become visible during the early years and only widen throughout the school years, TRS-EN can provide information on which EN skills should be further addressed in the instruction. Furthermore, TRS-EN can help to identify low-performing children at an early stage.
The authors would like to thank all contributors involved in collecting the data for the current manuscript. In addition, the authors thank all participants in the research group's seminars for their unwavering support through commenting.
No potential conflict of interest was reported by the author(s).
Summary of Previous Studies on Preschoolers' EN Skills Teacher Rating Scales.
Author(s) Sample Teacher Rating Scale Direct Assessment Concurrent validity (Teacher rating scale and direct assessment) Other results Limitations Baker et al. 760, 3- to 5-year-olds from low-income families in the US Seven item mathematics subscale from Academic Rating Scale (5-point Likert). Applied Problems subtest from Woodcock-Johnson III Tests of Achievement Correlation between the rating scale and direct assessment Excellent internal consistency α =.93. Inadequate concurrent measure and a narrow target group. Furnari et al. 435, 3- to 5-year-olds from low-income families in the US Seven items from the academic rating scale – mathematics (ARS-M), 5 items developed, in total 13 items (5-point scale). TEMA-3 for number sense and operations, geometry and measurement from TEAM. Moderate concordance with the concurrent measurement ( Internal consistency α =.97. Only 25% of variance was explained by the child's achievement. Narrow low-income target group Kilday et al. 318, 3.5- to 5-year-olds from the US Modified 12 item version of Academic Rating Scale (number sense, geometry and measurement; 5-point likert scale). Number sense using TEMA-3 and geometry and measurement by a modified version of TEAM Regression coefficients between Internal consistency on number sense subscale α =.92 and for geometry & measurement α =.89. Divided the scale in subscales without using statistical modeling. Kowalski et al. 122, 3- to 6-year-olds from the US The Galileo system: 68 items (dichotomous) computer-based scale for math skills. Twenty-two items developed based on the Galileo. Correlation with direct assessment Small but representative sample. Lambert, Kim, and Burts 1241, 3- to 4-year-olds from the US Teaching strategies GOLD: 7 numeracy skills items (10-point scale). Bracken School Readiness Assessment Correlation with direct assessment Internal consistency for mathematics α =.97 No information about family SES. Reid et al. 440, 3- to 5-year-olds from low-SES families in the US TRS-EAC: 5 numeracy items (5-point Likert). TEMA-3 Correlations between Internal consistency α =.97 for the early academic skills. Predictive validity for 45 children expressed correlations of.61 for numeracy. Narrow low SES target group. Predictive validity investigated for a small sub sample.
3 Note. SES = Socioeconomic status.
Correlations Between EN Assessments for 3- and 4-year-old Children Separately.
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. Concurrent 1. TRS-EN .982** .955** .741** .595** .553** .525** .709** .662** .534** .587** .625** .562** .556** 2. TRS-EN Counting .967** .897** .656** .562** .516** .509** .713* .651** .499** .570** .548** .475** .502** 3. TRS-EN Relational .910** .789** .696** .583** .509** .521** .669** .553** .530** .578** .574** .538** .486** 4. TRS-EN Arithmetic .639** .502** .611** .468** .454** .383** .405* .444** .366* .482** .400** .476** .222 5. ENT .891** .921** .512** .488** .463** .573** .818** .747** .716** 6. ENT Counting .638** .482** .412* .394* .515** .718** .751** .516** 7. ENT Relational .548** .492** .508** .509** .938** .550** .484** .510** .471** .524** .761** .608** .770** Subsequent 8. TRS-EN .630** .583** .562** .351* .345 .971** .827** .752** .676** .633** .581** 9. TRS-EN Counting .634** 605** .529** .340* .355* .983** .724** .665** .667** .637** .578** 10. TRS-EN Relational .490** .452** .468** .203 .233 .909** .858** .560** .495** .481** .410** 11. TRS-EN Arithmetic .507** .424** .531** .360* .209 .655** .570** .408** .622* .615** .517** 12. ENT .685** .597** .603** .532** .670** .565** .563** .546** .271 .906** .881** 13. ENT Counting .608** .542** .491** .485** .485** .540** .549** .457** .244 .857** .598** 14. ENT Relational .604** .520** .568** .448** .687** .486** .471** .522** .245 .908** .562**
- 4 Note. The results for the 4-year-olds are shown above the diagonal. The results for the 3-year-olds are shown below the diagonal.
- 5 ENT = Early Numeracy Test.
- 6 * p <.05. ** p <.001
Factor Loadings, R
3-year-olds 4-year-olds One-factor Three-factor One-factor Three-factor Factor loading R2 Counting Relational Arithmetic R2 Factor loading R2 Counting Relational Arithmetic R2 Item1 .793 .629 .812 .659 .930 .864 .938 .880 Item2 .743 .552 .758 .574 .899 .808 .903 .815 Item3 .725 .526 .738 .544 .859 .737 .863 .744 Item4 .703 .494 .735 .540 .880 .774 .888 .789 Item5 .736 .541 .764 .583 .848 .719 .854 .729 Item7 .810 .656 .824 .678 .664 .442 .669 .447 Item8 .846 .716 .858 .736 .845 .714 .848 .720 Item9 .643 .413 .668 .446 .856 .733 .865 .748 Item10 .626 .391 .646 .417 .802 .643 .811 .658 Item11 .846 .716 .868 .754 .951 .904 .958 .918 Item12 .837 .700 .857 .734 .909 .826 .912 .831 Item13 .720 .519 .744 .553 .907 .822 .911 .830 Item15 .866 .750 .920 .847 .871 .758 .888 .788 Item16 .849 .720 .867 .752 .863 .745 .876 .767 Item17 .581 .338 .609 .371 .821 .674 .836 .699 Item18 .535 .286 .561 .314 .888 .788 .905 .820 Item19 .746 .556 .764 .583 .875 .765 .889 .790 Item20 .685 .469 .722 .521 .896 .803 .915 .836 Item21 .861 .742 .956 .914 .805 .649 .892 .796 Item22 .837 .701 .906 .820 .719 .516 .786 .618
By Terhi Vessonen; Anna Widlund; Airi Hakkarainen and Pirjo Aunio
Reported by Author; Author; Author; Author