IELS 2018 OUTCOME MEASURES

Assessment domain(s)
  • Emergent literacy
  • Emergent numeracy
  • Self-regulation
  • Socio-emotional skills
TOP
Achievement and test scales
Scale Creation
  • The IELS scales for the 10 outcome domains were developed based on item response theory and population models using plausible values.
  • Each scale was constructed as theorized in the assessment framework.
    • However, analysis was undertaken where there was some evidence of more complex dimensionality.
    • The only case where a theoretically proposed dimension was separated into multiple dimensions as a result of the dimensionality analysis was empathy (emotion identification and emotion attribution).
  • Once scales were developed, the population model (IRT calibration model and latent regression model combined) was applied to draw five plausible values for each child (across each scale).
  • In order to generate plausible values for the ten domains of interest, multi-dimensional scaling of ten latent dimensions (domains) was conducted.
  • The multidimensional random coefficients multinomial logit model (MRCMLM) was used in two steps: national and international item calibrations, and national conditioning and production of plausible values.
  • As this is the first IELS study, the scale established will be linked to those in any future study in order to establish trends. The outcome domains were scaled and transformed such that they have an international mean of 500 and a standard deviation 100.

 

List of Achievement Scales
  • Literacy
  • Numeracy
  • Inhibition
  • Mental flexibility
  • Working memory
  • Emotion identification
  • Emotion attribution
  • Prosocial behaviour
  • Disruptive behaviour
  • Trust
TOP
Questionnaire and background scales
Scale Creation
  • Item dimensionality was first confirmed using structural equation modelling (SEM) based on the conceptual underpinning of the questions in the assessment framework and the identified structure from field test data.
  • The internal consistency of the proposed scales was then established.
    • Cronbach’s alpha was used to check internal consistency of each scaled index within the countries and to compare it between the countries.
    • The assessment of model fit was primarily conducted through reviews of the root mean square error of approximation (RMSEA), the comparative fit index (CFI), and the non-normed fit index (NNFI).
    • The IELS analyses relied on robust weighted least squares estimation (WLSMV) to estimate the confirmatory factor models.
  • IELS categorical items from the context questionnaires were scaled using IRT modelling (i.e., the one-parameter Rasch model) to derive estimates of each scale for each child.
  • Weighted likelihood estimates, in short, WLE (logits), for the latent dimensions were transformed to scales with an average of 0 and a standard deviation of 1 (with equally weighted samples).
  • A composite index reflecting socioeconomic background was derived nationally, using principal component analysis of three family background indicators
    • highest parental occupational status
    • highest educational level of parents (in years of education according to ISCED)
    • household income
  • Only scale indices are reported below.

 

List of Background Scales
  • Staff reports of child’s global development
  • Staff reports of child’s social and emotional skills
  • Staff reports of child’s cognitive and motor skills
  • SES index score (nationally standardised)
TOP
Sources - Measures
TOP