Background With the increased attention on the implementation of inquiry activities in primary science classrooms, a growing interest has emerged in assessing students’ science skills. Research has thus far been concerned with the limitations and advantages of different test formats to assess students’ science skills. Purpose This study explores the construction of different instruments for measuring science skills by categorizing items systematically on three subskill levels (science-specific, thinking, metacognition) as well as on different steps of the empirical cycle. Sample The study included 128 fifth and sixth grade students from seven primary schools in the Netherlands. Design and method Seven measures were used: a paper-and-pencil test (PPT), three performance assessments, two metacognitive self-report tests, and a test used as an indication of general cognitive ability. Results Reliabilities of all tests indicate sufficient internal consistency. Positive correlations between the PPT and the three performance assessments show that the different tests measure a common core of similar skills thus providing evidence for convergent validity. Results also show that students’ ability to perform scientific inquiry is significantly related to general cognitive ability. No relationship was found between the measure of general metacognitive ability and either the PPT or the three performance assessments. By contrast, the metacognitive self-report test constructed to obtain information about the application of metacognitive abilities in performing scientific inquiry, shows significant – although small – correlations with two of the performance assessments. Further explorations reveal sufficient scale reliabilities on subskill and step level. Conclusions The present study shows that science skills can be measured reliably by categorizing items on subskill and step level. Additional diagnostic information can be obtained by examining mean scores on both subskill and step level. Such measures are not only suitable for assessing students’ mastery of science skills but can also provide teachers with diagnostic information to adapt their instructions and foster the learning process of their students.

In most primary science classes, students are taught science inquiry skills by way of learning by doing. Research shows that explicit instruction may be more effective. The aim of this study was to investigate the effects of explicit instruction on the acquisition of inquiry skills. Participants included 705 Dutch fifth and sixth graders. Students in an explicit instruction condition received an eight-week intervention of explicit instruction on inquiry skills. In the lessons of the implicit condition, all aspects of explicit instruction were absent. Students in the baseline condition followed their regular science curriculum. In a quasi-experimental pre-test–post-test design, two paper-and-pencil tests and three performance assessments were used to examine the acquisition and transfer of inquiry skills. Additionally, questionnaires were used to measure metacognitive skills. The results of a multilevel analysis controlling for pre-tests, general cognitive ability, age, gender and grade level indicated that explicit instruction facilitates the acquisition of science inquiry skills. Specifically on the performance assessment with an unfamiliar topic, students in the explicit condition outperformed students of both the implicit and baseline condition. Therefore, this study provides a strong argument for including an explicit teaching method for developing inquiry skills in primary science education.

Computers are promising tools for providing educational experiences that meet individual learning needs. However, delivering this promise in practice is challenging, particularly when automated feedback is essential and the learning extends beyond using traditional methods such as writing and solving mathematics problems. We hypothesize that interactive knowledge representations can be deployed to address this challenge. Knowledge representations differ markedly from concept maps. Where the latter uses nodes (concepts) and arcs (links between concepts), a knowledge representation is based on an ontology that facilitates automated reasoning. By adjusting this reasoning towards interacting with learners for the benefit of learning, a new class of educational instruments emerges. In this contribution, we present three projects that use an interactive knowledge representation as their foundation. DynaLearn supports learners in acquiring system thinking skills. Minds-On helps learners to deepen their understanding of phenomena while performing experiments. Interactive Concept Cartoons engage learners in a science-based discussion about controversial topics. Each of these approaches has been developed iteratively in collaboration with teachers and tested in real classrooms, resulting in a suite of lessons available online. Evaluation studies involving pre-/post-tests and action-log data show that learners are easily capable of working with these educational instruments and that the instruments thus enable a semi-automated approach to constructive learning.