In a secondary data analysis based on data from a Dutch national science assessment in the fall of 2015, the added value of dynamic testing for determining research skills in the domain of Nature & Technology of grade 6 students in elementary school was examined compared to static testing. To this end, the research data (n=203) were used from one practical assignment, designed according to the principles of dynamic testing and involved the design and implementation of an experiment and a follow-up experiment with a ramp task, the Marble Run. Depending on progress on the subtasks, reading, process, and substantive help was provided, following a protocol of graduated prompts. Analyses aimed at determining the progress made in experiment 2 after supported participation in experiment 1 revealed groups with different learning potential. The largest progression observed was made by students who did not reach a solution to a subtask in experiment 1 even after all support, but succeeded independently in doing so in experiment 2. The relative proportion of nonsolvers in experiment 1 who made this progression was 36% setting the experimental variables 42% in repeated measurement and 56% drawing conclusions. Latent class analyses resulted in three groups, each of which went through a different history of support. The biggest difference concerned responsiveness to support with a crucial sub-skill for experimentation: control-of-variables. Furthermore, in one group, giving reading/ process support appeared to play an important part in finding solutions. The group of students who benefited the least from help appeared to have less knowledge of N&T and a more negative view of their own research skills.