A dynamic assessment tool was developed and validated using Mokken scale analysis to assess the extent to which kindergartners are able to construct unconfounded experiments, an essential part of scientific reasoning. Scientific reasoning is one of the learning processes happening within science education. A commonly used, hands-on, experimentation task was adapted to dynamically assess the use of the so-called control of variables strategy (CVS) by children 4 to 6 years of age. In this task, the children were challenged to design experiments using two ramps with up to four independent variables: weight of the ball, steepness of the slope, place of the starting gate, and surface texture of the slope. There were two scores of CVS use: experiment and variable correct score. The analysis showed it was possible to assess CVS use in a reliable and valid manner with the new assessment tool. Irrespective of the number of variables children were allowed to set, experiments validly measured CVS use. Given that the number of variables to be set increased the difficulty of the experiment, this can be used to scale children's CVS use. In other words, it is possible to differentiate between children on the basis of their CVS use. The children's use of CVS positively related to both their age and nonverbal reasoning ability. The present results thus show that it is feasible to evaluate the ability of kindergartners to construct unconfounded experiments using dynamic assessment. This means that kindergartners can use the CVS and might be seen as natural scientists, who can and will try to unravel the physical world around them. Their explorations appear to need sufficient guidance (i.e. within their zone of proximal development) to design multivariable experiments.