We present action research on computational thinking (CT) in elementary education, based on a schoolyear-long experiment in a typical no-frills Greek public school during the pandemic. As science and technology progress at an overwhelming pace, educational systems struggle with the impossibility of following and questioning the value of these -very important- advances in a school curriculum which must supply lifelong foundations rather than the moving target of “the state of the art”. Focusing on horizontal, transversal skills is a -partial- answer; CT is a case in point.

CT is by now established as an invaluable skill for all at higher education level. Bringing it (down?) to school education is mostly restricted to elite or experimental environments. In the Greek national curriculum, it is only mentioned as a side-benefit of informatics teaching with very little on how to cultivate it in practice. Our holistic research question: is CT actually a transversal interdisciplinary skill which can and should be developed starting at an early stage?

The experiment we designed and partially executed had several challenges: student attitude (informatics, one hour weekly, is for having fun with the computer, not for learning); insufficient access to hardware, software and digital content at school and unequal at home; and the perpetual “what’s the use?” question. We took these in stride as integral to the problem, wanting our results to be broadly generalizable and not attributable to special favorable school conditions.

We worked with 57 third and fourth graders including a substantial number of immigrants -typically of deficient Greek language skills and poorer economic conditions. Our baseline, depicted in a pupil profile, mapped their relation with computers (questionnaire for access and usage), school achievement (language and math grades) and spatial skills (a raven test). These were used to determine possible home use of learning scenarios and flipped classroom.

For the action research we utilized extensively Bebras-type activities, starting with a baseline test for assessing individual and collective progress. During the following six months, with the pandemic interrupting, upsetting, and somehow helping as pupils became accustomed to computer use for schoolwork, we developed and applied, in classroom and Webex, learning activities to further CT. They were based on past Bebras exercises, as well as developing algorithmic skills using Microbit, LightBot, and Hour-of-Code (Minecraft). The learning scenarios developed were case studies integrating CT into History and Literature, including logic and puzzle games for problem breakdown and solving, abstraction, and pattern recognition -the basic elements of computation thinking.

Pupils participated in a mid-term Bebras competition (with very encouraging results) and a similar final test with a clear improvement of their skills over their initial profiles. Besides the test our data included a diary with qualitative observations regarding the pupils’ psychological and emotional reactions during testing as well as their dealing with such activities under testing.

The experiment provides evidence that CT can be developed and advanced in typical school conditions, starting from early ages, when appropriate learning activities are provided. This is, in our opinion, where educational policy makers should focus their efforts for integrating computational thinking in elementary school curricula.

Keywords: Education, Computational Thinking.