Programming misconceptions have been a topic of interest in introductory programming education, with a focus on university level students. Nowadays, programming is increasingly taught to younger children in schools, sometimes as part of the curriculum. In this study we aim at exploring what misconceptions are held by younger, school-age children. To this end we design a multiple-choice questionnaire with Scratch programming exercises. The questions represent a selected set of 11 known misconceptions and relate to basic programming concepts. 145 participants aged 7 to 17 years, with an experience in programming, took part in the study. Our results show the top three common misconceptions are the difficulty of understanding the sequentiality of statements, that a variable holds one value at a time, and the interactivity of a program when user input is required. Holding a misconception is influenced by the mathematical effect of numbers, semantic meaning of identifiers and high expectations of what a computer can do. Other insights from the results show that older children answer more questions correctly, especially for the variable and control concepts. Children who program in Scratch only seem to have difficulties in answering the questions correctly compared to children who program in Scratch and another language. Our findings suggest that work should focus on identifying Scratch-induced misconceptions, and develop intervention methods to counter those misconceptions as early as possible. Finally, for children who start learning programming with Scratch, materials should be more concept-rich and include diverse exercises for each concept.@en

When teaching novices programming, misconceptions can occur. Misconception are incorrect beliefs about certain programming concept. For example, some novices think that a variable can hold multiple values, in the case of two consecutive assignment statements, such as x = 5; x = 7. While explaining variables introductory materials often use the metaphor of a box for a variable, which might contribute to the 'multiple values' hypothesis. To investigate this, we design and run a controlled experiment with 496 novice programmers, both children and adults. Half of our participants receive an introductory programming lesson in which we explain a variable as a box, while the other half of participants receive the explanation of a variable as being a label. They are subsequently questioned about their understanding of variables. Our results show that, for the simple questions involving one assignment, the box group performs better. However, for questions involving the misconception - with two consecutive assignment statements - the label group outperforms the box group. This however primarily occurs when considering variables of type string, for integers subjects interpret the statements as numeric values to be added.

@en