Rita Kelemen, Csaba Csíkos and János Steklács : Metacognitive strategies in mathematical problem solving: evidence from thinkaloud protocols and videotaped sessions Both in international and national declarations, one of the most important aims of school mathematics is to enable students to solve reallife mathematical problems. In recent years, several studies focused on the questions of how students use (or neglect) their everyday knowledge and experience when solving mathematical word problems in school context. These studies provided evidence related to students’ strong tendencies to exclude their realworld knowledge and realistic considerations from their mathematical word problem solving processes. The purpose of the present study is to use interviews and thinkaloud protocols to analyze and describe children’s problems solving behaviour patterns. The participants were twenty fifth grade students from two schools of Békés County, Hungary. One half of the sample was randomly selected from a group previously involved in metacognitionbased training in the fields of mathematics and reading, and the other half was randomly selected from their control group. We examined students’ realistic problemsolving processes by videotaping their thinkaloud sessions. The word problem used in this investigation was a simplified version of Kramarski, Mevarech and Lieberman’s pizzatask, which required the students to choose and order as many pieces of pizza as possible, from one of two restaurants up to 5000 Forints. Students’ observable behaviour was videotaped during a 20 minute long thinkaloud interview. We used Schoenfeld’s method to represent students’ behaviour patterns. The most interesting finding comes from a cluster analysis, in which student groups could be identified. One cluster consists of students who were members of the experimental group the year before, and they proved to be ‘good metacognitive strategy users’, even though they had below the average marks in maths. On the other hand, some students from the former control group could be classified as ‘poor metacognitive strategy users’. Our investigation has married two methodological ways of measuring metacognition. Using thinkaloud protocols enabled online observation of strategic elements of behaviour, while the posthoc analysis of videotapes enabled us to quantify observable behaviour patterns. MAGYAR PEDAGÓGIA 105. Number 4. 343358. (2005) Address for correspondence: Kelemen Rita és Csíkos Csaba, SZTE Neveléstudományi Tanszék, H6722 Szeged, Petőfi S. sgt. 3034.. 
