This study investigates how young children between 4-6 years old interact with personified robots during a lying situation. To achieve this, a temptation resistance paradigm was used, in which children were instructed to not look at a toy (behind their back) while the instructor (a robot dog, a humanoid or a human) left the room. Results revealed that regardless of the type of communication partner, children's peeking behaviour was similar across the 3 conditions, while there was a tendency of lying more towards the robots. The majority of the children (98%) showed semantic leakage while telling a lie, and most of them (89%) lied and denied their peeking behaviour. Additionally, children generally gave more verbal responses to the robot dog and to the humanoid in comparison with the interaction with the human. Furthermore, the mean pitch of children differed between the robot conditions, i.e. the mean pitch was significantly lower in the robot dog condition in comparison with the humanoid condition. Finally, facial expression analysis showed that children generally appeared happier when they were interacting to the robot dog compared to the humanoid or human.