Immersive Virtual Environments (IVEs) allow participants to interact with their 3D surroundings using natural hand gestures. Previous work shows that the addition of haptic feedback cues improves performance on certain 3D tasks. However, we believe this is not true for all situations. Depending on the difficulty of the task, we suggest that we should expect differences in the ballistic movement of our hands when presented with different types of haptic force-feedback conditions. We investigated how hard, soft and no haptic force-feedback responses, experienced when in contact with the surface of an object, affected user performance on a task involving selection of multiple targets. To do this, we implemented a natural egocentric selection interaction technique by integrating a two-handed large-scale force-feedback device in to a CAVE^TM-like IVE system. With this, we performed a user study where we show that participants perform selection tasks best when interacting with targets that exert soft haptic force-feedback cues. For targets that have hard and no force-feedback properties, we highlight certain associated hand movement that participants make under these conditions, that we hypothesise reduce their performance.