Location-based social networks, in addition to revealing users' online social network, also informs users' actual movements in the offline physical world. Due to this, they have recently been used in large-scale mobility and urban studies. In this paper, using a rigorous statistical methodology, we have found that a rank-distance distribution, which in recent research has been suggested to be a universal mobility law across cultural, demographic and national boundaries, does not follow a power-law distribution as originally claimed. Using a large-scale dataset obtained from Foursquare in Switzerland and New York City, we have shown that place transitions can be better explained using a log-normal and power-law with exponential cutoff model. Our study suggests that urban mobility patterns are more nuanced than previously reported and that goodness-of-fit tests need to be done in view of the generality of human mobility models.