As our cities continue to growth in an era of urbanization there is a need to harness the power of big data to support data driven planning. Yet we need to ensure this data in credible and reliable, particularly when obtained from smart phone apps through crowdsourced approaches. The goal of this paper is to gain insight in to the geographic representativeness and bias inherent in a crowdsourced dataset of bicycling routes. The empirical investigation uses the RiderLog bicycling data from 2010 – 2014 to contrast smartphone app collected bicycle commuting data with census journey to work data at the Statistical Area Level 2 census geography. The investigation focuses on Perth and the surrounding area of Western Australia. Both app and census data are converted to similar dyadic spatial interaction matrices. Correlations between the spatial interaction matrices are calculated with a modified t-test for assessing correlation in the presence of spatial autocorrelation. Results indicate the app generated data are representative in a number of SA2s located along the Swan River as well as among clusters of SA2s adjacent and close to the Indian Ocean. The app data are also found to exhibit a strong urban bias. Due to increased numbers of users, app generated data are more representative of the broader cycling population in urban areas than in rural areas. Means of extending this research in order to increase usability of crowdsourced data for bicycle transportation planning and management are also considered.