Children with reading disability can demonstrate different behavioral patterns or different reading disability profiles. Characterizing these profiles could be important for understanding the reason(s) for reading disability. Large samples sizes are often necessary for profile studies and for that reason researchers may try to pool data from many different research sites. Unfortunately, there is often missing data in large multi-site data sets that can be problematic for identifying different reading profiles, especially when the missing data are critical for differentiating between the reading profiles. We collected test scores from previous studies on reading development and disability to create a large data set. We tested different approaches for dealing with missing data to statistically estimate the missing data using a synthetic data set where we knew the reading profile for each case and then demonstrated the effectiveness of the methods with the real data (726 subjects). These methods can now be used to examine if there are reading disability profile differences in brain anatomy in this data set, for example. In addition, these methods can also be used for identifying profiles when there is missing data in other large data sets. You can read about the details of this study in our manuscript: Reading Profiles in Multi-Site Data with Missingness, which was published online in Frontiers in Psychology. https://www.frontiersin.org/articles/10.3389/fpsyg.2018.00644/full
