In this paper, we use secondary data to make a partial measurement system analysis of railway measurement cars and their obtained track geometry data. When a measurement car passes the same track section shortly after the previous passage, such as returning in the other direction after reaching a railway endpoint, the repeated measurements hold information of the measurement uncertainty of that car. Reasons for the measurement uncertainty can be sought in other variables that also are stored in the database, such as the individual car identity, the type of car, the speed of the car during measurement, and the travelled direction of the car. By also considering other known factors during the time of measurement as regressors, such as ground frost periods, enhanced modelling may be achieved and also indicate if such periods should be avoided to improve the measurement data quality.The results of this study suggest that the type of car had the largest influence on measurement variation out of the studied regressors. If the variation of a track geometry property on a track section is studied, the variation component belonging to the type of car can be deducted, improving data quality. We suggest that the method could also be used to find track sections that are prone to large seasonal variation, such as due to ground frost.