In general, there is a significant lack of data availability for undertaking comprehensive
ex-post appraisals, which limits the options for conducting such studies. Parthasarathi and Levinson (2010, p. 438) describe how “the data collection efforts were much more laborious and time consuming than anticipated. The unavailability of data in electronic format, lack of proper documentation, poor record keeping and data archiving procedures complicated the data collection process and subsequent analysis”. Similar comments are found in more or less all expost appraisals that the authors of the present review have analysed. Apart from the obvious problem of making the acquisition of sufficient data for the studies difficult, the issue presents a more complicated problem in the form of availability bias (Tversky & Kahneman, 1973). Extreme cases risk being over-represented in study samples when data availability is poor, since projects that deviate the most from expectations are likely to be subject to the most intense critical scrutiny and thus provide more readily available data sources (Nicolaisen, 2012; Siemiatycki,2009; Skaburskis & Teitz, 2003).