1565 # 0 & ' 0 ' ( + 65 1 ( 1 ( 69 # 4 ' 41 % ' 55 / 2418 ' / ' 06 1

In recent years a substantial number of organizations have gained experience in software process improvement (SPI). Furthermore, some researchers have studied such organizations by collecting and analyzing costs and benefits data on their SPI efforts. The objective of this report is to review and summarize the empirical evidence thus far on the costs and benefits of SPI. The intention is that this review would be utilized to support the business case for initiating and continuing SPI programs, to aid in the selection amongst the alternative improvement paradigms, to make more accurate estimates of the costs and benefits of such efforts, and to help set and manage the expectations of technical staff and management. The need for such a review is supported by the results of two recent surveys that were conducted by the SEI. The first survey was administered to individuals at the National SEPG Conference in 1993 and at an SPI tutorial during the Software Engineering Symposium in 1993 [25]. The respondents represented organizations that had mature SPI programs. More than seventy percent stated that they need information on the benefits of SPI (by choosing the "very high" or "high" response category in terms of characterizing their needs), which was also ranked as the highest need of the respondents. This indicates a need for consolidation of the evidence on the benefits of SPI. The second survey solicited information from organizations that had conducted software process assessments between 1992 and 1993 [26]. The results indicate that 77% of the respondents "Strongly Agree" or "Agree" that SPI has taken longer than expected and 68% stated that SPI has cost more than expected. This indicates a need for information to help estimate the costs of SPI and to set and manage expectations from SPI. Two general paradigms to SPI have emerged, as described by Card [10]. The first is the analytic paradigm. This is characterized as relying on "quantitative evidence to determine where improvements are needed and whether an improvement initiative has been successful". The second, the benchmarking paradigm, "depends on