TY - GEN
T1 - Trust yourself! or maybe not
T2 - 35th Brazilian Symposium on Software Engineering, SBES 2021, held in conjunction with the Brazilian Conference on Software: Theory and Practice, CBSoft 2021
AU - Matsubara, Patricia
AU - Steinmacher, Igor
AU - Maldonado, José
AU - Gadelha, Bruno
AU - Conte, Tayana
N1 - Publisher Copyright:
© 2021 ACM.
PY - 2021/9/27
Y1 - 2021/9/27
N2 - Software effort estimates are uncertain, given that they are probabilistic assessments of the future. Evaluating their uncertainty involves assigning them an appropriate confidence level and is paramount for satisfying commitments in software projects. However, estimators tend to be overconfident about their estimates, hampering the accuracy of their uncertainty assessments. Our research goal is to identify the factors related to overconfidence and uncertainty assessments in software estimation. To do so, we carried out a Systematic Literature Mapping (SLM), based on automated and snowballing searches. Our findings include eight factors related to overconfidence and uncertainty assessment. Some of them resulted in unexpected implications for practice. We also identified valuable and easy-to-use metrics that software practitioners can apply smoothly in their daily practice. Additionally, very few field and respondent studies exist about the topic. The software engineering area can significantly benefit from investigating how much practitioners know about the overconfidence effect, as well as of a better comprehension of the perceived importance, practices, and accuracy of uncertainty assessments in the software industry.
AB - Software effort estimates are uncertain, given that they are probabilistic assessments of the future. Evaluating their uncertainty involves assigning them an appropriate confidence level and is paramount for satisfying commitments in software projects. However, estimators tend to be overconfident about their estimates, hampering the accuracy of their uncertainty assessments. Our research goal is to identify the factors related to overconfidence and uncertainty assessments in software estimation. To do so, we carried out a Systematic Literature Mapping (SLM), based on automated and snowballing searches. Our findings include eight factors related to overconfidence and uncertainty assessment. Some of them resulted in unexpected implications for practice. We also identified valuable and easy-to-use metrics that software practitioners can apply smoothly in their daily practice. Additionally, very few field and respondent studies exist about the topic. The software engineering area can significantly benefit from investigating how much practitioners know about the overconfidence effect, as well as of a better comprehension of the perceived importance, practices, and accuracy of uncertainty assessments in the software industry.
KW - Overconfidence
KW - Software effort estimation
KW - Uncertainty assessments
UR - http://www.scopus.com/inward/record.url?scp=85117146625&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85117146625&partnerID=8YFLogxK
U2 - 10.1145/3474624.3474643
DO - 10.1145/3474624.3474643
M3 - Conference contribution
AN - SCOPUS:85117146625
T3 - ACM International Conference Proceeding Series
SP - 452
EP - 461
BT - CBSOFT 2021 - Brazilian Conference on Software; Proceedings - 35th Brazilian Symposium on Software Engineering, SBES 2021
PB - Association for Computing Machinery
Y2 - 29 September 2021 through 1 October 2021
ER -