
346 Empirical Research in Software Engineering
TABLE 8.5 (Continued)
Construct Validity Threats
Threat Type Threat Description Threat Mitigation
Studies that
Encounter
TheseThreats
Measurement bias Incomplete fault data may create bias as only xed
faults are considered.
The study should account for bias because of incomplete
faults.
N: S14; S18; S33;
S34; S41.
Appropriate threshold values for metrics are subjective
in nature. Thus, results may not hold or change on
other threshold values leading to bias.
As threshold values are subjective in nature, the study
should validate other threshold values to increase the
signicance of the results.
N: S14, ...