Test reliability refers to a scale's consistency of measurement. Cronbach's Alpha, a measure of reliability, is used to measure the internal consistency of each scale. Perfect reliability is 1.00 and the accepted standard for this type of instrument is .70 - .80 (Murphy & Davidshofer, 2001). BDS tests meet or exceed these thresholds.
In testing, the term validity refers to the extent that a test measures what it was designed to measure. A test cannot be accurate without being valid.
When individuals known to have more severe problems or symptoms receive higher scale scores than individuals known to have fewer problems or symptoms, the test is said to have evidence of construct validity (DeVon et al., 2007). It is anticipated that a repeat offender will have higher scale scores than first-time offenders, indicating more severe symptoms or problems.
Each BDS test contains a Truthfulness Scale. Truthfulness Scale development has been influenced by MMPI Truthfulness Scale methodology. Research has demonstrated that truthfulness is linked to positive treatment outcomes (Barber, et al. 2001; Simpson, 2004). While denial (refutation, problem minimization or lying) has been linked to negative treatment outcomes (Marshall, Thornton, Marshall, Fernandez & Mann, 2001); resistance (Simpson, 2004); problem minimization (Murphy & Baxter, 1997); treatment dropout (Daly & Pelowski, 2000; Evans, Libo & Hser, 2009); and recidivism (Grann & Wedin, 2002; Nunes, Hanson, Firestone, Moulden, Greenberg & Bradford, 2007).
A truthfulness-related problem is identified when a Truthfulness Scale score is at or above the Problem Risk range. Problem (70 to 89th percentile) scorers are typically cautious, guarded and defensive respondents. Problem scorer's test answers should be dealt with carefully in a prudent manner Severe problem scores (90 to 100th percentile) invalidate the test.
Behavior Data Systems (BDS) classifies offenders into four risk ranges:
• Low risk (0 to 39th percentile),
• Medium Risk (40th to 69th percentile),
• Problem Risk (70th to 89th percentile),
• Severe Problem (90th to 100th percentile)
Risk ranges represent degree of severity. Risk ranges were established by converting raw scores using cumulative percentage distributions. Early instrument development included the use of content experts to confirm the proposed risk ranges. Data analyses, in combination with field reports from experienced evaluators have confirmed that these percentile categories provide accurate identification of problem behavior.
Early instrument development included the use of content experts to confirm the proposed risk ranges. Data analyses, in combination with field reports from experienced evaluators have confirmed that these percentile categories provide accurate identification of problem behavior.