Reliability of Scales

How Reliability was Measured

Reliability of multi-item scales was assessed using Cronbach’s alpha (α), which ranges from 0 to 1. Conventionally, values of α ≥ .70 are considered acceptable, ≥ .80 good, and ≥ .90 excellent.

Reliability estimates were based on responses from 186 participants who completed the TESTERLY Career Inventory for personal curiosity and vocational guidance purposes. Participants discovered the service organically through search engines and posts on platforms such as Reddit and Facebook. For more information on the sample, including demographic and geographic details, see the description of the sample.

RIASEC interest scales

Summary: Each Holland Code RIASEC scale was measured with 13 items. The average Cronbach’s α was .90 (excellent), with individual scales ranging from .88 to .92.

Specific Interest Scales

Summary: Across the 59 specific interest scales, the average Cronbach’s α was .89. Of these, 29 scales showed reliability above .90 (excellent), and 29 fell between .80 and .90 (good). Only one scale - Personal Service - had a reliability below .80 (α = .76), which is still in the acceptable range. To improve this, two additional items have since been added to the Personal Service scale, ensuring that all specific interest scales achieve a reliability of at least .80.

Of the 59 specific interests, most (45) of them are measured using six items. To capture all relevant aspects of certain scales, 13 interests included additional item content.

Personality / Work Styles

Summary: Across the 17 personality scales, the average Cronbach’s α was .85. Four traits (Leadership, Social Orientation, Stress Tolerance, and Innovation) showed reliability above .90, while two traits (Independence and Cooperation) had reliability between .70 and .80.

These results are strong for measures of personality traits, which traditionally show lower internal consistency than measures of specific interests. Personality traits are broader, more complex constructs and typically require more items to measure reliably. In this case, all but two traits were assessed with 10 items. The two scales with reliability below .80 are consistent with expectations, as those traits are particularly diverse in nature.