Engelsk titel: Optimizing the RMET to measure bias not performance differences
Läs online
Författare:
Pfuhl, Gerit
;
Ekblad, Leif
Email: gerit.pfuhl@uit.no
Språk: Eng
Antal referenser: 41
Dokumenttyp:
Artikel
UI-nummer: 18010246
Sammanfattning
Background: Human social interaction and communication is complex. Sending a verbal message is often accompanied by intonations, facial expressions, grimaces, and body postures. Nonverbal signals are potentially open for misinterpretation. One popular test for assessing the interpretation of facial expressions is the “Reading the Mind in the Eyes” Test (RMET). This test has been used to relate Theory of Mind abilities along the autistic spectrum. However, this test was normed on a small sample of students, and answers were coded binary as either correct or wrong.
Methods: We recruited from various forums, blogs, and personal websites over 10,000 people. To assess autistic traits (neurodiversity), we used the Aspie Quiz, which agrees well with the AQ test (Ekblad, 2013). Importantly, we included an “I don’t know” answer option. Further, participants could freely indicate which emotion they read in the eyes. Applying an iterative process, we derived alternative mental state descriptors.
Results and conclusion: This optimized RMET increased the ability to differentiate between people with few or many autistic traits, respectively. By using logistic regression, the test is able to measure difference in bias, not just performance. We found a pronounced negativity bias among people who scored high on many autistic traits. This bias may contribute to a vicious circle of avoiding social interactions.