Assessment of Interpersonal Motivation in Transcripts (AIMIT): an inter- and intra-rater reliability study of a new method of detection of interpersonal motivational systems in psychotherapy

Assessing Interpersonal Motivations in Transcripts (AIMIT) is a coding system aiming to systematically detect the activity of interpersonal motivational systems (IMS) in the therapeutic dialogue. An inter- and intra-rater reliability study has been conducted. Sixteen video-recorded psychotherapy sessions were selected and transcribed according to the AIMIT criteria. Sessions relate to 16 patients with an Axis II diagnosis, with a mean Global Assessment of Functioning of 51. For the intra-rater reliability evaluation, five sessions have been selected and assigned to five independent coders who where asked to make a first evaluation, and then a second independent one 14 days later. For the inter-rater reliability study, the sessions coded by the therapist-coder were jointly revised with another coder and finally classified as gold standard. The 16 standard sessions were sent to other evaluators for the independent coding. The agreement (κ) was estimated according to the following parameters for each coding unit: evaluation units supported by the 'codable' activation of one or more IMS; motivational interaction with reference to the ongoing relation between patient and therapist; an interaction between the patient and another person reported/narrated by the patient; detection of specific IMS: attachment (At), caregiving (CG), rank (Ra), sexuality (Se), peer cooperation (PC); and transitions from one IMS to another were also scored. The intra-rater agreement was evaluated through the parameters 'cod', 'At', 'CG', 'Ra', 'Se' and 'PC' described above. A total of 2443 coding units were analysed. For the nine parameters on which the agreement was calculated, eight ['coded (Cod)', 'ongoing relation (Rel)', 'narrated relation (Nar)', 'At', 'CG', 'Ra', 'Se' and 'PC'] have κ values comprised between 0.62 (CG) and 0.81 (Cod) and were therefore satisfactory. The scoring of 'transitions' showed agreement values slightly below desired cut-off (0.56). Intra-rater reliability was very good (κ values for Cod = 0.90; κ for all IMS = 0.78). Data seem to support the validity of the AIMIT method in terms of reliability, and encourage to further implementation of the AIMIT approach.