Research Methodology for Testing the Algorithmic Alienation Index and Systemised Self Hypothesis v1

Abstract

This research methodology specifies a staged, mixed-method programme to develop and test the proposed Algorithmic Alienation Index (AAI) and the hypothesised Systemised Self (SS). It makes a necessary distinction between Algorithmic Alienation (AA), an established theoretical construct with partial qualitative empirical grounding in Kanbay, Akçam, and Arkan (2026) and Arkan, Kanbay, and Akçam (2026), and the newer claims under examination. The AAI is a proposed operational instrument, not a validated measure of AA; SS is a falsifiable hypothesis, not a diagnosis or demonstrated endpoint. International English-speaking adults aged 18–60 would be recruited with quotas across country/region, age, gender, education, and self-reported intensity of personalised-feed use. The proposed hybrid data strategy makes a controlled research feed the primary objective behavioural source; participant side, consented tracking and data donation provide naturalistic evidence; voluntary platform backend data are supplementary rather than a feasibility condition. The programme comprises content validation and cognitive interviews, exploratory factor analysis (EFA; target n = 800), independent confirmatory factor analysis (CFA; target n = 1,200), a 12-month cohort, a controlled feed friction shock experiment, a field randomised technical encouragement study, cognitive tasks, interpretative phenomenological analysis (IPA) interviews, and a blinded longitudinal expert all-data (LEAD)-style panel. Approximately 30-items are included in the AAI for psychometric reduction throughout testing. Competing one factor, four correlated factor, second order, and cognitive affective models will be assessed. The theorised weights (.30/.25/.25/.20) and resistance multiplier are explicitly treated as hypotheses; base, resistance, and adjusted scores will always be reported separately. SS classification will be evaluated against a blinded LEAD–IPA multimethod reference standard that permits indeterminate and alternative explanation conclusions, avoiding both a false claim and incorporation bias. Direct randomisation supports causal inference in the controlled feed; field encouragement will use intention to treat primary estimates and instrumental variable complier average causal effect analyses secondarily. Longitudinal random intercept cross lagged panel models intend to address temporal ordering without causal overclaim. Governance prioritises data minimisation, no message-content capture by de

Keywords