GATA: Test–retest reliability of measurement outcomes

Social researchers have long sought to overcome the vulnerability of the self-report method to a variety of effects based on respondents’ inability or unwillingness to answer correctly. One obvious strategy for solving this problem is to try to extract information in such a way that the respondent’s attitude towards the object under study can be assessed by his or her actions “indirectly” related to that object, without relying on the respondent’s “direct” answers as to what he or she thinks that attitude is. One of the many tools that claim to be able to probe the respondent’s consciousness in this way is the Graphical Associative Test of Attitudes (GATA). This paper presents a summary of the test-retest reliability of GATA. 18 repeat tests two (12 tests) and four (6 tests) weeks after the initial meas¬urement showed that GATA can potentially produce reliable results. At the same time, this ability is not completely stable, it depends on the subject being assessed and the time between the tests being compared. While the reliability of GATA over a two-week period is often satisfactory (Spearman ρ > 0.700 in 4 out of 12 tests) and comparable to that of the self-report method, over a four-week period, only one out of six GATA evaluations maintains such a high level of reliability. Possible dimensions of future research aimed at improving the reliability of GATA output are discussed in the paper.