Introduction: The HGSHS:A is one of the most commonly used measures of
hypnotic suggestibility. However, this test suffers from low feasibility due to a
time requirement exceeding 1 h, and from a questionable representation of the
normal population. Recently, a short version of HGSHS-5:G was developed and
published, and now the first results are available. The scope of this investigation
was to verify the assumption of equally positioned and normally distributed
scores, resulting in equally sized suggestibility groups in a number of different
studies with full or short versions of HGSHS, and to compare the results of the
11-item score with the 5-item score, the latter being calculated from either the
full version or the short version test.
Methods: Data from 21 studies with testing for HGSHS were analyzed, 15 using
the HGSHS:A full version and six using the HGSHS-5:G short version, for a total
of 2,529 data sets. Position and distribution of both the 11-item score and the
5-item score were tested. Linear regression analysis was used to compare the
two scores, as well as cross-table and weighted Cohen’s kappa to determine
the match of grouping into low and high suggestibility. To evaluate contributing
factors to the observed differences in the study results, a multifactorial analysis
of variance was performed.
Results: In the different studies, position and distribution of scores, as well as
group sizes for low and high suggestibles, varied. All score distributions were
found to be non-normal and shifted to the right from the middle score; the shift
was more extensive with the 11-item score. The correlation between both scores
calculated from full version tests was moderate (R2 = 0.69), as was the match
of suggestibility grouping (κ = 0.58). Studies using the short version involving
less student-dominated populations showed sufficient agreement with the full
version, but lower scores were caused by an increase in the zero score.
Conclusion: A normal population is not represented in most applications of
HGSHS, and grouping into low and high suggestibles varies, mainly due to
different positions of score distributions. A direct comparison of full and short
versions of HGSHS tested in the same subjects is still missing.