The SCORE-15 (Stratton, P., Lask, J., Bland, J., Nowotny, E., Evans, C., Singh, R., Janes, E., & Peppiatt, A. (2013). Detecting therapeutic improvement early in therapy: Validation of the SCORE-15 index of family functioning and change: Validation of the SCORE-15 index. Journal of Family Therapy, 36(1), 3–19. https://doi.org/10.1111/1467-6427.12022) is a questionnaire designed so that family members of families engaged in family/systemic therapy can rate, not themselves, but their family, actually their family functioning. This is in line with much systemic therapy theory that sees the aim as changing the system/family, and change in individuals as secondary to that. It is a 15 item version of an earlier 40 item version (Stratton, P., Bland, J., Janes, E., & Lask, J. (2010). Developing an indicator of family function and a practicable outcome measure for systemic family and couple therapy: The SCORE. Journal of Family Therapy, 32(3), 232–258. https://doi.org/doi: 10.1111/j.1467-6427.2010.00507.x) but has probably largely superceded that.
SCORE has been sponsored by AFT, the UK based Association for Family Therapy and Systemic Practice and is copyright to AFT. There is excellent supporting material at https://www.aft.org.uk/page/score and AFT makes the instrument, and related instruments available without reproduction fees.
Detail #
One interesting issue when analysing data from SCORE measures, also true for questionnaire ratings of therapeutic environments, is that although this is a questionnaire datasets will typically have ratings from more than one person rating the same family (enviroment). This means that individuals’ responses don’t have the crucial “independence of observations” that underpins most statistical analyses which seek to address whether there are systematic effects in the data. For example, we might be interested in the question “do families with a male index client have different SCORE-15 scores from those with female index clients?” If these were self-ratings of independent clients we could approach this question with “tests” which would give a p-values or by computing confidence intervals for the difference in mean scores. Those methods depend on independence of observations but we don’t have this. In principle that takes us into multilevel modelling (MLM) but unless one is working with very, very large families, the typical numbers of members per family won’t be enough for those methods to work.
This is also not just an issue of inter-rater reliability as, although one might expect some agreement, it is also an axiom of families that not all members see them the same way. Ideal quantitative ways to handle such fascinating data are still developing.
Try also … #
Independence of observations
Inter-rater reliability
Multilevel modelling (MLM)
Nesting and nested data
Chapters #
Chapter 4
