Page 96 [96]
OCR
ZSUZSANNA MIRNICS community members. Ihese may include data on work performance, work absence, fluctuation statistics etc. Community data or other objective data may be taken as well, about a client’s economic contribution to his or her community." To be useful in an effectiveness analysis, measures of supervision outcome must be both reliable and valid. Reliability means that the measure gives the same finding over multiple uses (assuming no change in what is being measured) and provides the same findings when used by different researchers. Validity means that the measure assesses the outcomes that it is supposed to measure and provides data that are generalizable. Neither reliability nor validity is intended as a theoretical concept; each is established by pre-testing the measuring instrument". Experimental designs are also highly recommended. The classical type assigns clients either to receive intervention or not to receive it, according to a random selection procedure. The functioning of each subject is measured following the process. Client functioning under the control condition is used to “subtract” the “effects” due to intervention from the “effects” due to other factors. When this type of supervision versus no-supervision research design is used, explanations of apparent improvements by factors other than supervision itself can be rejected with reasonable confidence. As long as clients have been assigned randomly to the different conditions, it can be assumed (within identifiable probability limits) that the obtained “effects” are due to the intervention. Control groups are necessary; otherwise such inferences about causal factors are extremely difficult to develop. A multiple group (supervision vs. supervision) design is slightly different: here, superiority of one intervention or method over another can be tested. In such comparison studies clients are assigned randomly to method A, method B etc.!? Another suggested approach is use so-called mediator variables in statistical analyses. Mediators show important relations between an intervention and outcome, but may not explain the precise process of change." Establishing a mediator has several reguirements. First, I would not want multiple mediators 10 Ibid.; Office of Technology Assessment, Washington D.C., Congress of the United States. Series of papers on technology assessment, Background paper 3, The Efficacy and Cost Effectiveness of Psychotherapy, October 1980, www.princeton.edu/~ota/disk3/1980/8020/802005.PDF, accessed 1 September 2015. See for instance: J. C. Nunnally — H. Bernstein, Psychometric theory, 3rd ed., New York, McGraw-Hill, 1994. Office of Technology Assessment, The Efficacy and Cost Effectiveness of Psychotherapy. A.E. Kazdin, Mediators and mechanisms of change in psychotherapy research, Annual Review of Clinical Psychology 3 (2007) 1-27. + 94 +
