In this chapter we’ve presented some distinctions that are important in the analysis of data. We’ve explained effort (the amount of work performed), effectiveness (the results of the work), and...

In this chapter we’ve presented some distinctions that are important in the analysis of data. We’ve explained effort (the amount of work performed), effectiveness (the results of the work), and efficiency (the comparison of results and efforts expended). These terms can help you understand that accountability is not simply a matter of raising effectiveness levels; accountability must also exhibit an optimal ratio of effort expended for the effectiveness obtained. We also discussed practical, statistical, and theoretical significance. The first refers to social and personal criteria of acceptable behaviors in various social settings, the second relates to mathematical rules, and the third involves theoretical expectations. Sometimes they coincide, but not necessarily; they are independent concepts of equal importance for evaluationinformed practice. However, we want you to understand the implications of the several combinations of significance that we discussed early in this chapter. We presented guidelines for determining the number of observations in a phase and for determining when to change an intervention. We also discussed the use of follow-ups in singlesystem designs, suggesting that follow-ups be used to obtain vital information on the lasting quality of your services. We introduced the issue of autocorrelation, noting that it can be a very serious problem for rigorous evaluation. We showed how to test for the presence of autocorrelated data, and we discussed what to do about it. The latter included a presentation on how to transform autocorrelated data into data that are usable by the several analytic techniques. We also showed how to test for autocorrelation and how to transform autocorrelated data using an easy-to-use personal computer program, SINGWIN, which comes with this book. Finally, we presented a discussion of probability tables that can be used with some analytic procedures and we showed how SINGWIN could be used in place of these tables if you prefer.