Glowik, John2019-11-282019-11-281977http://hdl.handle.net/2440/122176The results of this thesis deal with a numerical investigation of data exhibiting split-line regression characteristics. Chapter 1 gives a general introduction and a more informative description of split-line regressions. In Chapter 2 we develop some procedures to test for significant non-linearity characteristics in data sets. In particular four methods are considered in detail and applied to examples in Chapter 5. In Chapter 3 we consider some previously developed procedures for finding the least squares estimate for the intersection point of two regression lines. We also develop new methods of approach in order to find an estimate for the intersection point, γ. In Chapter 4 we consider some inferential problems, where the asymptotic distribution and confidence interval for γ are discussed. In Chapter 5 we look at some experimental and generated data sets, comparing with respect to time, the various methods considered for finding an estimate of γ. In the Appendix we list, in FORTRAN IV, the routines which were developed and used for the various methods of approach.enThesis