As part of a broad program to determine ft values for superallowed O+ --> O(+)beta transitions, the half-lives of Sc-42, V-46, Mn-50 and Co-54 have been measured with isotopically pure samples prepared with an on-line isotope separator. Emitted beta(+) particles were observed with a large-solid-angle, high-efficiency, low-background gas proportional counter, from which signals were multiscaled. The counting electronics had a well-defined, non-extendable dead time. As no exact data analysis is possible, three different procedures were compared. The most stringent method involved regression analysis where a common half-life was simultaneously fitted to the individual decay curves of hundreds of samples and their Poisson statistical weights were distorted to reflect dead-rime losses. The analysis procedures were tested on hypothetical decay data that were created by binning a time sequence of events, obtained from interval distributions reflecting the given half-lives and intensities of two decay components, and allowing for series and non-extendible dead times that reflected experimental conditions. On the basis of these tests it is demonstrated that the analysis does not introduce any bias at the 0.01% level. The deduced half-lives are Sc-42: (680.67 +/- 0.28) ms, V-46: (422.57 +/- 0.13) ms, Mn-50: (283.29 +/- 0.08) ms, Co-54: (193.28 +/- 0.07) ms. These results establish important elements in a demanding test of the Electroweak Standard Model.