In August 2011 at the International Symposium on Integrated Functionalities in Cambridge, a whole session was devoted to the electrocaloric effect, which is undergoing a modest renaissance. Surprisingly, the various reports showed that the indirect method of measuring cooling (described in the following sections) did not usually agree with the direct method (actually measuring temperature change with a thermometer). However, there was no obvious systematic error: sometimes the indirect temperature change was larger and sometimes smaller. The discrepancies were beyond the experimental errors. The majority of the present thesis is dedicated to careful reexamination of some of the assumptions made in the indirect method, both during measurement and in the subsequent data analysis that leads to inferred temperature changes. Experimentally, I conclude that the most serious systematic error is likely to be the unwarranted assumption that polarization and field measurements, recorded in hysteresis loops that are traced within a millisecond or less, are all taken at the same temperature. In reality, the experience of the material during such loops is neither isothermal nor adiabatic. Other systematic errors relate to data analyses and are discussed in detail. In some ways, therefore, this thesis has a negative flavour. But it is not designed to criticise prior work. Rather, it is intended to discriminate between reliable experimental procedures and those less convincing. This is a line of research with important technology transfer possibilities, and hence the numerical values of electric cooling must be unusually reliable if we are to avoid unwise capital investment as a country.