Photothermal radiometry has been widely used to measure the thermal diffusivity of bulk materials. In the case of thin plates and filaments, a one-dimensional heat propagation model including heat losses has been developed, predicting that the thermal diffusivity can be obtained by recording both the surface temperature amplitude and phase profile slopes ("slope method"). However, this method has given highly overestimated values of the thermal diffusivity of poor-conducting films and filaments. In this paper we analyze the effect of the experimental factors affecting the thermal diffusivity measurements of thin plates and filaments using infrared thermography, in order to establish the experimental conditions needed to obtain accurate and reliable values of the diffusivity of any kind of material using the slope method. We present the calculations of the surface temperature of thin isotropic and anisotropic plates heated by a modulated and tightly focused laser beam, showing that the slope method is also valid for this kind of pointlike heating. Special attention is paid to the effect of surface heat losses (convective and radiative) on the diffusivity measurements of small-dimension and poor-conducting materials. Lock-in thermography measurements performed in the best experimental conditions on a wide set of samples of different thermal properties (thin isotropic and anisotropic plates and filaments) confirm the validity of the slope method to measure accurately the thermal diffusivity of samples of these shapes.