Phased array ultrasonic attenuation method has received increasing attention and application in material characterization. However, the combined effect of frequency, grain size, and sound path on the phased array ultrasonic spectrum and attenuation has not been precisely understood. In this study, this combined effect is identified. An experimental strategy and an attenuation coefficient correction method are proposed. A set of Ni-based alloys with varying grain sizes and thicknesses, known as GH4742, is utilized for experimental investigations. Two 5 MHz and 10 MHz nominal frequency phased array probes are used for the back-wall echo acquisition. The results show that the spectral shift and attenuation coefficient are not the simple linear relationship with the sound path. The higher diffraction losses in the far field are close and tend to constant in the range of high frequency. The interface loss and other measurement factors in phased array ultrasound account for more than 60% of the total attenuation in the fine-grain material. Compared with the scattered model, the reference specimen calibration method can also effectively correct the experimental attenuation coefficient.