Social discounting is when resource allocation decreases as social distance increases. Studies fitting different quantitative models to social discounting data have shown that a q-exponential function based on Tsallis' statistics best fits loss data, whereas a hyperbolic power function best fits gain data. However, a social discounting sign effect, where losses are discounted less than equivalent gains, has not been consistently found. This study fit four different quantitative social discounting models to gain and loss data for 40 United States (US) participants. We compared quantitative model fits to previous studies collected with Japanese and German participants to determine if (1) different quantitative social discounting models best fit loss and gain data, (2) US participants discounted less gains than Japanese participants, but not losses, and (3) US participants showed the sign effect. Results showed that the q-exponential function and the hyperbolic power function best fit median loss and gain data, respectively. There were no significant absolute differences between cultures for gains or losses, and US participants showed a robust sign effect. While most results for US participants were consistent with previous data, future cross-cultural social discounting studies are needed that manipulate sign as well as reward magnitude to determine best quantitative model fits. Social discounting results are also discussed in relation to their application to important health behaviors such as smoking and obesity.