Policy representation has been one of the foremost topics in political science. The pre-condition is the stability of policy preferences. While individual opinions may be influenced by many sources, scholars have found the macro level of opinion to be stable. The disaggregation of survey data may lead to the problem of a large standard deviation being encountered due to the small number of observations in some counties. Therefore, multilevel regression with post-stratification (MRP) is used to estimate public opinion toward budget spending on social welfare between 2007 and 2013. These MRP estimates are validated. However, this data analysis shows mixed results regarding the stability of public opinion in Taiwan.