Limitations and Future work
The result from regression model can not explain the agesimphenomenon, to find out the underlaying relationship between ageism and causation, I need to look into other aspects, such as race, sexual orientation and health status, however, many types of these variables have been deleted in the dataset in order to reduce the chance to identify individuals. To fix this, I plan to involve more dataset in the future work, such as regional distribution data and employment data. Also, it could be meaningful to look into a time range and different areas to see whether social development has effect on ageism phenomenon.
As for the method part, if more variables involved, using decision tree to see the discrimination probability would be useful.