If i were to conduct a sip idea using public data in order to collect original data I would use the US census site. I would conduct a survey viewing the better places to live in America bases off several different variables. I would randomly select states from each region of the united states in order to ensure i have a variety of states with different weather and circumstances. Then i would compare family incomes per state then adding in the variable of average temperature per year and also the number of the population in each state may be variables affecting income. I could do hypothesis testing with income through simple one tailed tests comparing souther states to northern states income. Then i could compare income of states with their average temperature or population size through hypothesis testing.
For an original data sip idea, I would target the expenditures made by college students who use their own money or their parents money to pay for things such as clothes, food, alcohol, gas, coffee, etc. After randomly selecting students to participate in this survey, i would ask them for their demographic information including income and gender. Hypothesis testing would include do men or women pay more for these expenditures on their own, do higher incomes mean more parents pay for their kids, do higher income students buy more alcohol than students with lower incomes, etc.