Australia is now conducting a survey on same sex marriage, to decide if they take the vote to the parliament or not. Although it is called a survey, the population is treating it as a referendum. The government could decide on its own to legalize the same-sex marriage, the same way it decides on topics such as immigration laws.
Are there any economic arguments that help a government to decide which topics are polled from the population and which are not?
Polling is expensive so only issues that a lot of people feel strongly about will get polled.
You can imagine that every citizen $i$ would pay $r_i$ monetary units to have a certain policy changed. This number may also be negative to reflect that some people like the current policy better than the proposed change.
Ideally the government would like to know $\sum r_i$, if its positive or negative. But I am sure there are also a lot of practical political considerations. However polling itself has a significant cost. Let us denote this by $C$. The government probably has some vague idea about $\sum r_i$. E.g. most people don't care about the color of chewing gum wraps, so the $\sum r_i$ for banning silver coloring is likely negligable. In this case a poll to determine the exact value would be wasteful as
$$
\sum r_i - C < 0.
$$
So when you pick something to poll it is probably a salient issue, something that a lot of people know about and have strong opinions about. As I said, I am sure that these considerations are usually secondary to political ones.
