From the Policy Concepts in 1000 Words: Policy change and measurement post:
[see also Policy Concepts in 1000 Words: Policy change
Policy Concepts in 1000 Words: How do policy theories describe policy change?]
The first thing we learn when we study public policy is that no-one is quite sure how to define it. Instead, introductory texts focus on our inability to provide something definitive. That is OK if we want to pretend to be relaxed about life’s complexities, but not if we want to measure policy change in a reasonably precise way. How can we measure change in something if we don’t know what it is?
A partial solution is to identify and measure types of public policy. For example we might treat policy as the collection of a large number of policy instruments or decisions, including:

Public expenditure. This includes deciding how to tax, how much money to raise, on which policy areas (crime, health, education) to spend and the balance between current (e.g. the wages of doctors) and capital (building a new hospital) spending.
Economic penalties, such as taxation on the sale of certain products, or charges to use services.
Economic incentives, such as subsidies to farmers or tax expenditure on certain spending (giving to charity, buying services such as health insurance).
Linking government-controlled benefits to behaviour (e.g. seeking work to qualify for unemployment benefits) or a means test.
The use of formal regulations or legislation to control behaviour.
Voluntary regulations, such as agreements between governments and other actors such as unions and business.
Linking the provision of public services to behaviour (e.g. restricting the ability of smokers to foster children).
Legal penalties, such as when the courts approve restrictions on, or economic sanctions against, organizations.
Public education and advertising to highlight the risks to certain behaviours.
Providing services and resources to help change behaviour.
Providing resources to tackle illegal behaviour.
Funding organizations to influence public, media and government attitudes.
Funding scientific research or advisory committee work.
Organizational change, such as the establishment of a new unit within a government department or a reform of local government structures.
Providing services directly or via non-governmental organizations.
Providing a single service or setting up quasi-markets.

I say ‘partial solution’ because this approach throws up a major practical problem: we do not have the ability to track and characterise all of these instruments in a satisfactory or holistic way. Rather, we have to make choices about what information to use (and, by extension, what to ignore) to build up a partial, biased, picture of what is going on. Here are some of the practical problems we face:
[please see the post for more]