What are proxy measures?
A proxy is an indirect measure of the desired outcome which is itself strongly correlated to that outcome. It is commonly used when direct measures of the outcome are unobservable and/or unavailable.
Also reading from https://olc.worldbank.org/sites/default/files/1.pdf.
We can say that Proxy major is mostly used on when the researcher has to solve problems which can’t be addressed in a direct way. So instead of asking them a direct question, you ask them an indirect question which would ultimately lead to the answer to close to the answer.
Choosing data source which can be answerable to the Proxy measure’s characteristics is very important. For example in the study of "Measuring income and poverty using Proxy Means Tests" the essence of the proxy means the test method is using household characteristics to predict household welfare. Therefore, we need a the data source that would contain information about both household characteristics (household composition, dwelling qualities, assets) and household welfare (normally measured by consumption volume), and that would allow us to draw a correspondence between the two.
This means that we normally need results of a household survey. Household Income and Expenditure Surveys (sometimes called household economic surveys) or Living Standards Measurement Study surveys are examples of surveys that contain data that could be used in developing a PMT model.
Regardless of the exact type of survey that we want to use, there are two important criteria that our the data source has to meet: (This criteria needs to be identified for each measure and would differ for each project)
• The data should be as recent as possible—as the living conditions of the population change, the relationship between particular household characteristics and its welfare is also prone to change. Ideally, the survey should have been conducted in the last five years.
• The survey results should be representative for the area. If we want to use the PMT scores for targeting social safety nets beneficiaries nationwide, we should use a survey that is nationally representative.
Pros & Cons :
We would be able to address or solve part of the problem by using PMT model, as in this case not all the population at the given location would represent the population.
We might need to identify a diffrent set of measures for each group of the sample we want to target, and at the end, all data points should have some common ground so that data points can be normalized.
Research can be found here for More reading.