Lecture note on secondary data analysis: for Housing Studies students

Lecture note on secondary data analysis: for Housing Studies students

 

Highlight 5 main ideas of the secondary research method of secondary data analysis in the context of Housing Studies dissertation projects.

Secondary data analysis means answering a new research question by re-analysing data that another organisation or researcher originally collected for a different purpose. For a Housing Studies dissertation, it is especially useful when fieldwork access, time, or budget is limited.

Five main ideas

1.    Use analysable data, not only published statistics
Look for datasets with respondent-, household-, property-, or neighbourhood-level records (“microdata”), rather than only summary tables in annual reports. This allows you to test relationships—for example, whether overcrowding varies by household income, tenure type, or district.

2.    Match your research question to measurable variables
Break broad housing concepts into observable variables. For instance, “housing affordability” could be represented by rent-to-income ratio, monthly mortgage payment, household income, or housing expenditure burden. The video stresses that variables need to be available within the dataset and sufficiently close to the concept in the research question.

3.    Check the codebook before committing to a dataset
A codebook or data dictionary explains each variable, question wording, response categories, numeric codes, missing values, sample coverage, and collection method. Without it, a dataset may be impossible to interpret correctly; it is essential for judging whether “public housing,” “crowding,” “household,” or “income” has been defined in a way suitable for your dissertation.

4.    Ensure variables can be linked in the same dataset
To investigate an association, the relevant variables generally must appear for the same units of analysis. For example, a dissertation question such as “Is housing dissatisfaction associated with overcrowding among public-rental households?” requires records that contain tenure, crowding measures, and satisfaction responses for each relevant household—not three separate published tables.youtube

5.    Refine the question transparently around available data
Secondary-data projects often require a compromise between the ideal question and what existing data actually measure. You may begin with “How does housing quality affect wellbeing?” but revise it to “How is self-reported housing dissatisfaction associated with self-reported wellbeing among renters?” if those are the available variables. Clearly explain this operationalisation, its limitations, and why the measures are defensible in the methodology chapter.

Housing dissertation application

A practical workflow is:

1.    Define the population and unit of analysis: individual, household, dwelling, estate, district, or transaction.

2.    List the concepts in your question and the variables needed for each.

3.    Locate relevant public, institutional, survey, census, transaction, or administrative datasets.

4.    Read the dataset documentation and assess its date, geography, sampling, missing data, and access restrictions.

5.    Align the final research question, hypotheses, method, and limitations with what the dataset can validly support.

For a Hong Kong-focused project, this approach could support topics such as affordability, public-housing mobility, housing satisfaction, overcrowding, spatial inequality, ageing households, or the relationship between tenure and wellbeing—provided the selected dataset includes properly documented variables for those constructs.