Lecture note on positivism for housing studies dissertation project study

Lecture note on positivism for housing studies dissertation project study

 

In the context of study of research philosophies, please clarify four main ideas of positivism and offer 4 main how-to advices on research methods practices in the context of Housing Studies dissertation project works.

Positivism treats housing issues as phenomena that can be studied through observable evidence, measurement, and systematic testing. In a Housing Studies dissertation, it is most suitable when your purpose is to estimate patterns, test relationships, or assess whether one measurable factor is associated with another—for example, whether overcrowding is associated with lower housing satisfaction.

Four main ideas of positivism

1.    An objective reality exists independently of the researcher
Positivism assumes that housing conditions—such as rent, floor area, tenure, waiting time, household size, or reported satisfaction—exist independently of a researcher’s opinions. The researcher should therefore minimise personal values and interpretation when collecting and analysing evidence.

2.    Valid knowledge comes from empirical observation and measurement
Claims should be based on data that can be observed, recorded, counted, or measured, rather than intuition or untested assumptions. In housing research, this may mean using administrative records, Census data, structured questionnaires, rent data, or standardised housing-condition indicators.

3.    Social phenomena can be explained through variables and causal patterns
Positivist research aims to identify regular relationships—often framed as cause-and-effect or statistically testable association—between clearly defined variables. A study might test whether rent burden, dwelling size, tenure, and commuting time predict housing satisfaction. It seeks explanation and, where justified, prediction rather than primarily interpreting personal meaning.

4.    Research should be systematic, transparent, and replicable
A positivist study normally specifies its concepts, sampling, measurement rules, hypotheses, and statistical procedures in advance. Another researcher should be able to inspect the process and, ideally, repeat it using the same design and obtain comparable results. This is why standardised instruments, reliability checks, and explicit analysis procedures matter.

A useful distinction: positivism is not simply “using a questionnaire.” It is the philosophical logic behind a design: an assumed measurable reality, deductive hypothesis testing, researcher distance, and a goal of generalisable explanation. A questionnaire can also be used within other philosophies, depending on how it is designed and interpreted.

Four research-practice advices

1. Convert broad ideas into measurable variables

Start with a focused explanatory question and define every central concept operationally.

For example:

“To what extent do rent burden, overcrowding, tenure, and neighbourhood accessibility predict housing satisfaction among private renters in Hong Kong?”

Concept	Possible operational measure
Housing satisfaction	Mean score from 4–6 Likert-scale items, such as satisfaction with space, condition, safety, and location
Rent burden	Monthly rent ÷ monthly household income × 100
Overcrowding	Persons per room, persons per square metre, or a recognised overcrowding threshold
Tenure	Private renter, public renter, owner-occupier, subsidised owner
Accessibility	Minutes to work/study or distance/time to the nearest MTR station

Use established definitions and validated measures wherever possible. In Hong Kong, the 2021 Population Census includes variables such as accommodation floor area, tenure, rent, mortgage payments, household characteristics, and whether a quarter is a subdivided unit—useful benchmarks for defining or contextualising your measures.

2. State testable hypotheses before collecting data

Derive hypotheses from theory and prior evidence, then test them with data. Avoid writing hypotheses only after looking at results.

Examples:

• H1: A higher rent-to-income ratio is associated with lower housing satisfaction.
• H2: Households experiencing overcrowding report lower housing satisfaction than non-overcrowded households.
• H3: After controlling for household income and household size, public-rental and private-rental households differ significantly in housing satisfaction.

This follows a deductive logic:

Theory/literature→ Hypotheses→ Data collection→ Statistical test→ Support, rejection, or revision

Be cautious in your wording. A cross-sectional survey can usually demonstrate an association, not definitive causation. For example, “rent burden is significantly associated with lower satisfaction” is safer than “rent burden causes dissatisfaction,” unless the design supports a causal claim.

3. Design sampling and data collection to reduce bias

Define the population precisely—for example, “Hong Kong residents aged 18 or above who are the household head or a household member responsible for housing decisions and currently rent private accommodation.”

Then:

• Use probability sampling where feasible, such as stratified sampling by district, tenure, or housing type.
• If access limitations require convenience or online sampling, state this honestly and avoid claiming that the results represent all Hong Kong households.
• Pilot your questionnaire with approximately 10–20 relevant respondents to identify unclear wording, missing response options, excessive length, and sensitive questions.
• Use neutral, single-focus questions. Do not ask: “Do high rents and poor conditions make you dissatisfied?” Split this into separate measurable items.
• Record eligibility, recruitment channels, response rate, missing data rules, and fieldwork dates.

Hong Kong official household surveys illustrate the value of an explicit sampling frame: the General Household Survey samples from an address list and collects information on tenure, rent, housing type, household size, and income. Its design does not cover institutional residents or people living aboard vessels—an example of why a dissertation must state who is and is not included. Thematic Household Surveys similarly use an address-based sampling frame and random selection of quarters.

4. Use appropriate statistical tests and report quality checks

Match the analysis to the question and measurement level:

• Use descriptive statistics first: frequencies, percentages, means, medians, and standard deviations.
• Use chi-square tests for relationships between categorical variables, such as tenure type and whether overcrowding is present.
• Use t-tests or ANOVA to compare mean housing-satisfaction scores across two or more groups.
• Use correlation to assess the direction and strength of association between two continuous variables, such as rent burden and satisfaction.
• Use multiple regression when assessing whether several factors jointly predict an outcome while controlling for likely confounders such as income, age, household size, district, or tenure.

For a multi-item scale such as housing satisfaction, check internal consistency—for example, with Cronbach’s alpha—and report it. Also report sample size, missing-data treatment, significance level, effect size where possible, and limitations. Positivist work values transparent procedures because results should be open to scrutiny and replication.

A feasible dissertation design

For a part-time MBA Housing Studies project, a manageable positivist design could be:

• Aim: Examine factors associated with housing satisfaction among private renters in selected Hong Kong districts.
• Design: Cross-sectional quantitative survey, supplemented by Census or C&SD descriptive statistics.
• Sample: 150–250 eligible adult renters, with quotas or stratification by district and dwelling type if a probability sample is impractical.
• Instrument: A structured questionnaire with demographic items, housing characteristics, rent burden, overcrowding, commuting time, neighbourhood ratings, and a multi-item housing-satisfaction scale.
• Analysis: Descriptive statistics; reliability test for the satisfaction scale; correlation; multiple regression.
• Claim boundary: Identify statistically supported associations in the achieved sample, while acknowledging potential self-selection bias and the inability of cross-sectional data to establish causality.

This approach aligns your ontology (a measurable external housing reality), epistemology (knowledge obtained through observation), methodology (deductive quantitative inquiry), and methods (survey, secondary data, and statistics).