What Is the Research Subject? How to Define and Write It in Your Thesis
What Is a Research Subject
A research subject is the issue, phenomenon, relationship or characteristic that your study aims to observe and explain. In a quantitative thesis, the research subject is usually represented through research variables and the relationships between them. For example, a study on the effect of service quality on customer satisfaction has the relationship between service quality and customer satisfaction as its research subject, not all the customers who completed the questionnaire.
You can clarify it with three questions: What issue is the study examining, which variables make up that issue, and which relationship between the variables do you want to test? Your answers should match the thesis title, objectives, research questions, hypotheses and questionnaire.
A general template is: The research subject of this study is [the relationship or phenomenon] between [the independent variable] and [the dependent variable] in the context of [the research context]. If the study includes a mediator or moderator, state its role clearly instead of only listing the variable name.
The research subject does not have to be a group of people. Workers, students or customers provide the data and are usually described as survey respondents or units of analysis. The research subject is what you use that data to draw a conclusion about.
If you are building your proposal from the beginning, you can also review scientific research and What is scientific research to place this term correctly throughout the research process.
Why the Research Subject Matters in Quantitative Research
Defining the research subject correctly helps you decide what data to collect. When the subject is the effect of perceived usefulness on usage intention, the questionnaire must measure both constructs. Items that only describe demographic characteristics will not be enough to test the hypotheses.
The research subject also determines the model and analysis method. A model with latent variables measured by multiple items may require Cronbach's Alpha, EFA and regression or SmartPLS. A study that only describes a percentage or level may not need a causal model.
While reading the literature, record four things: the construct being studied, how it is measured, the relationship being tested and the context in which it is applied. This gives your reading a clearer purpose than simply collecting as many papers as possible. If your study uses a questionnaire and .sav or .csv data, the section on quantitative research can help you distinguish descriptive research from research that tests relationships.
The research subject is also a basis for checking whether the study is internally consistent. If the thesis title concerns purchase intention but the objective measures satisfaction, you have two different research subjects. If the hypothesis concerns customers in Ho Chi Minh City but the sample also includes people who have never used the service, the research design needs to be reviewed.
Write the research subject before finalising the questionnaire. This helps you avoid taking an existing questionnaire, keeping all its variables and then trying to explain them in Chapter 1.
How Much Research Subject Is Enough
A research subject does not have a numerical cutoff like Cronbach's Alpha or R². It is methodologically adequate when it clearly describes the issue being studied, the relevant variables, the relationship to be tested and the context of application. The table below is a design checklist, with a suitable methodological source for each part.
| Content to check | Signs that it is adequate | How to cross-check and source |
|---|---|---|
| Research concept | Clearly states the phenomenon or relationship to be explained | Compare it with the approach to business research design in (Nguyễn Đình Thọ, 2011) |
| Research variables | Includes independent, dependent and additional variables when the model requires them | Each variable must have a suitable scale and items, following the two-step measurement and structural model procedure in (Anderson and Gerbing, 1988) |
| Unit of analysis | Clearly identifies individuals, households, businesses or organisations | The unit of analysis must match the question and the collected data, following the methodological guidance in (Nguyễn Đình Thọ, 2011) |
| Relationship to be tested | Can be converted into an objective, question or hypothesis | The hypothesis must be linked to the relationships between variables in the model, according to (Bollen, 1989) |
| Measurement approach | Each construct has corresponding items or indicators | For a quantitative scale, you can use Likert based on the foundation provided by (Likert, 1932) |
| Research sample | Respondents have relevant experience with the subject | Sample size should be considered in relation to the number of items, often using 5 to 10 observations per item as a reference (Hair et al., 2010) |
| Contextual scope | States the location, time and surveyed group | The scope must limit your conclusions to the context in which the data were collected, according to (Nguyễn Đình Thọ, 2011) |
This table is a design checklist, not a scoring system. Do not write that “the research subject achieves 80%” because a research subject is not validated by one index. Measures such as KMO, factor loading and Cronbach's Alpha evaluate the data and scale after data collection. They do not prove that the description of the research subject is correct.
If the study includes several groups, state whether you are examining the overall relationship or comparing groups. For example, “the effect of shopping experience on customers' repurchase intention in Hanoi” is different from “a comparison of that effect between customers under 25 and customers aged 25 and above.”
How to Read the Research Subject in the Output
The research subject does not appear as a separate line in SPSS output. You identify it by connecting three parts: the variable names in Variable View, the variables entered into the analysis and the group described in Frequencies, Descriptive Statistics, Correlations or Coefficients.
For example, suppose you are studying the effects of perceived usefulness and perceived ease of use on intention to use an online learning application. In SPSS, the items may be named PU1 to PU4, PEU1 to PEU4 and BI1 to BI3. After assessing the scales, you calculate the mean score for each construct and then enter PU, PEU and BI into the regression.
The table below is illustrative output, not the result of a real study. The column names follow a common SPSS output format so you know which part to read.
| SPSS output table | Row or column to read | Illustrative data | Meaning for the research subject |
|---|---|---|---|
| Descriptive Statistics | N for PU, PEU, BI | 248 | There are 248 valid observations entered into the analysis |
| Correlations | Pearson Correlation: PU with BI | 0.482 | The two constructs have a linear relationship in the illustrative sample |
| Model Summary | R Square | 0.361 | PU and PEU explain part of the variation in BI |
| ANOVA | Sig. | 0.000 | The illustrative regression model is statistically significant according to the p-value rule |
| Coefficients | PU, Standardized Beta | 0.341 | PU has a positive effect coefficient in the illustrative model |
| Coefficients | PEU, Standardized Beta | 0.219 | PEU has a positive effect coefficient in the illustrative model |
| Coefficients | PU, Sig. | 0.000 | There is statistical evidence to consider the hypothesis that PU affects BI |
| Coefficients | PEU, Sig. | 0.006 | There is statistical evidence to consider the hypothesis that PEU affects BI |
Do not use the table above to write results for your own data. Replace it with the values in the file currently open, and check the number of observations in every table. If N in Descriptive Statistics is 248 but one variable contains many missing values, the number of observations in the regression may be lower.
When reading the output, return to the original question: which variable affects which variable, and in which group of people are you testing that relationship? A neat Coefficients table that uses the wrong sample still does not answer the correct research question.
What to Do When the Research Subject Is Not Adequate
If the description is still too broad, work through the steps below. This lets you correct the design instead of trying to explain a study that has already moved off course.
Rewrite the Subject as a Specific Relationship
Replace “studying online shopping behaviour” with “studying the effects of trust and perceived risk on online purchase intention.” The revised sentence tells you which variables need to be measured and which model needs to be tested.
Compare the Subject with the Objectives and Hypotheses
Each specific objective should answer one part of the research subject. If an objective says “propose solutions to improve satisfaction” but the model does not measure any factor related to satisfaction, revise the objective or establish the basis for the model before running the data.
Recheck the Sampling Criteria
Respondents must have experience relevant to the phenomenon. For a study on repurchase intention, the screening question should establish that respondents have purchased the product or used the service during the period you specify. If the wrong sample has already been collected, changing a variable name in SPSS will not solve the problem.
Narrow the Scope of the Conclusion
If your data come only from students at one university, state your conclusion for that group instead of writing about all Vietnamese consumers. This is a defensible response when the limitation comes from resources and sample design, as long as you state it clearly in the thesis.
Record Every Change
Save the proposal, questionnaire and data file before each revision. When your supervisor asks why the research subject differs from the first version, you can explain the change using the objectives, sample criteria and variables that were actually measured.
Distinguishing the Research Subject from the Research Object and Scope
These three terms are often mixed together in one Chapter 1 sentence. The research subject answers “what issue are you studying?” The research object or survey population answers “from whom are you collecting data?” The research scope answers “where, when and to what extent are you studying it?”
For the study “The effect of service quality on the satisfaction of customers using digital banking in Hanoi,” the research subject is the relationship between service quality and satisfaction. The survey population is customers who have used digital banking. The geographical scope is Hanoi. The time scope is the period when the data were collected. The content scope is the service-quality dimensions selected for the model.
You can present it as follows:
- Research subject: the relationship between the service-quality dimensions and satisfaction.
- Survey population: customers who have used digital banking services.
- Geographical scope: customers living or transacting in Hanoi.
- Time scope: the period when the questionnaire was distributed and collected.
- Content scope: the variables and hypotheses specified in the research model.
If the study includes interviews or qualitative data in its reference materials, keep your implementation consistent with your own quantitative design. What is qualitative research can help you understand the difference in approach, but you should not add a method outside your design to the thesis simply because a reference study used it.
Common Errors
The first error is writing “the research subject is students.” This usually describes the respondents but does not say whether you are studying their behaviour, perceptions or a relationship between variables.
The second error is treating the research subject as the thesis topic. The thesis title may be long, while the subject should be condensed into the central issue and relationship. Separate the title into context, variables and relationship before writing the definition sentence.
The third error is using the scope in place of the subject. “In Ho Chi Minh City in 2026” only states the location and time. It does not say which phenomenon you are studying.
The fourth error is changing variables after viewing the results without recording the reason. If you remove an item because its corrected item-total correlation is low or its factor loading does not meet the criterion, save the output from each round and explain the decision using the criterion you selected. Cronbach's Alpha at 0.7 or above is generally accepted according to (Nunnally, 1978), while an exploratory scale can refer to a level from 0.6 according to (Hair et al., 2010). These thresholds assess the scale. They do not replace the task of defining the research subject.
The final error is drawing conclusions beyond the sample. Data from one customer group, one university or one locality do not automatically represent the entire market. Put the research subject, survey population and scope side by side before writing Chapter 4.
Frequently asked questions
What is a research subject in a thesis?
It is the issue, phenomenon or relationship that the thesis aims to explain. In quantitative research, the subject is usually represented by variables and hypotheses, such as the effect of service quality on satisfaction.
Are the research subject and survey population the same?
They are related but different. The survey population is the person or unit that provides the data, while the research subject is the issue about which you draw a conclusion from that data.
Do I need to include the location and time in the research subject?
The location and time belong to the research scope. You can place them in the same introductory section to clarify the context, but the subject description must still state the central issue or relationship.
Is the research subject the dependent variable?
The dependent variable is only one component of the subject in many quantitative studies. The subject may include independent variables, a dependent variable, a mediator, a moderator and the relationships between them.
What should I do if the research subject is too broad?
Narrow it to a specific context, survey group, model and relationship. Then check that the thesis title, objectives, hypotheses, questionnaire and analysis method still match one another.
Can I revise the research subject after collecting data?
You can adjust the wording if the data still measure the same issue, but do not change the subject simply to hide an unsuitable sample or an unexpected result. If you need to remove items or narrow the conclusion, record the reason and state the research limitation.
How to write this in your thesis: Open your proposal and data file, write one sentence using the pattern [relationship] between [variables] in [context], then compare it with your questionnaire and variable list before running the next analysis. If you need to check the model, variables and real data in SPSS or SmartPLS, you can use the M4 analysis module.