For most business FYPs, a direct-effect model is the safest conceptual framework, a mediation model fits when you can explain a mechanism between two variables, a moderation model fits when the effect should differ across groups, and an input-process-output (IPO) layout fits projects built around a programme, system or process. The table below compares the four on the questions supervisors ask first, and the rest of the article shows how to choose, draw and defend one.
How do the four conceptual framework models compare?
The criteria are the ones that decide whether a model survives a proposal defence: the research question it answers, what it needs from your data, how hard it is to analyse and how easily it can be misread.
| Model | Question it answers | What it needs | Analysis difficulty | Main risk | Best for |
|---|---|---|---|---|---|
| Direct-effect | Does X relate to Y? | Validated scales for each variable | Lower: correlation and regression | Looks thin if there is no theory behind the link | First-time researchers, short timelines |
| Mediation | Does X relate to Y through M? | A theory that explains M, a scale for M, enough respondents | Higher: indirect effect testing | Calling it causal when the design is cross-sectional | Projects with a clear mechanism, such as trust or satisfaction |
| Moderation | Does the X to Y link change with W? | A reason why W changes the link, enough cases per group | Medium to higher: interaction terms or group comparison | Choosing W without a reason | Projects that compare groups, such as age, gender or experience |
| IPO | What goes into a process and what comes out? | A defined programme, system or process with measurable inputs and outputs | Varies with the design | Mixing it with a theory-based model without saying so | Training, operations, system implementation or case projects |
Whichever you choose, check your faculty’s guide and your supervisor’s preference, because some faculties expect a theory-based framework and others accept an IPO layout.
What is a conceptual framework, and how is it different from a theoretical framework?
A conceptual framework is your own diagram of the variables in the study and the relationships you will test. A theoretical framework is the established theory that explains why those relationships should exist. The conceptual framework is built on top of the theory. If you are not sure where the boundary lies, read the difference between a theoretical framework and a conceptual framework before you draw anything.
In a business FYP, the diagram normally appears at the end of Chapter 2 and leads straight to the hypotheses. Every arrow in the diagram should match one hypothesis, and every box should match one scale in your questionnaire.
When should you choose a direct-effect model?
A direct-effect model has independent variables on the left, one dependent variable on the right and an arrow for each hypothesis. Example: social media marketing, service quality and price perception, each linked to customer purchase intention.
- Choose it when your problem statement asks which factors relate to an outcome, your timeline is a single semester and your population is one group.
- Strengths. It is easy to draw, easy to defend and easy to test with correlation and multiple regression.
- Weakness. If you only list variables from other studies, examiners ask why these variables and what mechanism connects them. A theory in Chapter 2 answers that.
A direct-effect model with three or four independent variables and a clear theory is usually easier to analyse and to defend than a complicated model you cannot test with the time and sample you have.
When should you choose a mediation model?
A mediation model adds a variable between the cause and the outcome: X affects M, and M affects Y. Baron and Kenny (1986), Journal of Personality and Social Psychology, 51(6), 1173 to 1182, is the classic reference for the moderator-mediator distinction. Preacher and Hayes (2008), Behavior Research Methods, 40(3), 879 to 891, describe asymptotic and resampling strategies for assessing indirect effects in multiple mediator models, and are commonly cited for bootstrap tests of mediation.
Illustrative example: social media marketing leads to customer trust, and trust leads to purchase intention. The mediation question is whether the link between marketing and intention runs partly or wholly through trust.
- Choose it when you have a theory that explains the mediator and a validated scale for it.
- Strengths. It shows a mechanism, which supervisors value when it is justified.
- Weaknesses. It needs more data and more analysis steps. In a cross-sectional survey you cannot show that M comes between X and Y in time, so write “is consistent with mediation”, not “proves mediation”.
When should you choose a moderation model?
A moderation model asks whether the strength or direction of an effect depends on a third variable. Example: the link between social media marketing and purchase intention may be stronger for younger consumers than older ones. Here age is the moderator.
- Choose it when there is a theoretical or practical reason to expect different effects for different groups, and your sample has enough people in each group.
- Strengths. It makes a generic model more specific and gives a more interesting discussion chapter.
- Weaknesses. A moderator chosen only because it is available in the questionnaire, such as gender or age, looks arbitrary. Explain why the effect should differ.
Baron and Kenny (1986) also distinguish moderators from mediators, which is useful if your supervisor asks you to justify the difference.
When should you choose an IPO model?
An input-process-output layout organises the framework as inputs (resources, programme features, independent variables), a process (what the organisation or system does) and outputs (the results you measure). It suits projects about training, operations, systems implementation or a case company.
Illustrative example for a case-based FYP: inputs are staff skills and available equipment; the process is a new order-handling procedure; the outputs are order accuracy and customer complaints. Each output needs an indicator you can measure from records or a survey.
- Choose it when your project is about a defined process or programme rather than about relationships between attitudes.
- Weakness. It can be vague about why the inputs should lead to the outputs. If your faculty expects theory, add a theory to explain the process, and say so.
Not every faculty uses the IPO layout, so confirm with your supervisor before you adopt it. If your project is a case company, our guide to turning your industrial training report into a business administration FYP shows how to link a case to a thesis structure.
How do you choose among the four?
Answer these questions in order. The first answer that fits is usually your model.
- Does your problem statement ask about factors that relate to one outcome? Start with direct-effect.
- Can you name a mechanism and find a validated scale for it? Consider mediation.
- Do you expect the effect to differ across groups you can sample in enough numbers? Consider moderation.
- Is the project about a process or programme with measurable inputs and outputs? Consider IPO.
- Can you analyse it with the software and time you have? If not, simplify. Our guide to which statistical test to use for your FYP data matches model type to test.
A practical rule: one model, one dependent variable, three or four paths. A framework with ten arrows is difficult to measure and harder to defend.
How do you draw the framework so examiners accept it?
- Put independent variables on the left, the dependent variable on the right. Mediators go between them, and moderators above or below the arrow they change.
- Label every arrow with a hypothesis code, H1, H2 and so on, and make the codes match the text.
- Use the exact variable names that appear in your objectives and questionnaire.
- Add control variables in a separate box if you plan to use them.
- Write a caption and a short paragraph explaining the diagram in words; never leave the figure alone.
- Cite the theory that supports each arrow, using the original source.
An illustrative caption: “Figure 2.1. Conceptual framework: social media marketing, service quality and price perception predict purchase intention, with trust as a mediator (H4) and age as a moderator (H5).” Notice the caption names every relationship so the reader can check it against the hypotheses.
How does the framework connect to the rest of the FYP?
The framework is the hinge of the thesis. It follows the problem statement and objectives, it determines the variables you define, and it drives the questionnaire and the analysis. See how to write a problem statement for research for the starting point, how to write operational definitions for a marketing FYP for turning each box into something measurable, and how to write a literature review on online purchase intention for an example of building the evidence for each arrow.
Before you submit the proposal, run a short consistency check. Count the arrows in the diagram, count the hypotheses in the text and count the scales in the questionnaire; the three numbers should agree, allowing for scales used for control variables. Then check that each hypothesis is phrased with the same variable names that the diagram uses. Mismatches at this stage are common, and they are easy to fix before an examiner reads the chapter, but difficult to explain after data collection.
What mistakes make a framework fail at proposal defence?
- Variables without a theory. The diagram looks like a list of everything found in the literature.
- Arrows without hypotheses. Every arrow needs a hypothesis and a test.
- Too many paths. More than a handful of paths needs a large sample and advanced analysis.
- Causal claims from correlational data. Be careful with the verbs: relate, associate, predict.
- Different names in the figure and the questionnaire. Keep the same terms throughout.
- Copying a framework from another study unchanged. Adapt it to your population and cite the source.
How can Tesify help you turn the framework into chapters?
Once the model is fixed, Tesify helps you structure your chapters around it. It is used by 9,000+ students across 15,000+ chapters, and the thesis stays 100% written by you.
Frequently asked questions
What is the best conceptual framework model for a business FYP?
For most FYPs, a direct-effect model grounded in a clear theory is the safest. Choose mediation, moderation or IPO only when your problem statement and data support them.
What is the difference between mediation and moderation?
A mediator explains how or why X affects Y by sitting between them. A moderator changes the strength or direction of the X to Y link. Baron and Kenny (1986) discuss the distinction.
Do I need a theory behind my conceptual framework?
Yes, in most business faculties. The theory explains why the relationships in your diagram should exist, and it is cited from the original source in Chapter 2.
What does IPO mean in a conceptual framework?
Input, process, output. It organises the study around what goes into a process, what the process is and what results from it. Check with your supervisor that your faculty accepts it.
How many variables should my framework have?
Keep it manageable: one dependent variable and a small number of independent variables, plus a mediator or moderator only if justified. Your sample size and analysis skills set the limit.
Can I use more than one dependent variable?
Yes, but each adds hypotheses and analysis. For a single-semester FYP, one dependent variable is easier to defend.
Where does the conceptual framework go in the FYP?
Usually at the end of Chapter 2 after the literature review and theory, followed by the hypotheses. Check your faculty’s template.
How do I test mediation in an FYP?
Baron and Kenny (1986) describe the classic steps, and Preacher and Hayes (2008) describe bootstrap strategies for indirect effects. Ask your supervisor which approach your faculty expects.
Can I copy a framework from a journal article?
You can adapt one if you cite it and change it to fit your population and variables. Copying it unchanged is risky, because your measures and context differ.
Should the diagram include control variables?
If you plan to control for variables such as age or gender, show them in a separate box so the main relationships stay clear.