A marketing project in Nigeria usually tests whether one marketing activity affects one consumer outcome. You write the research questions by turning each specific objective into a question, and the hypotheses by stating, for each question, that no significant relationship or effect exists (H0) and that one does (H1). Three to five matched sets, each naming a measurable variable, is the standard.
The rest of this page is the worked detail, using the sectors Nigerian marketing projects are actually written about: fast-moving consumer goods, telecoms, banks and the small retailers of Lagos, Onitsha and Kano.
What does a marketing project in Nigeria usually test?
Almost every undergraduate marketing project in a Nigerian department is built on one of a small number of relationships between a marketing activity and a consumer response. The activity is the independent variable and the response is the dependent variable:
| Marketing activity (independent variable) | Consumer response (dependent variable) | Typical setting |
|---|---|---|
| Social media marketing | Brand awareness, purchase intention | Undergraduates, young consumers in a named university or city |
| Sales promotion (discounts, bonus packs, raffles) | Consumer patronage, impulse buying | Supermarket or open-market shoppers |
| Product packaging and labelling | Purchase decision | Buyers of a named consumer brand |
| Service quality (SERVQUAL dimensions) | Customer satisfaction, loyalty | Bank or telecom customers in a branch or town |
| Celebrity endorsement, advertising appeal | Buying behaviour, brand preference | Users of a named product category |
| Pricing strategy | Patronage, perceived value | Small and medium retailers, market traders |
| Corporate social responsibility | Corporate image, customer loyalty | Host communities, customers of a named firm |
| Relationship marketing | Customer retention | Customers of a named service provider |
The reason this matters for your hypotheses is that a hypothesis needs two variables, and a topic that names only one (“An appraisal of social media marketing in Nigeria”) has nothing to hypothesise. If your topic is still in that shape, it will be sent back before the hypotheses are ever read; the repair is in the piece on getting a project topic approved after a rejection.
How do you turn a specific objective into a research question?
Mechanically. Each specific objective in Chapter One begins with a verb of intent (to examine, to determine, to ascertain); the research question restates it as a question about the same two variables, in the same order, with no new words. Take a project titled Effect of social media marketing on the brand awareness of selected fast-moving consumer goods among undergraduates of the University of Lagos:
| Specific objective | Research question |
|---|---|
| To examine the effect of social media advertising on the brand awareness of selected FMCG brands among UNILAG undergraduates | What is the effect of social media advertising on the brand awareness of selected FMCG brands among UNILAG undergraduates? |
| To determine the relationship between influencer endorsement and brand recall of the selected brands | What is the relationship between influencer endorsement and brand recall of the selected brands? |
| To ascertain the extent to which user-generated content influences purchase intention for the selected brands | To what extent does user-generated content influence purchase intention for the selected brands? |
The discipline is one objective, one question, same variables, same order. A research question that introduces a variable the objective never mentioned is the first thing a supervisor circles. The section order in which objectives and questions sit inside the chapter is set out in the guide to writing Chapter One section by section.
How do you write the null and alternative hypotheses?
Nigerian marketing departments overwhelmingly state hypotheses in the null form, sometimes with the alternative beside it. The null asserts that nothing is going on: no significant effect, no significant relationship, no significant difference. The alternative asserts the opposite. The word “significant” belongs in the hypothesis and nowhere in the research question. For the three objectives above:
H01: Social media advertising has no significant effect on the brand awareness of selected FMCG brands among UNILAG undergraduates.
H11: Social media advertising has a significant effect on the brand awareness of selected FMCG brands among UNILAG undergraduates.H02: There is no significant relationship between influencer endorsement and brand recall of the selected brands.
H12: There is a significant relationship between influencer endorsement and brand recall of the selected brands.H03: User-generated content does not significantly influence purchase intention for the selected brands.
H13: User-generated content significantly influences purchase intention for the selected brands.
Three verbs carry three different shapes, and the shape decides the test later. “Effect” and “influence” imply that one variable changes another, which points to regression. “Relationship” implies that two variables move together without a claim about cause, which points to correlation. If you write “effect” and then run a correlation in Chapter Four, the mismatch will be raised at the defence.
Which variables go into each hypothesis, and how do you make them measurable?
A hypothesis is testable only when both variables have an operational definition: a statement of exactly what will be measured and how. Marketing constructs are abstract, so this is where most projects become untestable without anyone noticing until Chapter Four.
| Construct | Operational definition for the project | Measured by |
|---|---|---|
| Social media advertising | Exposure to paid brand posts on Instagram, X, TikTok and Facebook in the last month | Five Likert items on frequency and attention, summed to a score |
| Brand awareness | Ability to recognise and recall the selected brands | Unaided recall count plus four Likert items on recognition |
| Influencer endorsement | Exposure to and trust in brand content from named influencers | Four Likert items on exposure and credibility |
| Purchase intention | Stated likelihood of buying the brand in the next three months | Three Likert items on intention |
Keep to two variables per hypothesis. A third variable that changes the strength of the relationship, income or gender for instance, is a moderating variable and gets its own hypothesis (“There is no significant difference in brand awareness between male and female undergraduates”), which is a difference hypothesis and meets a different test. When a construct such as service quality has named dimensions, as SERVQUAL’s tangibles, reliability, responsiveness, assurance and empathy do, you may write one hypothesis per dimension, but say so in the objectives first.
Which theory should the hypotheses come from?
A hypothesis reads as arbitrary unless a theory in Chapter Two predicts it. Nigerian marketing panels expect the hypotheses to be consistent with the theory the study is anchored on, and they ask about it. The theories that fit the relationships above are a small set: the theory of planned behaviour (Ajzen, 1991) for purchase intention; the technology acceptance model (Davis, 1989) for mobile money, online shopping and app adoption; the SERVQUAL model (Parasuraman, Zeithaml and Berry, 1988) for service quality and satisfaction; customer-based brand equity (Keller, 1993) for awareness and image; the AIDA model for advertising effects; and the marketing mix for pricing and promotion topics. Ajzen’s and Davis’s papers are single journal articles with DOIs and were checked against Crossref on 30 August 2026. How to present the theory and write the paragraph that connects it to your variables is covered, for a sister discipline, in the guide to writing the theoretical framework of an education project; the four-move shape is the same in marketing.
Which statistical test will each hypothesis meet in Chapter Four?
Deciding the test while you write the hypothesis is the single habit that prevents the Chapter Four crisis, because it forces the variables into a measurable form now.
| Hypothesis shape | Wording cue | Test for Likert-score data | Test for categorical data |
|---|---|---|---|
| One variable affects another | effect, influence, impact | Simple or multiple linear regression | Chi-square test of independence |
| Two variables move together | relationship, association | Pearson correlation (Spearman if not normal) | Chi-square |
| Two groups differ | difference between male and female, urban and rural | Independent samples t-test | Chi-square |
| Three or more groups differ | difference across age groups, income levels | One-way ANOVA | Chi-square |
The decision rule is stated once in Chapter Three and applied to every hypothesis: reject H0 where the p-value is below 0.05, otherwise fail to reject. The full matching of question shape to test, with the assumption checks and the substitute when an assumption fails, is in the answer to which statistical test to use for a final year project, and the table layout for reporting each result is in the guide to writing Chapter Four.
How many hypotheses should a marketing project have?
As many as there are specific objectives, which in a Nigerian undergraduate project means three to five. Fewer than three reads as thin; more than five is rarely completed properly in one session and produces a Chapter Four that runs out of pages. Number them H01 to H0n in the order of the objectives, and keep that order in Chapter Four so the panel can follow objective, question, hypothesis and table down the same numbered line.
Which faults get the hypotheses returned?
- A hypothesis that does not match an objective. Every H0 must trace to a numbered objective with the same variables. An orphan hypothesis is the commonest fault.
- “Significant” in the research question. Research questions ask; hypotheses test. Keep the statistical word in the hypothesis only.
- A variable nobody can measure. “Brand image” with no operational definition and no items on the questionnaire cannot be tested, and the supervisor knows it before you do.
- Hypotheses about the company instead of the consumer. “Guinness Nigeria’s promotional budget has no significant effect on its market share” needs internal company data no student can obtain. Hypothesise about respondents you can reach.
- Directional wording in the null. “Social media marketing has no positive effect” is a one-tailed claim most departments do not expect from an undergraduate. Use “no significant effect” and let the test be two-tailed unless your supervisor asks otherwise.
- Verb and test that disagree. “Effect” tested with correlation, or “relationship” tested with a t-test. Pick the verb for the test you will run.
A complete worked set for a telecoms service quality project
Topic: Effect of service quality on customer satisfaction among MTN subscribers in Enugu metropolis. Anchored on the SERVQUAL model, which names five dimensions of service quality, so the objectives are written per dimension and the hypotheses follow.
Objective 1: To examine the effect of network reliability on customer satisfaction among MTN subscribers in Enugu metropolis.
RQ1: What is the effect of network reliability on customer satisfaction among MTN subscribers in Enugu metropolis?
H01: Network reliability has no significant effect on customer satisfaction among MTN subscribers in Enugu metropolis.Objective 2: To determine the effect of responsiveness of customer care on customer satisfaction among the subscribers.
RQ2: What is the effect of responsiveness of customer care on customer satisfaction among the subscribers?
H02: Responsiveness of customer care has no significant effect on customer satisfaction among the subscribers.Objective 3: To ascertain whether customer satisfaction differs significantly between prepaid and postpaid subscribers.
RQ3: Does customer satisfaction differ between prepaid and postpaid subscribers?
H03: There is no significant difference in customer satisfaction between prepaid and postpaid subscribers.
Tests, decided now: H01 and H02 by multiple regression with satisfaction as the dependent variable, H03 by an independent samples t-test, all at the 0.05 level. Each construct is measured by a block of Likert items adapted from the SERVQUAL instrument, which means the Chapter Three instrument section can name its source rather than invent one.
Generate the matched set from your own topic
Tesify drafts the objectives, research questions and null hypotheses for your marketing topic as one aligned set, so the variables match row by row and the verb in each hypothesis already points to the test you will run. Rework any row after supervisor corrections without retyping the rest. There is a free plan and no card is required.
Frequently asked questions
What is the difference between a research question and a hypothesis in a marketing project?
The research question asks what the relationship between two variables is; the hypothesis states, in a form that can be tested statistically, that no significant relationship exists (null) or that one does (alternative). Every research question should have one hypothesis pair beneath it with the same variables.
Should I write the null hypothesis or the alternative hypothesis?
Most Nigerian marketing departments require the null form, and many accept the null alone. Where your departmental format asks for both, state the null first and the alternative directly beneath it, numbered H0 and H1 with the same subscript.
How many hypotheses should a marketing project have?
One per specific objective, which usually means three to five. More than five is rarely tested properly in an undergraduate project and produces a Chapter Four that runs out of space.
What is an example of a null hypothesis for a marketing project?
Sales promotion has no significant effect on consumer patronage of selected supermarkets in Ibadan. It names an independent variable, a dependent variable, a population and a location, and it can be tested with data collected from shoppers.
Can a marketing hypothesis be about a company rather than consumers?
Only if the company data is published or you have written access. Hypotheses about a firm’s internal budget, sales volume or market share usually cannot be tested by an undergraduate, so departments prefer hypotheses about consumers or customers you can reach with a questionnaire.
Which statistical test is used to test marketing hypotheses?
It depends on the wording. Effect and influence hypotheses are tested with regression, relationship hypotheses with Pearson or Spearman correlation, difference hypotheses with a t-test or ANOVA, and hypotheses on categorical responses with chi-square, all at the 0.05 significance level.
What does significant mean in a marketing hypothesis?
That the effect or relationship observed in the sample is unlikely to have arisen by chance, judged by a p-value below 0.05. It does not mean large or important. A significant effect can be small, which is why the strength of the relationship is reported as well.
Do research questions and hypotheses have to match the objectives exactly?
Yes. Same variables, same order, same population. A hypothesis that introduces a variable the objectives never mention is the most common reason the section is returned by a supervisor.
Which theories are used to support marketing hypotheses in Nigeria?
The theory of planned behaviour for purchase intention, the technology acceptance model for adoption of digital channels, SERVQUAL for service quality and satisfaction, customer-based brand equity for awareness and image, AIDA for advertising effects, and the marketing mix for pricing and promotion.