Business Research Methods

The students collected data from the scan room without informing the mothers, so there was no informed consent. This breaches participants' autonomy, privacy and confidentiality in a sensitive medical context.

A theoretical framework is the set of established theories and concepts that underpin a study. It explains why the variables are expected to be related and guides the research questions, hypotheses and analysis.

Sampling is the process of selecting a subset of units (a sample) from the whole population so that findings from the sample can be generalised to the population. It saves time and cost compared with studying every unit.

An academic proposal is prepared by students or scholars to meet a degree requirement, with a strong theoretical and literature base, formal structure and an evaluation by a university committee. A non-academic proposal is prepared for an organisation or funder, focuses on solving a practical problem, and stresses budget, timeline and expected benefits.

At the 99% confidence level the significance level is 0.01. The researcher rejects the null hypothesis when the p-value is less than 0.01 ($p < 0.01$), and fails to reject it otherwise.

A mediating variable is an intervening variable that explains how or why an independent variable affects a dependent variable. The independent variable influences the mediator, which in turn influences the dependent variable.

Example: training (independent) raises employee skills (mediator), which raises job performance (dependent).

This is a Type I error.

• The null hypothesis is that the accused is innocent ($H_0$: innocent).
• The judge sentences the person, which means $H_0$ is rejected.
• The person is actually innocent, so $H_0$ is true.
• Rejecting a true null hypothesis is a Type I error (false positive), with probability $\alpha$.

A Type II error would be freeing a person who is actually guilty (failing to reject a false null).

Observation is the systematic watching and recording of behaviour, events or conditions as they occur. It is used in the following circumstances:

1. Behaviour that people cannot or will not report accurately. Examples are customer movement in a store, or children or animals who cannot answer questionnaires.
2. Actual rather than stated behaviour is needed. Observation avoids recall bias and social desirability bias in what respondents say they do.
3. Physical or process data is required. Examples are queue lengths, workplace conditions and employee-customer interaction.
4. Exploratory studies. It helps in understanding a context before designing a survey.

It is justified because it gives first-hand, direct and more objective data. Its limits are time, cost, observer bias and inability to capture motives.

All ordinal scales are nominal scales is the true statement.

• A nominal scale only classifies items into categories with no order (for example gender or religion).
• An ordinal scale also classifies items into categories, and in addition ranks them in order (for example low, medium, high).
• So every ordinal scale has the classification property of a nominal scale plus order. Every ordinal scale is therefore also nominal.
• The reverse is false: a nominal scale has no order, so it cannot be an ordinal scale.

Literature review is the critical study of earlier research on a topic. Its role in academic research is as follows:

1. Identifies the research gap. It shows what is already known and what remains unstudied, which justifies the study.
2. Avoids duplication. It prevents repeating work already done.
3. Builds the theoretical base. It provides theories and concepts for the conceptual framework and hypotheses.
4. Guides methodology. It shows suitable research designs, variables, measures and statistical tools used before.
5. Sharpens the problem. It helps in framing the research problem, objectives and questions.
6. Supports interpretation. The findings can be compared with earlier studies in the discussion.

A research paradigm is a set of basic beliefs and assumptions about the nature of reality, knowledge and method that guides how research is done.

On the basis of reasoning used:

1. Deductive (positivist) paradigm. The researcher begins with a theory, derives hypotheses and tests them with data, moving from general to specific. It is usually quantitative.
2. Inductive (interpretivist) paradigm. The researcher begins with observations and data and builds patterns, themes and theory, moving from specific to general. It is usually qualitative.
3. Abductive (pragmatic) paradigm. The researcher moves between data and theory to find the best explanation, combining both approaches. It is often used in mixed-method research.

Inferential statistics are methods used to draw conclusions about a population from sample data, through estimation and hypothesis testing (for example t-test, ANOVA, regression).

The scientific research process is a systematic sequence of steps for obtaining reliable knowledge.

Diagram

Identify problem
       |
Review literature
       |
Formulate hypotheses / research questions
       |
Design research (variables, sampling, instruments)
       |
Collect data
       |
Analyse data
       |
Interpret results (accept / reject hypotheses)
       |
Report findings
       |
(feedback to the problem and theory)

Steps

1. Identify the problem. Select and define the area of study and the gap to be filled.
2. Review literature. Study earlier research and theory to refine the problem.
3. Formulate hypotheses or research questions. State the expected relationships in testable form.
4. Research design. Decide the approach, population, sampling plan, variables and data collection instruments.
5. Collect data. Gather primary or secondary data in the field.
6. Analyse data. Edit, code, tabulate and apply statistical tests.
7. Interpret results. Decide whether the hypotheses are supported and relate the findings to theory.
8. Report. Present the findings, conclusions and recommendations in a written report. New questions found feed back to the first step.

Purpose of reliability test. Reliability shows the consistency and stability of a measuring instrument: whether it gives the same results when repeated under the same conditions. A researcher tests it to make sure the instrument is free of random error, so that findings can be trusted and repeated. Common measures include Cronbach's alpha (internal consistency), test-retest and equivalent form.

Test-retest reliability versus equivalent form reliability

Purpose of validity test. Validity is the extent to which an instrument measures what it is supposed to measure. A researcher tests it to make sure the data and conclusions are accurate and relevant to the concept studied, and are not distorted by systematic error. A valid instrument must also be reliable, but a reliable instrument need not be valid.

Types of validity

1. Content validity. The instrument covers all aspects of the concept. It is judged by experts. Example: a test of banking knowledge should cover all banking topics.
2. Criterion-related validity. The instrument's scores are compared with an external standard. It has two forms: concurrent validity, where the standard is measured at the same time, and predictive validity, where the instrument predicts a future outcome (for example a selection test predicting job performance).
3. Construct validity. The instrument measures the theoretical construct it claims to measure. It includes convergent validity (related measures correlate) and discriminant validity (unrelated constructs do not correlate).
4. Face validity. At first look, the instrument appears to measure the intended concept. It is the weakest form.

Sampling techniques are classified into probability and non-probability sampling.

Diagram

                 SAMPLING TECHNIQUES
                /                   \
     Probability sampling     Non-probability sampling
     (known chance)           (chance not known)
        |                           |
  - Simple random             - Convenience
  - Systematic                - Judgemental (purposive)
  - Stratified                - Quota
  - Cluster                   - Snowball
  - Multi-stage

Probability sampling gives every unit a known, non-zero chance of selection, so results can be generalised.

• Simple random: each unit has an equal chance, by lottery or random numbers.
• Systematic: every $k$th unit is selected after a random start, where $k = N/n$.
• Stratified: the population is divided into homogeneous strata, and a random sample is drawn from each.
• Cluster: the population is divided into groups (clusters), some clusters are chosen at random, and all units in them are studied.
• Multi-stage: sampling is done in several stages, for example district, then ward, then household.

Non-probability sampling does not give every unit a known chance, so generalisation is limited.

• Convenience: units that are easy to reach are chosen.
• Judgemental (purposive): the researcher selects units by own judgement.
• Quota: fixed numbers are taken from each category until quotas are filled.
• Snowball: initial respondents refer others, used for hard-to-find populations.

(a) Research framework

The study explains what drives customers' favourable and unfavourable experiences in restaurants.

Independent variables                    Dependent variable
-------------------------                -------------------------
Social interaction  ----\
Core service        -----+----------->   Customer experience
Physical environment ---/                (favourable / unfavourable)
                              ^
                              |
                  Moderating variable: customer demographics (age, gender, visit frequency)

• Social interaction: the behaviour, courtesy and responsiveness of employees, and interaction with other customers.
• Core service: the quality of food and beverages, and unexpected extras such as a little more food.
• Physical environment: the ambience, layout, cleanliness, lighting and music.
• Customer experience (dependent): the customer's overall favourable or unfavourable evaluation of the visit.
• Moderating variable: customer characteristics such as age, gender and visit frequency, which may strengthen or weaken the relationships.

(b) Statement of the problem

Restaurants in Nepal are growing quickly, yet many struggle to retain customers. Customer experience is now recognised as a main source of competitive advantage (Pine & Gilmore, 1998). Limbu and Kutal (2022) found that social interaction, core service and physical environment drive experiences in restaurants. However, most studies are qualitative and based on small samples, and the extent to which each driver affects overall experience has not been measured (Shrestha, 2020). Managers therefore lack evidence on where to put resources. This study fills the gap by testing the effect of these drivers on customer experience in restaurants of the study area.

(c) Research questions, objectives and hypotheses

Research questions

1. What is the effect of social interaction on customer experience in restaurants?
2. What is the effect of physical environment on customer experience in restaurants?

Objectives

1. To examine the effect of social interaction on customer experience.
2. To analyse the effect of physical environment on customer experience.

Null hypotheses

• $H_{01}$: Social interaction has no significant effect on customer experience.
• $H_{02}$: Physical environment has no significant effect on customer experience.

(d) Research design

I would use a descriptive and causal (explanatory) cross-sectional research design with a quantitative survey.

• The study describes customers' experience and explains the effect of drivers on it, so a descriptive-causal design is suitable.
• Data would be collected once from a sample of restaurant customers through a structured questionnaire, using stratified or convenience sampling.
• The data would be analysed with descriptive statistics, correlation and regression to test the hypotheses.
• The original study used an exploratory qualitative design (critical incident technique). A follow-up study can measure the identified drivers quantitatively.