BT 783: Quantitative Research Design

• The lecture is for BT 783: Advanced research method with the focus of quantitative research design.
• The lecture is designed by V.K Acheamfour at the Kwame Nkrumah University of Science & Technology, Kumasi, Ghana.

Introduction

• Research design is a comprehensive plan consisting of how research components are executed and how the major parts are structured to achieve the research aims and objectives (Creswell and Creswell, 2018).
• There are varying opinions on research designs
• Research design comprises of three sub-components, which are the applied research method, research philosophy, and research strategy (Creswell, 2014).
• Research design comprises of six distinct layers and is known as the onion ball (Saunders et al., 2016)
• The lecture focuses on the procedures for conducting research based on quantitative research design.

Part I: Understanding Quantitative Research

• Quantitative research design utilizes statistics or numbers to quantify the world.
• Quantitative approaches use statistics to numerically describe phenomena and ascertain relationships.
• Quantitative research establishes empirical relationships, but it is less suited to explain causal mechanisms or constituents behind statistical relationships.

Quantitative vs. Qualitative Research

• Quantitative and qualitative methods are complementary and used in the social sciences

The quantitative research process

• Hypothesis: A provisional or unconfirmed statement derived from a theory, tested via verification or falsification
• Operationalization: Transforms a concept into a variable, crucial for achieving content validity
• Theory: Provides a simplified explanation of how the world works.
• Measurement
• Statistical Analysis
• Sampling

Key points about theory and hypothesis:

• Theories explain content and attributes through concepts.
• Measurement ease varies between concise and abstract concepts.
• Variables measure concepts.
• Content validity is crucial when operationalizing a concept.

Measurement of Variables

• The two categories of variables are:
  • Categorical data: data must be classified or put in a ranked order
    • Includes ordinal and nominal data
  • Quantifiable data: values are measured numerically as a quantity
    • Includes discrete, continuous, ratio, and interval data

Part II: Doing Quantitative Research (The Survey)

• Surveys are a major technique for gathering information about people or firms.
• This research method collects data systematically using standardized procedures
• Researchers commonly use random or representative samples to study a population's attitudes, perceptions, or behaviors.

Types of surveys

• Cross-sectional: gathers information from individuals once.
  • Draws inferences at a particular point in time.
  • Cannot establish causality.
• Findings from cross-sectional studies must be supported by theory, logic, or intuition.
• Longitudinal: repeats the same questions, and is used to analyze changing behaviors and attitudes in a population over time.
  • The three types of studies are trend, cohort, and panel.

Questionnaire design

• A survey researcher can ask questions about what people think, what they do, what attributes they have, and how much knowledge they have about an issue
• These are some factors to consider in questionnaires:
  • Ordering of questions should move from general to specific, impersonal to personal, and easy to difficult.
  • The number of questions should include as many questions as necessary and as few questions as possible.
  • Question wording should be: clear, simple, and precise, Avoid the following:
    • Vague
    • Biased
    • Pointless

Other considerations for designing questionnaires include:

• Social desirability: inclination to give socially desirable responses, or responses that make them look good against the background of social norms or common values.
• Open-ended questions allow respondents to answer in their own words, while closed-ended questions require them to select an answer from a predetermined set of choices
• Coding, where question responses need to be transformed into numbers for data analytical purposes

Variables

• The different ways to operationalize survey questions are:
  • String
  • Continuous
  • Ordinal
  • Nominal variables

Population and sample size

• Target population should include all subjects the research is interested in.
• Interviewing the whole population is too expensive and or not logistically possible.
• You must find the sampling frame and select only a sample of the population.
• A sampling frame consists of all units from which the sample is drawn.

The different types of sampling are:

• Representative sample mirrors the population’s features
• Random sample gives every individual an equal chance of selection
• Biased sample is neither representative nor random and causes selection, non-response, and response biases.

Sample size determination strategies

• Use a census for small populations
• Using sample size of a similar study
• Using published tables
• Using formulae to calculate a sample size

Sampling Error

• It depicts the degree to which the results is different from the results from population.
• Non-random sampling: individuals don’t possess equal selection chances.
• The common non-probabilistic sampling techniques are convenience, purposive, volunteer, and snowball sampling.

Conducting the survey

• Survey methods can be conducted through:
  • Face to face
  • Telephone survey
  • Mail-in survey
  • Online survey

Part III: Doing Quantitative Research (Data Analysis)

• Quantitative research data analysis is mostly conducted with the aid of specialized software such as:
  • SPSS
  • Stata which have functions for data analysis, data management, and graphics

Analysis

• Reliability: The extent to which results can be reproduced under the same conditions
  • Assessed by: Checking consistency of results across time, observers, and parts of the test
• Validity: The extent to which results measure what they should
  • Assessed by: Checking results against established theories and other measures of the same concept

Univariate statistics

• Frequency tables show raw frequencies and percentages.
• Central tendency measures include mean, median, mode, and range.
  • Mean: the average value
  • Median: the middle number in a distribution.
  • Mode: the most frequent value.
  • Range: provides an indication how widely spread the data is.
• Spread: refers to the distribution of data
  • Deviation: is the difference of the observation and the sample mean
  • Sample variance presents the average of the squared deviations.
  • Standard deviation measures the variability around the mean
  • Standard error how close the mean of the sample is to the population mean.
• Other:
  • The relative importance index (RII) ranks variables
  • 1-sample t-test compares mean to an established mean
  • Factor analysis performs data reduction and identification of themes

Parametric vrs Non-parametric

• Statistics are used because it is usually impossible to collect data from all members of a population, hence we use a sample to make true inference about the population.
• Parametric procedures assume a distribution shape while nonparametric procedures do not. Tools for assessing data normality:
  • Kolmogorov-Smirnov
  • Shapiro-Wilk
  • Histogram

Bivariate and Multi-variate statistics

Regression estimators:

• OLS: Dependent variable is continuous.
• LOGIT: Dependent variable is dichotomous.
• ORDERED LOGIT: Dependent variable is ordered categorical.
• MULTINOMIAL LOGIT: Dependent variable is multiple categorical.