Introduction

to
Quantitative
Research
O R I G I N A L I T Y: P R O F. T E R R Y R YA N A N D D R . L I Z
FITZGERALD
E D I T E D BY: D R . H I N D B I TA R
LLOs
The students should be able to:
1. explain different Quantitative data collection techniques (Knowledge)
2. use different data collection techniqes to collect Quantitative data
(Cognitive)
3. differentiate between probability and non-probability samples
(Cognitive)
4. differentiate between the three types of research design (Cognitive)
5. explain type I and II errors (Knowledge)
Research and Research Methods
Which Research method you choose will
depend on:
◦Your research questions
◦Your underlying philosophy of research
◦Your preferences and skills
Quantitative Approach
oAttempts to explain phenomena by collecting and analysing numerical
data
oTells if there is a “difference” but not necessarily why
oData collected are always numerical and analysed using statistical
methods
oRandomisation to reduce subjective bias
oIf there are no numbers involved, its not Quantitative
oSome types of research lend themselves better to quant approaches than
others
Data Sources in Quantitative Research
◦ Surveys
◦ Questionnaire
◦ Observations (counts of numbers and/or coding data into
numbers)
◦ Secondary data (government data; SATs scores etc)

Large number is needed

CONT.
Be aware of the following in designing your Questionnaire :
• Must have a theoretical framework or conceptual structure
• Keep language simple
• Keep questions short
• Avoid double-barreled questions
• Avoid leading questions
• Avoid questions in the negative
• Ask only questions for which the respondent should possess knowledge needed
• Try to make sure the question means the same thing to all respondents
CONT.
•Remove ambiguity
•Avoid direct questions on sensitive topics
•Ensure the question’s frame of reference is clear
•Avoid creating opinions
•Use personal wording if you want respondent’s own feelings
•Avoid unnecessary/objectionable detail
Probability Samples
Allow generalization to population

best chance to create a sample that is truly representative of the population

1. Simple random sampling: The lottery system and using number generating software/
random number table

2. Stratified random sampling: the researcher divides a more extensive population into smaller
groups that usually don’t overlap but represent the entire population. Then draw a sample
from each group separately. Dividing population into strata, sampling from these randomly

https://www.questionpro.com/blog/probability-sampling/
 CONT.
3. Random cluster sampling: select participants randomly that are spread out geographically.
◦ you wanted to choose 100 participants from the entire population of the U.S.
◦ it is likely impossible to get a complete list of everyone
◦ the researcher randomly selects areas (i.e., cities or counties) and randomly selects from within those
boundaries

4. Systematic sampling: choose every “nth” individual to be a part of the sample

◦ select every 3th person to be in the sample

https://www.questionpro.com/blog/probability-sampling/
Non-Probability Samples
1. Convenience sampling
2. Snowball sampling
Types of Research Design
1. Randomized experiments
2. Non-randomized experiments
◦ Quasi-experiment / nonequivalent groups
3. Non-experiments
Experiments
1. The distinguishing feature of an experiment is
deliberate manipulation
2. The point is to test a causal proposition
3. Can be randomized or not
Features of Experiments
1. Experimental units (e.g., people, classrooms, businesses,
communities...)
2. Treatments (independent variables)
3. Outcomes (dependent variables)
4. Comparisons
5. Assignment (of experimental units)
Randomized Experiments
❖Employ initial random assignment
❖More difficult to do in field than in lab
❖More difficult to do with humans than with objects
❖Use one of the Probability Samples techniques to select
your subjects
 Non-randomized (Quasi-) Experiments
Have

1. Treatments

2. Outcomes

3. Experimental units

4. Comparisons

Lack

Randomized assignment

Require the researcher to rule out possible pre-treatment differences between groups
Hypothesis
❑ Start with null hypothesis
• i.e. H0 – that there will be no difference
❑ A good hypothesis must be based on a good research
question. It should be simple, specific and stated in advance
(Hulley et al., 2001).
Type I and II
Errors (Banerjee
et al., 2008)
Type I error, also known as a “false
positive”:

if an investigator rejects a null

hypothesis that is actually true in
the population

Type II error, also known as a

"false negative"

if the investigator fails to reject a

null hypothesis that is actually false
in the population.
Analyzing Quant. Data
❑ Always good to group and/or visualise the data initially → outliers/cleaning data

❑What average are you looking for?

❑Mean, median or mode?

❑Spread of data:
❑ skewness/distribution
❑ range, variance and standard deviation
CONT.
❑ Trying to find the signal from the noise
❑Generally, either a difference (between/within groups) or a correlation

❑Choosing the right test to use:

❑(depends what sort of data you have – interval/ratio vs nominal/ordinal and how it is distributed)

❑Correlation does not imply causation!

Interpreting the Results
Significance level – a fixed probability of wrongly rejecting the null hypothesis H0, if it is in fact
true. Usually set to 0.05 (5%).

p value - used in hypothesis testing to help you support or reject the null hypothesis
◦ The p value is the evidence against a null hypothesis
◦ The smaller the p-value, the stronger the evidence that you should reject the null hypothesis
◦ A small p (≤ 0.05), reject the null hypothesis
◦ This is strong evidence that the null hypothesis is invalid

Effect size – numerical way of expressing the strength or magnitude of a reported relationship,
be it causal or not
Example
Matched users were those who learning styles were matched with the lesson plan e.g.
sequential users with a sequential lesson plan. Mismatched participants used a lesson plan that
was not matched to their learning style, e.g. sequential users with a global lesson plan.

H0 – there will be no statistically significant difference in knowledge gained between users from
different experimental groups
H1 – students who learn in a matched environment will learn significantly better than those who
are in mismatched environment
H2 – students who learn in a mismatched environment will learn significantly worse than those
who learn in a matched
Statistical testing was carried out using a univariate ANOVA in SPSS, to
determine if there was any significant difference in knowledge gained.
Initial conjecture suggests that the mismatched group actually performed
better than the matched group.
However, the difference between the two groups was not significant
(F(1,80)=0.939, p=0.34, partial eta squared = 0.012) and hence hypotheses 1
and 2 can be rejected.
You have to worry about the following:
❑Is my sample size big enough?

❑Have I used the correct statistical test?

❑have I reduced the likelihood of making Type I and/or Type II errors?

❑Are my results generalisable?

❑Are my results/methods/results reproducible?

❑Am I measuring things the right way?

Be careful
❑Some things can’t be measured – or measured accurately

❑Doesn’t tell you why

❑Can be impersonal – no engagement with human behaviours or individuals

❑Data can be static – snapshots of a point in time

❑Can tell a version of the truth (or a lie?)