Research Statistics
A Practical Guide for Student Researchers
Prepared as an educational reference and independent study guide.
Page 1
Table of Contents
1. Research Variables and Measurement
2. Questionnaires and Likert Scales
3. Descriptive Statistical Treatment
4. Correlation and Association
5. Choosing Statistical Tests
6. Writing Results and Conclusions
Page 2
Chapter 1: Research Variables and Measurement
Variables
A research variable is a characteristic that can take different values. Researchers define variables
operationally so that they can be observed or measured consistently.
Independent and Dependent Variables
In many explanatory studies, the independent variable represents a predictor or condition, while the
dependent variable represents an outcome. In correlational research, it is usually more appropriate to
discuss variables and their associations rather than imply experimental causation.
Levels of Measurement
• Nominal: categories without an inherent order.
• Ordinal: ordered categories.
• Interval: equal intervals without a meaningful absolute zero.
• Ratio: equal intervals with a meaningful zero.
Page 3
Chapter 2: Questionnaires and Likert Scales
Questionnaire Design
Questionnaire items should be clear, focused, and aligned with the research questions. Avoid ambiguous
wording, double-barreled questions, unnecessary jargon, and items that respondents cannot reasonably
answer.
Likert-Type Items
Likert-type items commonly ask respondents to indicate their degree of agreement or frequency using
ordered response categories. Researchers should define the response anchors clearly.
Validity and Reliability
Validity concerns whether an instrument measures what it is intended to measure. Reliability concerns
consistency of measurement. Content validation, pilot testing, and internal consistency analysis can
strengthen instrument development.
Page 4
Chapter 3: Descriptive Statistical Treatment
Frequency and Percentage
Frequencies count responses, while percentages express those counts relative to the relevant total. They
are particularly useful for categorical variables and respondent profiles.
Mean and Standard Deviation
For appropriately treated quantitative or composite scores, the mean summarizes central tendency and the
standard deviation summarizes dispersion.
Weighted Mean
Weighted means may be used when response categories are assigned numerical weights and the
research design justifies summarizing responses in this manner. Interpretation should follow the study's
stated scale and assumptions.
Page 5
Chapter 4: Correlation and Association
Pearson's r
Pearson's product-moment correlation coefficient measures the strength and direction of a linear
relationship between two quantitative variables. It is commonly used when its assumptions are reasonably
satisfied.
Spearman's Rank Correlation
Spearman's rho is a rank-based measure of monotonic association and can be useful for ordinal data or
when Pearson correlation assumptions are not appropriate.
Interpreting Correlation
The coefficient ranges from -1 to +1. The sign describes direction; the magnitude describes strength of
linear association. Statistical significance and practical importance should be considered separately.
Important Caution
A statistically significant correlation does not prove that one variable causes another. Correlational
research describes relationships, not experimental cause-and-effect.
Page 6
Chapter 5: Choosing Statistical Tests
Descriptive Questions
Questions asking about the distribution or level of a variable may use frequency, percentage, mean,
median, or standard deviation depending on the variable.
Comparing Groups
Two-group comparisons may involve an independent-samples t test or a nonparametric alternative when
assumptions or measurement conditions require it. Comparisons involving more than two groups may use
ANOVA or suitable alternatives.
Association Questions
Pearson's r and Spearman's rho are possible choices for association, depending on variable characteristics
and assumptions. Categorical associations may be examined using a chi-square test.
Beyond the Test Name
Statistical treatment should follow the research question, research design, measurement level, sampling
approach, assumptions, and data characteristics. There is no single statistical test that is automatically
correct for every Likert-scale study.
Page 7
Chapter 6: Writing Results and Conclusions
Results
Present findings objectively. Include the statistic, relevant descriptive information, and a clear
interpretation. Tables should be labeled and referenced in the text.
Discussion
Explain what the findings mean in relation to the research problem and existing literature. Avoid claims that
go beyond the study design.
Conclusion and Recommendations
Conclusions should directly answer the research questions and be supported by the findings.
Recommendations should be practical, specific, and consistent with the evidence.
Research Ethics
• Obtain appropriate informed consent or assent as required.
• Protect respondent confidentiality.
• Explain the purpose and voluntary nature of participation.
• Store data securely and limit access appropriately.
• Report results honestly, including limitations.
Page 8
Study Notes & Key Takeaways
Use the following space to record definitions, formulas, examples, questions, and personal insights from
the material.
End of guide
Page 9