The order of a quantitative chapter
A clear order helps the reader follow, and examiners check each step. Report in the sequence below and keep the headings parallel to your research questions or hypotheses.
| Section | Contents |
|---|---|
| Data preparation | Response rate, exclusions, missing data, outliers, coding of variables |
| Sample description | Demographics of respondents, compared with the population where possible |
| Reliability and validity | Cronbach's alpha for each scale, factor analysis if used |
| Descriptive statistics | Means, standard deviations, ranges, correlations |
| Assumption checks | Normality, linearity, multicollinearity and homoscedasticity for the tests used |
| Hypothesis tests | Results for each hypothesis in order, with test statistics, p values and effect sizes |
| Summary of findings | A table showing each hypothesis and whether it was supported |
Describe the data first
Report the number of invitations, completed responses and the response rate, and then explain what you excluded and why (for example, 11 responses with more than 20 percent missing data). State how you handled missing values and outliers. Present descriptive statistics in a table.
| Variable | n | Mean | SD | Cronbach's alpha |
|---|---|---|---|---|
| Supervisor support | 150 | 3.82 | 0.71 | 0.88 |
| Onboarding quality | 150 | 3.41 | 0.84 | 0.81 |
| Intention to stay | 150 | 3.65 | 0.92 | 0.86 |
Say what the figures mean in plain language: respondents rated supervisor support relatively high (mean 3.82 on a five-point scale), and all scales showed good internal consistency (alpha above 0.80).
A worked comparison of two groups
Suppose you compare intention to stay between employees who received a structured onboarding program (group A) and those who did not (group B).
Independent samples t-test (hypothetical)
Group A: n = 60, mean = 3.9, SD = 0.8. Group B: n = 60, mean = 3.5, SD = 0.9.
Difference in means = 0.4. Standard error = square root of (0.8 squared / 60 + 0.9 squared / 60) = square root of (0.010667 + 0.0135) = square root of 0.024167 = 0.1555.
t = 0.4 / 0.1555 = 2.57 with about 117 degrees of freedom, giving p = .011, which is below .05.
Effect size (Cohen's d) = 0.4 / pooled SD. Pooled SD = square root of ((0.64 + 0.81) / 2) = square root of 0.725 = 0.85. So d = 0.4 / 0.85 = 0.47, a small to medium effect.
Write-up: Employees who received structured onboarding reported higher intention to stay (M = 3.90, SD = 0.80) than those who did not (M = 3.50, SD = 0.90), t(117) = 2.57, p = .011, d = 0.47.
Always report the effect size along with the p value. A p value tells you whether the difference is likely to be due to chance, but not whether it matters, and with a large sample tiny differences can be significant.
Reporting a regression
Multiple regression estimates how much each predictor explains the outcome while holding the others constant. Report the overall fit, each coefficient and the checks you performed.
| Predictor | B | SE | Beta | t | p |
|---|---|---|---|---|---|
| Supervisor support | 0.48 | 0.08 | 0.42 | 6.00 | < .001 |
| Onboarding quality | 0.20 | 0.08 | 0.18 | 2.50 | .014 |
| Tenure (years) | 0.03 | 0.05 | 0.04 | 0.60 | .55 |
Model summary (hypothetical): R squared = .34, adjusted R squared = .33, F(3, 146) = 25.1, p < .001. In words: the model explains 34 percent of the variance in intention to stay. Supervisor support was the strongest predictor (beta = .42, p < .001); onboarding quality made a smaller but significant contribution (beta = .18, p = .014); tenure was not significant.
Before reporting, check the assumptions: the relationship is roughly linear, residuals are approximately normal, variances are similar across levels of the predictors (homoscedasticity), predictors are not too highly correlated (variance inflation factors below about 5 to 10) and observations are independent. If assumptions fail, use a transformation, a robust method or a different test, and say so.
Summarize hypotheses and separate results from discussion
| Hypothesis | Test | Result | Decision |
|---|---|---|---|
| H1: Supervisor support is positively related to intention to stay | Regression | Beta = .42, p < .001 | Supported |
| H2: Onboarding quality is positively related to intention to stay | Regression | Beta = .18, p = .014 | Supported |
| H3: Tenure is positively related to intention to stay | Regression | Beta = .04, p = .55 | Not supported |
Keep this chapter about what the data show. Save explanations, links to theory, comparisons with earlier studies and practical implications for the discussion chapter. A common pattern is that the results chapter says supervisor support was the strongest predictor, and the discussion says why that fits or challenges the literature. For qualitative results, see our guide to thematic analysis.
Working on this assignment now? Get a price for help with your paper.
Get an instant quoteReading a correlation
A correlation coefficient (r) summarizes the strength and direction of a linear relationship between two variables, from -1 to +1.
Correlation and its test (hypothetical)
The correlation between supervisor support and intention to stay is r = 0.52 in a sample of 150.
Shared variance = r squared = 0.52 squared = 0.27: about 27 percent of the variation in one variable is associated with the other. Test: t = r x square root of (n - 2) / square root of (1 - r squared) = 0.52 x 12.17 / 0.854 = 7.41 with 148 degrees of freedom, so p is below .001.
Write-up: Supervisor support was positively correlated with intention to stay, r(148) = .52, p < .001.
| Size of r (absolute) | Common description |
|---|---|
| 0.10 | Small |
| 0.30 | Medium |
| 0.50 or more | Large |
These labels come from conventions that vary by field. Say what the size means in your context and remember that correlation does not show cause.
Presenting tables clearly
- Number and title every table Table 4.2. Descriptive statistics and correlations for study variables.
- Keep one idea per table Do not combine unrelated results.
- Use consistent decimals Two for most statistics, three for p values.
- Add notes under the table Sample size, scale range, and the meaning of any symbols.
- Refer to each table in the text And say what the reader should notice.
- Move long output to an appendix Keep the chapter readable.
A reader should be able to understand the table without the text, and the text should add interpretation, not repeat every number.
Assumptions and what to do when they fail
| Assumption | How to check | If it fails |
|---|---|---|
| Normality of residuals | Histogram or Q-Q plot of residuals | Transform the variable or use a robust or non-parametric method |
| Linearity | Scatterplot of predictor against outcome | Add a transformation or a curved term |
| Equal variances | Residuals versus fitted values plot; Levene's test for groups | Use Welch's t-test or robust standard errors |
| No severe multicollinearity | Variance inflation factors; correlations among predictors | Combine or drop overlapping predictors |
| Independence | Study design; clustering of respondents | Use methods for clustered data |
Report the check and the decision in one or two sentences. Examiners look for evidence that you knew the assumptions mattered, not for perfect data.
Common mistakes
- Reporting p values only Add effect sizes and confidence intervals where possible.
- Skipping assumption checks Examiners look for them.
- Mixing results and discussion Interpret later.
- Too many tables Use tables for key results; move the rest to the appendix.
- Confusing correlation and causation Unless the design supports it, use associated with, not causes.
- Inconsistent decimals Choose two decimals for most statistics and p to three, and keep to it.
Check your style guide (often APA) for the exact format. If you want help with statistical analysis, you can order MBA dissertation help or business analytics assignment help.