About this example
This page is the results chapter (Chapter 4) of an example doctoral or DBA thesis in business and management. The study surveyed employees of Indian manufacturing and service firms about total quality management (TQM) practices and their firm's quality, operational and business performance, and asked whether the two sectors differ and whether TQM practices are associated with performance. The chapter shows how to report reliability, group comparisons and correlations, and how to write about results that are not statistically significant.
It is an illustrative teaching example, not a real student's work. All numbers are illustrative data: they are round, internally consistent values created for teaching and do not describe real firms. Do not submit any part of it as your own work; your university's academic integrity rules apply.
The example: sample and measures
4.1 Introduction
This chapter reports the analysis of 400 completed questionnaires. Section 4.2 describes the respondents and their firms, Section 4.3 the reliability of the scales, Section 4.4 the comparisons between sectors (H1 to H4) and Section 4.5 the associations between TQM practices and perceived performance (H5). Analyses were run in SPSS. All items used a five-point scale from 1 (strongly disagree) to 5 (strongly agree), and scale scores are item means; treating such composite scores as interval data is common and generally robust (Norman, 2010).
4.2 Respondents and firms
| Characteristic | Category | n | % |
|---|---|---|---|
| Sector | Manufacturing | 184 | 46 |
| Service | 216 | 54 | |
| Job level | Senior management | 60 | 15 |
| Middle management | 140 | 35 | |
| Supervisory | 112 | 28 | |
| Non-supervisory | 88 | 22 | |
| Experience in the firm | Less than 4 years | 120 | 30 |
| 4 to 5 years | 100 | 25 | |
| 6 to 10 years | 112 | 28 | |
| More than 10 years | 68 | 17 |
Respondents were spread across job levels, and most had worked in their firm for at least four years, which suggests they were familiar with its quality practices. Early and late respondents did not differ on the main scales, which gives some reassurance about non-response bias (Armstrong & Overton, 1977).
4.3 Reliability of the scales
| Scale | Items | Cronbach's alpha |
|---|---|---|
| Leadership | 5 | 0.86 |
| Strategic planning | 4 | 0.81 |
| Customer focus | 5 | 0.88 |
| Information and analysis | 4 | 0.79 |
| People management | 5 | 0.87 |
| Process management | 5 | 0.84 |
| Perceived quality performance | 4 | 0.85 |
| Perceived operational performance | 4 | 0.83 |
| Perceived business performance | 4 | 0.86 |
All scales showed acceptable internal consistency (alpha 0.79 to 0.88). The six TQM practice scales were averaged into a composite TQM score for the sector comparison, as specified in the methodology.
The example: comparing the sectors (H1 to H4)
4.4 Differences between manufacturing and service firms
Independent-samples t tests compared the two sectors on the composite TQM score and the three perceived performance scales (Table 4.3). Mean scores were close to 4 in both sectors, indicating general agreement that the practices were in place and that performance was good, with standard deviations of about 0.6 to 0.7.
Before the tests were run, the distributions of the scale scores were inspected. They were moderately skewed towards agreement, as is common with Likert-type composites, but with more than 180 respondents in each group the t test is robust to this degree of non-normality. The standard deviations of the two groups were similar, so the pooled-variance form of the test is reported; Welch's version, which does not assume equal variances, gave the same conclusions. Eleven questionnaires with more than a fifth of items missing had been excluded before analysis, as set out in the methodology, leaving the 400 analysed here.
| Scale | Manufacturing M (SD) | Service M (SD) | Difference [95% CI] | t (398) | p | d |
|---|---|---|---|---|---|---|
| TQM practices (composite) | 3.98 (0.62) | 4.03 (0.58) | −0.05 [−0.17, 0.07] | −0.83 | .41 | −0.08 |
| Perceived quality performance | 3.91 (0.71) | 3.99 (0.66) | −0.08 [−0.21, 0.05] | −1.17 | .24 | −0.12 |
| Perceived operational performance | 3.87 (0.69) | 3.95 (0.64) | −0.08 [−0.21, 0.05] | −1.20 | .23 | −0.12 |
| Perceived business performance | 3.82 (0.74) | 3.88 (0.70) | −0.06 [−0.20, 0.08] | −0.83 | .41 | −0.08 |
No difference was statistically significant, so H1 to H4 were not supported: the data did not reject the null hypotheses of equal means. The confidence intervals show that any true differences are likely to be small, at most about a fifth of a scale point, and the standardised differences are close to zero (Lakens, 2013). To test whether the sectors were practically equivalent, a two one-sided tests procedure was run with equivalence bounds of ±0.25 scale points, set in advance as the smallest difference managers would regard as meaningful. All four 90% confidence intervals fell within these bounds (for example, −0.15 to 0.05 for TQM practices), so the sectors can be treated as equivalent within that margin (Lakens, 2017).
The example: TQM practices and performance (H5)
4.5 Associations between TQM practices and perceived performance
| Variable | 1 | 2 | 3 | 4 |
|---|---|---|---|---|
| 1. TQM practices (composite) | 1 | |||
| 2. Perceived quality performance | 0.62 | 1 | ||
| 3. Perceived operational performance | 0.55 | 0.58 | 1 | |
| 4. Perceived business performance | 0.48 | 0.51 | 0.64 | 1 |
TQM practices were positively and moderately to strongly correlated with all three performance measures (r = 0.48 to 0.62), supporting H5. The association was strongest for quality performance and weakest for business performance, which is consistent with quality practices affecting quality outcomes most directly. Because practices and performance were reported by the same person in the same questionnaire, part of these correlations may reflect common method variance (Podsakoff et al., 2003), and correlation alone does not show that TQM practices improve performance.
4.6 Summary of findings
| Hypothesis | Result | Interpretation |
|---|---|---|
| H1 to H4: sectors differ in TQM practices and perceived performance | Not supported | Differences were small; sectors equivalent within ±0.25 scale points |
| H5: TQM practices are associated with perceived performance | Supported | Moderate to strong positive correlations; common method variance possible |
How to adapt this to your own study
- Check that every construct in your title is measured in your data, and rename the construct if it is not.
- Report scale anchors once and give standard deviations with every mean.
- Make category bands mutually exclusive and exhaustive.
- Investigate alphas above about 0.95 and correlations above about 0.9 before reporting them.
- For each test, report the estimate, its confidence interval, an effect size and the exact p value.
- Write "did not reject" for non-significant results and, if you need to claim no difference, plan an equivalence test with a justified bound.
- Name the main threat to interpretation, such as common method variance, next to the result it affects.
Common mistakes in this chapter
- "Accepting" the null hypothesis after a non-significant test.
- Reporting means without standard deviations or scale anchors.
- Presenting very high alphas or correlations as strengths.
- Overlapping or incomplete category bands.
- Promising objective outcomes (such as financial performance) but measuring perceptions.
- Reading correlations from a single-source survey as causal effects.
- Pasting tables as images, which readers cannot search or check.
| What examiners look for | Where this example does it | Check in your own draft |
|---|---|---|
| Measures that match the research question | Perceived performance named throughout; 4.6 note | Does each construct in your title appear in your data? |
| Appropriate reliability reporting | Table 4.2 and note on high alphas | Have you checked for redundant items? |
| Complete statistical reporting | Table 4.3: M, SD, difference, CI, t, p, d | Is every test reported in full? |
| Correct interpretation of null results | 4.4: "not rejected" and equivalence test | Have you avoided "accepting" a null? |
| Awareness of bias | 4.5: common method variance | Have you named the main threat to each result? |
| Clear summary linked to hypotheses | Table 4.5 | Can the reader see each hypothesis and its outcome? |
Key takeaways
- Non-significant does not mean no difference; report the interval and avoid "accepting" a null (Altman & Bland, 1995).
- Use an equivalence test with a pre-stated bound to claim practical sameness (Lakens, 2017).
- Very high alphas and correlations signal measurement problems (Taber, 2018).
- Single-source surveys invite common method variance (Podsakoff et al., 2003).
- Make your title match what you measured.
