Type: Results Chapter | Subject: Marketing | Level: Undergraduate | Word Count: ~2200 words
This model chapter was produced by an Essays UK specialist as reference material for learning purposes only. For support in this field, see our marketing dissertation support.
This chapter should present the results of your online survey investigating the relationship between brand trust, perceived value, social media engagement and brand loyalty among UK undergraduate students, reporting sample characteristics, scale reliability and hypothesis-testing outcomes only. Interpretation against the existing literature belongs in the Discussion chapter.
Note: this is a single chapter presented as reference material; the full dissertation would include the remaining chapters.
This chapter presents the results of the cross-sectional online survey conducted to examine the relationships between brand trust, perceived value, social media engagement and brand loyalty among undergraduate students at a UK university, in relation to fast-fashion apparel brands. Data were analysed using IBM SPSS Statistics version 28. The chapter is organised into six sections: the sample profile and response rate; data screening procedures; the reliability of the four measurement scales used; descriptive statistics for each construct; the results of correlation and regression analysis addressing the three hypotheses set out in Chapter Three; and a summary of the findings. In keeping with standard dissertation convention, this chapter reports the statistical results only; interpretation of these findings in relation to the existing literature is presented in Chapter Five. The results are reported in the order in which they were generated during analysis, and no results are omitted or selectively presented: all pre-registered hypothesis tests and diagnostic checks conducted for this dissertation are included below, consistent with transparent reporting practice for undergraduate quantitative dissertations (Bryman, 2016).
The survey link was distributed to 320 undergraduate students via the university’s marketing programme mailing list and associated student social media groups over a three-week period. A total of 241 responses were submitted, of which 27 were excluded prior to analysis: 18 were incomplete beyond the demographic section and 9 failed one or both attention-check items embedded in the questionnaire. This left 214 valid responses for analysis, representing a usable response rate of 66.9 per cent. Table 1 summarises the demographic profile of the achieved sample.
| Characteristic | n | % |
|---|---|---|
| Gender: Female | 138 | 64.5% |
| Gender: Male | 71 | 33.2% |
| Gender: Prefer not to say / other | 5 | 2.3% |
| Year 1 | 58 | 27.1% |
| Year 2 | 79 | 36.9% |
| Year 3 | 63 | 29.4% |
| Year 4 / placement year | 14 | 6.5% |
| Faculty: Business and Management | 96 | 44.9% |
| Faculty: Social Sciences | 51 | 23.8% |
| Faculty: Arts and Humanities | 34 | 15.9% |
| Faculty: STEM | 33 | 15.4% |
The sample was predominantly female (64.5 per cent) and drawn most heavily from Years 2 and 3 of study (66.3 per cent combined), with just under half of respondents (44.9 per cent) enrolled in a Business or Management-related programme. This profile is broadly consistent with the gender and subject distribution reported for marketing undergraduate cohorts nationally (Higher Education Statistics Agency, 2023), although the sample should not be treated as statistically representative of the wider student population given the non-probability sampling method used.
Respondents were also asked to nominate the fast-fashion brand they purchased from most frequently, in order to anchor subsequent items to a consistent reference brand for each respondent, following the nominated-brand approach used in comparable student brand-loyalty research. Seventy-two per cent of respondents reported purchasing from their nominated brand at least once a month, and a further 21 per cent reported purchasing at least once every two to three months, indicating that the achieved sample was, in the main, reasonably engaged with the brand category under investigation rather than composed of infrequent or lapsed customers.
A non-response comparison was also conducted to assess whether the 27 excluded cases differed systematically from the 214 valid cases retained for analysis. Comparing the two groups on the demographic variables available for both, gender and year of study, using chi-square tests of independence, revealed no statistically significant differences, gender: χ²(2) = 1.84, p = .40; year of study: χ²(3) = 3.21, p = .36, providing some reassurance that the exclusion of incomplete and attention-check-failing responses did not introduce a substantial demographic bias into the achieved analytic sample relative to the full set of respondents who began the survey.
Prior to substantive analysis, the dataset was screened for missing values, normality and multivariate outliers. Item-level missing data were minimal, affecting fewer than 1.5 per cent of cases across any single item, and Little’s (1988) Missing Completely at Random test was not significant, χ²(38) = 41.62, p = .32, supporting the use of expectation-maximisation imputation for the small number of missing values rather than listwise deletion. Skewness and kurtosis values for all four composite scales fell within the conventional ±1 range, indicating an acceptably normal distribution for parametric analysis (Tabachnick and Fidell, 2019); Shapiro-Wilk tests were statistically significant for all four scales, which is unsurprising given the sample size and is not treated as disqualifying given that skewness and kurtosis values themselves were within acceptable limits. Mahalanobis distance was calculated for all cases against the three predictor variables to identify multivariate outliers; no case exceeded the critical chi-square value of 16.27 at p < .001 for three degrees of freedom, and no cases were therefore removed from the dataset prior to hypothesis testing. Specific skewness values ranged from -0.31 (Brand Trust) to 0.24 (Social Media Engagement), and kurtosis values ranged from -0.42 to 0.38 across the four scales, both comfortably within the ±1 threshold applied above. Levene’s test for equality of variances, run in preparation for the supplementary group comparisons reported later in this chapter, indicated no violation of the homogeneity of variance assumption for Brand Loyalty scores across gender or year-of-study groups (all p > .10).
Because all constructs were measured via self-report within a single online survey administered at one point in time, the dataset is potentially vulnerable to common method variance, whereby correlations between variables are inflated by shared measurement method rather than the underlying constructs themselves. Following Podsakoff et al.’s (2003) recommended procedural remedies, item order was randomised across the four scales, response anchors were varied slightly between sections, and respondents were assured of anonymity to reduce socially desirable responding. Harman’s single-factor test was also conducted as a post-hoc diagnostic check: an unrotated exploratory factor analysis of all eighteen items produced four factors with eigenvalues greater than 1, and the first, largest factor accounted for 31.4 per cent of the total variance, below the 50 per cent threshold conventionally taken to indicate a substantial common method variance problem. While this test has recognised limitations as a sole diagnostic, its result, taken together with the procedural remedies applied at the design stage, provides reasonable, though not conclusive, grounds for treating common method variance as unlikely to be a major threat to the validity of the correlations reported below.
Prior to hypothesis testing, the internal consistency of each multi-item scale was assessed using Cronbach’s alpha. All four scales exceeded the conventional threshold of .70 (Nunnally and Bernstein, 1994), indicating acceptable reliability for the achieved sample. Table 2 reports the number of items, mean score, standard deviation and Cronbach’s alpha for each construct.
| Construct | Items | M | SD | Cronbach’s α |
|---|---|---|---|---|
| Brand Trust | 5 | 3.92 | 0.68 | .84 |
| Perceived Value | 4 | 3.68 | 0.71 | .79 |
| Social Media Engagement | 4 | 3.41 | 0.79 | .81 |
| Brand Loyalty | 5 | 3.77 | 0.74 | .88 |
Corrected item-total correlations for all scale items exceeded the recommended minimum of .30 (Field, 2018), and no item’s removal would have improved the alpha coefficient for its parent scale by more than .02, supporting the retention of all items in their original form for the descriptive and inferential analyses that follow. Mean scores across the four constructs ranged from 3.41 (Social Media Engagement) to 3.92 (Brand Trust) on the five-point Likert scale, indicating a generally positive but not extreme pattern of responses. Mean inter-item correlations within each scale ranged from .42 to .61, within Field’s (2018) recommended .15 to .50 band for unidimensional scales of this kind, offering further item-level support for the internal consistency indicated by the Cronbach’s alpha values reported above. Figure 1 presents these mean scores graphically.
Figure 1: Mean construct scores on a 1–5 Likert scale (n = 214)
Three hypotheses were tested using Pearson product-moment correlation followed by multiple linear regression, with Brand Loyalty entered as the dependent variable. H1 proposed a positive relationship between Brand Trust and Brand Loyalty; H2 proposed a positive relationship between Perceived Value and Brand Loyalty; H3 proposed a positive relationship between Social Media Engagement and Brand Loyalty. Table 3 presents the correlation matrix for the four study variables.
| Variable | 1 | 2 | 3 | 4 |
|---|---|---|---|---|
| 1. Brand Trust | — | |||
| 2. Perceived Value | .49** | — | ||
| 3. Social Media Engagement | .33** | .29** | — | |
| 4. Brand Loyalty | .58** | .46** | .27** | — |
**p < .01 (two-tailed)
All three predictor variables were significantly and positively correlated with Brand Loyalty at the bivariate level, providing initial support for H1, H2 and H3. The strongest correlation was observed between Brand Trust and Brand Loyalty (r = .58, p < .001), followed by Perceived Value (r = .46, p < .001) and Social Media Engagement (r = .27, p < .001). Using Cohen’s (1988) conventional benchmarks for correlation effect sizes (small = .10, medium = .30, large = .50), the correlation between Brand Trust and Brand Loyalty was large in magnitude, while the correlations involving Perceived Value and Social Media Engagement were medium and small-to-medium respectively.
Before conducting the regression analysis, the standard assumptions of multiple linear regression were checked. Scatterplots of each predictor against the standardised residuals indicated an acceptably linear relationship in each case, and the plot of standardised residuals against predicted values showed no discernible pattern, supporting the assumption of homoscedasticity. Tolerance values for the three predictors ranged from .81 to .91, corresponding to variance inflation factors of between 1.10 and 1.23, well below the conventional threshold of 5 (Hair et al., 2019), indicating that multicollinearity did not pose a threat to the stability of the regression coefficients. The Durbin-Watson statistic was 1.98, close to the ideal value of 2, indicating no evidence of problematic autocorrelation among the residuals.
A standard multiple regression was then conducted to establish the combined and independent predictive contribution of the three variables. The overall model was statistically significant, R² = .42, adjusted R² = .41, F(3, 210) = 50.72, p < .001, indicating that the three predictors jointly accounted for approximately 42 per cent of the variance in Brand Loyalty scores. At the individual predictor level, Brand Trust (β = .45, p < .001) and Perceived Value (β = .21, p < .001) each made a statistically significant independent contribution to the model, while the standardised coefficient for Social Media Engagement was not statistically significant once Brand Trust and Perceived Value were controlled for (β = .08, p = .163). Standardised residuals for all cases fell within ±3, and Cook’s distance did not exceed 1 for any case, indicating that no single respondent exerted undue influence on the regression model. Ninety-five per cent confidence intervals for the unstandardised regression coefficients were [0.38, 0.61] for Brand Trust, [0.14, 0.33] for Perceived Value and [-0.03, 0.19] for Social Media Engagement, with the interval for Social Media Engagement crossing zero, consistent with the non-significant standardised coefficient reported above.
On this basis, H1 and H2 were supported. H3 was supported at the bivariate level but not supported once the shared variance with Brand Trust and Perceived Value was accounted for in the regression model.
As a supplementary, exploratory analysis, an independent-samples t-test was conducted to compare Brand Loyalty scores between female (n = 138, M = 3.81, SD = 0.73) and male respondents (n = 71, M = 3.71, SD = 0.76); the difference was small and did not reach statistical significance, t(207) = 0.94, p = .35, 95% CI [-0.11, 0.31]. A parallel one-way ANOVA comparing Brand Loyalty scores across the four year-of-study groups was similarly non-significant, F(3, 210) = 1.42, p = .24. These supplementary comparisons are reported for completeness and were not among the pre-registered hypotheses tested above.
The achieved sample of 214 undergraduate students produced four reliable measurement scales, each exceeding the recommended Cronbach’s alpha threshold. Descriptive results indicated moderately positive mean scores across all constructs. Correlation analysis supported all three hypothesised bivariate relationships, while multiple regression indicated that Brand Trust and Perceived Value were the stronger independent predictors of Brand Loyalty, together accounting for the majority of the explained variance, with Social Media Engagement’s contribution not reaching statistical significance once these two variables were controlled for. In hypothesis terms, H1 (Brand Trust → Brand Loyalty) and H2 (Perceived Value → Brand Loyalty) were both supported at the bivariate and multivariate level; H3 (Social Media Engagement → Brand Loyalty) was supported at the bivariate level only. The supplementary gender and year-of-study comparisons reported above indicated no statistically significant demographic differences in Brand Loyalty within this sample. No cases were excluded from the analysis on data-quality grounds, and all four measurement scales demonstrated acceptable reliability and an approximately normal distribution suitable for the parametric tests reported above.
It should be reiterated that the statistical results reported in this chapter are correlational rather than experimental in design, and that the regression model, while identifying Brand Trust and Perceived Value as the stronger independent statistical predictors of Brand Loyalty within this sample, does not by itself establish the direction or mechanism of any underlying causal relationship between the constructs measured. Table 1, Table 2 and Table 3, together with Figure 1, are referred to again in Chapter Five, where the pattern of results summarised here is considered against the wider brand loyalty and consumer behaviour literature and against the study’s methodological limitations.
Need a Model Dissertation Chapter Written to Your Exact Brief?
Our 350+ UK-qualified writers deliver referenced model documents from £15 per 250 words, with free plagiarism and AI-detection reports.
You May Also Like
All work is written by human writers. 100% AI free, guaranteed.
100% money back guarantee if you find plagiarism in our work.
COMPANY DETAILS
