Feel at Home Intervention

1. Executive Summary

This quantitative analysis examines data from 510 first-year Engineering students at Stellenbosch University who participated in the "I Belong Here" / Feel at Home intervention during Welcoming Week 2026. In this intervention, students read and watched stories from diverse senior students about their university transition experiences, completed a reflective writing activity connecting the stories to their own journey, and then responded to a survey measuring their reactions across six dimensions.

It is essential to understand that the survey dimensions do not measure general wellbeing or life satisfaction. They measure students' reactions to the specific intervention — how they perceived the stories, what emotions the stories evoked, and their resulting views on belonging, diversity, and support-seeking. This distinction is critical for interpretation.

2. The Intervention & What the Scores Measure

The "I Belong Here at Stellenbosch University" intervention (subtitled "Making Differences Count") uses a social-belonging intervention design. Students read or watch stories from five senior students — Erica, Nathan, Steven, Anne, and Christopher — who come from diverse backgrounds and share their ups and downs navigating the transition to university. After engaging with the stories, students complete a reflective writing activity connecting the stories to their own experience, and then respond to the survey that produces the six dimension scores.

What Each Dimension Measures

3. Methodology

All analyses were performed in Python using pandas, numpy, and scipy.stats. The dataset contained N = 510 complete records with zero missing values across all six composite score dimensions. All scores are on a 1–7 scale. Two items are reverse-scored within their composites: General_12 ("Was the information negative?") and Belonging_4 ("I expect the social experience to be difficult"), using the formula 8 − raw score.

Normality was assessed via Shapiro-Wilk tests (random sample of n = 500, seed = 42). All six dimensions violated normality (p < .05), so Spearman rank-order correlations were used for all correlation analyses. Threshold analysis used pragmatic cut-points of < 4.0 (below scale midpoint = concern) and < 3.0 (lower range = heightened concern). Item-level analyses and cross-tabulations were conducted to illuminate what drives dimension-level scores.

4. Descriptive Statistics

Mean Scores — Outcome Dimensions vs. Reaction Dimensions

Outcome dimensions (what the intervention aims to change):

Reaction dimensions (how students experienced the intervention):

5. Item-Level Analysis

Dimension-level means can obscure important patterns. The item-level analysis reveals which specific questions drive composite scores up or down, providing a more nuanced picture of how students experienced the intervention.

General Opinion — Why Is It the Lowest Dimension?

General Opinion (M = 4.75) is pulled down by three specific "change" items. Students largely found the stories positive, useful, and relatable — but many report that the stories did not change their existing understanding or challenge their assumptions about SU. This pattern is consistent with students who arrived already well-informed about university life.

Positive Emotions — What Resonated Most?

The stories evoked strongest feelings of positivity, optimism, and motivation — the forward-looking emotions that support a successful transition. "Relieved" and "stimulated" scored lower, which makes sense: the stories are informative rather than adrenaline-inducing, and relief requires prior anxiety to relieve.

Negative Emotions — Most Students Unaffected, But a Subgroup Wasn't

Belonging — One Item Stands Out

6. Threshold Analysis

Using < 4.0 as the concern threshold (below scale midpoint) and < 3.0 as the heightened concern threshold. Note: for General Opinion and Positive Emotions, scores below threshold likely reflect reduced resonance with the intervention content rather than personal distress. For Belonging and Negative Emotions (Reversed), scores below threshold carry more clinical significance.

Students Below Threshold by Dimension

Multi-Dimension Patterns

7. Student Profiles & Cross-Tabulations

Cross-tabulations reveal whether students who scored low on one dimension also struggled on others — or whether their profiles suggest different underlying patterns.

Profile A: Students with Low General Opinion (< 4.0, n = 110)

Profile B: Students with High Negative Emotions (Reversed < 4.0, n = 64)

Profile C: Students with Low Belonging (< 4.0, n = 29)

Most Common Co-Occurrences Below Threshold

8. Correlation Analysis

Spearman rank-order correlations (all dimensions non-normal, Shapiro-Wilk p < .05).

Spearman Correlation Matrix (ρ)

Colour key: strong (|ρ| ≥ .5)  moderate (.3–.5)  weak (.1–.3)  near-zero / negative

Key Correlation Insights

9. Distribution Analysis

Quartile Breakdown

Ceiling Effects

Seeking Support and Appreciating Differences show significant ceiling effects, with 51.6% and 36.7% of students respectively scoring ≥ 6.5 (out of 7.0). These dimensions may not differentiate well among high-functioning students. In contrast, General Opinion (3.9% at ceiling) and Positive Emotions (4.1% at ceiling) distribute more evenly across the scale and provide better discrimination.

10. Key Findings & Implications

The Intervention Achieves Its Core Outcomes

The Feel at Home intervention is designed to foster belonging, normalise diverse pathways to success, and encourage help-seeking. On all three outcome dimensions, scores are high to very high: Seeking Support (M = 6.35), Appreciating Differences (M = 6.12), and Sense of Belonging (M = 5.49). The intervention's core messages — that students from all backgrounds can succeed, that asking for help is normal and expected, and that there is no single way to be a successful student — are clearly landing with the majority of the cohort.

The Moderate General Opinion & Positive Emotions Scores Are Expected

The two "reaction" dimensions (General Opinion M = 4.75 and Positive Emotions M = 4.81) are not warning signs — they reflect the known demographic pattern of differential resonance. The stories feature students overcoming financial barriers, being first-generation university attendees, and navigating imposter syndrome from under-resourced backgrounds. Students whose own backgrounds mirror these stories rate the intervention higher; students from more privileged backgrounds, whose assumptions and understanding aren't challenged, naturally rate lower on the "change" items that drive these composites down.

The item-level evidence supports this: students overwhelmingly agreed the stories were positive (M = 5.75), that they paid attention (M = 5.44), and that they could relate to the storytellers (M = 5.24). What they disagreed with was whether the stories changed their view (M = 3.39–3.66) — a pattern fully consistent with already being well-informed about university life.

Where the Genuine Concerns Lie

The Independence of Negative Emotions Matters

The near-zero correlation between Negative Emotions and General Opinion (ρ = −.021) is the most important statistical finding in this report. It means a student can find the stories useful, feel they belong, value diversity, intend to seek support — and still feel overwhelmed, anxious, or uncertain. The intervention's positive effects do not automatically buffer against negative emotional activation. Some students may be recognising genuine vulnerabilities in the stories that mirror their own situation, and this recognition triggers both positive engagement and negative affect simultaneously.

This has direct implications: screening students only on positive metrics (belonging, satisfaction, engagement) will miss the emotionally distressed subgroup. The Connected programme tools focused on emotional regulation, coping with uncertainty, and managing anxiety are essential complements to the belonging-focused intervention.

Implications for the Connected Programme

11. Limitations & Next Steps

Limitations

This analysis is based on a single unstratified cohort measured at one time-point during Welcoming Week 2026. Without biographical data, we cannot confirm the hypothesised advantaged/disadvantaged split driving differential intervention reactions. The ceiling effects on Seeking Support and Appreciating Differences limit discrimination among high-scoring students and may partially reflect social desirability or the positive framing of orientation activities. The thresholds used (< 4.0 and < 3.0) are pragmatic cut-points, not clinically validated boundaries. Composite scores were used as provided — item-level reliability analysis (Cronbach's alpha) was not conducted.

Reproducibility

All analyses were computed using Python 3.x with pandas, numpy, and scipy.stats. The analytic pipeline is deterministic and fully reproducible given the source CSV. Shapiro-Wilk tests used n = 500 with random_state = 42.

Next Steps

The immediate priority is linking biographical information to enable stratified analysis across all six dimensions. This will allow identification of specific demographic groups where the intervention resonated most strongly, where negative emotions clustered, and where anticipated social difficulty is highest. A follow-up measurement later in the first semester would provide longitudinal data on whether the intervention's effects on belonging and support-seeking persist, and whether anticipated social difficulty translates into actual loneliness.

Citation