Seminarpaper 2
Magdalena Fink
2026-06-24
# ==============================================================================
# 1. Load required libraries
# ==============================================================================
library(readxl)
library(dplyr)## Warning: Paket 'dplyr' wurde unter R Version 4.5.3 erstellt##
## Attache Paket: 'dplyr'## Die folgenden Objekte sind maskiert von 'package:stats':
##
## filter, lag## Die folgenden Objekte sind maskiert von 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(stringr)
# ==============================================================================
# 2. Load dataset
# ==============================================================================
df_raw <- read_excel("ihsm25_survey.isle.xlsx", sheet = "data")
# ==============================================================================
# 3. Function: Parse WHO-5 responses
# ==============================================================================
parse_who <- function(x) {
x_clean <- tolower(trimws(as.character(x)))
case_when(
str_detect(x_clean, "all|6|5") ~ 5,
str_detect(x_clean, "most|4") ~ 4,
str_detect(x_clean, "more than half|3") ~ 3,
str_detect(x_clean, "less than half|2") ~ 2,
str_detect(x_clean, "some|1") ~ 1,
str_detect(x_clean, "no time|0") ~ 0,
TRUE ~ as.numeric(gsub("[^0-9]", "", x_clean))
)
}
# ==============================================================================
# 4. Data preparation / recoding
# ==============================================================================
df_clean <- df_raw %>%
mutate(
WHO1 = parse_who(`G04Q05[Q1]`),
WHO2 = parse_who(`G04Q05[Q2]`),
WHO3 = parse_who(`G04Q05[Q3]`),
WHO4 = parse_who(`G04Q05[Q4]`),
WHO5 = parse_who(`G04Q05[Q5]`),
financial_strain = case_when(
G06Q14 == "< 400€" ~ 1,
G06Q14 == "400€-500€" ~ 2,
G06Q14 == "500€ - 700€" ~ 3,
G06Q14 == "700€ - 1000€" ~ 4,
G06Q14 == "> 1000€" ~ 5,
TRUE ~ as.numeric(gsub("[^0-9]", "", as.character(G06Q14)))
),
network_sat = case_when(
G05Q24 == "Very unsatisfied" ~ 1,
G05Q24 == "Unsatisfied" ~ 2,
G05Q24 == "Neutral" ~ 3,
G05Q24 == "Satisfied" ~ 4,
G05Q24 == "Very satisfied" ~ 5,
TRUE ~ NA_real_
),
family_contact = case_when(
G06Q25 == "Never" ~ 1,
G06Q25 == "1 time" ~ 2,
G06Q25 == "2-3 times" ~ 3,
G06Q25 == "Nearly every day" ~ 4,
G06Q25 == "Every day" ~ 5,
TRUE ~ NA_real_
)
)
# ==============================================================================
# 5. Dependent variable: Mental distress
# ==============================================================================
df_clean$mental_distress <- rowMeans(
cbind(
5 - df_clean$WHO1,
5 - df_clean$WHO2,
5 - df_clean$WHO3,
5 - df_clean$WHO4,
5 - df_clean$WHO5
),
na.rm = TRUE
)
# ==============================================================================
# 6. Social support index
# ==============================================================================
df_clean$net_z <- as.numeric(scale(df_clean$network_sat))
df_clean$fam_z <- as.numeric(scale(df_clean$family_contact))
df_clean$social_support <- rowMeans(
cbind(df_clean$net_z, df_clean$fam_z),
na.rm = TRUE
)
# ==============================================================================
# 7. Centering + interaction term
# ==============================================================================
df_clean$financial_c <- as.numeric(scale(df_clean$financial_strain, scale = FALSE))
df_clean$support_c <- as.numeric(scale(df_clean$social_support, scale = FALSE))
df_clean$interaction <- df_clean$financial_c * df_clean$support_c
# ==============================================================================
# 8. Sample filtering
# ==============================================================================
df_final <- df_clean %>%
filter(!is.na(mental_distress),
!is.na(financial_c),
!is.na(support_c))
cat("Final sample size:", nrow(df_final), "out of 286\n")## Final sample size: 264 out of 286# ==============================================================================
# 9. Sample description
# ==============================================================================
# Sample size
nrow(df_final)## [1] 264# Gender distribution
table(df_final$G07Q27)##
## female male other
## 158 103 2prop.table(table(df_final$G07Q27)) * 100##
## female male other
## 60.0760456 39.1634981 0.7604563# Study program distribution
table(df_final$G07Q28)##
## Business, Economics & Law (e.g., Management, Finance, International Relations, Law)
## 67
## Health & Life Sciences (e.g., Medicine, Nursing, Biology, Psychology, Public Health)
## 70
## other
## 26
## Social Sciences & Humanities (e.g., Sociology, Education, Languages, Arts, Philosophy)
## 48
## STEM & Technology (e.g., Engineering, Computer Science, Physics, Mathematics)
## 50prop.table(table(df_final$G07Q28)) * 100##
## Business, Economics & Law (e.g., Management, Finance, International Relations, Law)
## 25.670498
## Health & Life Sciences (e.g., Medicine, Nursing, Biology, Psychology, Public Health)
## 26.819923
## other
## 9.961686
## Social Sciences & Humanities (e.g., Sociology, Education, Languages, Arts, Philosophy)
## 18.390805
## STEM & Technology (e.g., Engineering, Computer Science, Physics, Mathematics)
## 19.157088# ==============================================================================
# 10. Hierarchical regression models
# ==============================================================================
m1 <- lm(mental_distress ~ financial_c, data = df_final)
m2 <- lm(mental_distress ~ financial_c + support_c, data = df_final)
m3 <- lm(mental_distress ~ financial_c + support_c + interaction, data = df_final)
# ==============================================================================
# 11. Results
# ==============================================================================
anova(m1, m2, m3)## Analysis of Variance Table
##
## Model 1: mental_distress ~ financial_c
## Model 2: mental_distress ~ financial_c + support_c
## Model 3: mental_distress ~ financial_c + support_c + interaction
## Res.Df RSS Df Sum of Sq F Pr(>F)
## 1 262 198.59
## 2 261 193.08 1 5.5106 7.519 0.006529 **
## 3 260 190.55 1 2.5306 3.453 0.064269 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1summary(m3)##
## Call:
## lm(formula = mental_distress ~ financial_c + support_c + interaction,
## data = df_final)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.95684 -0.65221 -0.04892 0.56390 2.36198
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 2.19692 0.05275 41.646 < 2e-16 ***
## financial_c -0.04633 0.04829 -0.959 0.33821
## support_c -0.21458 0.08011 -2.679 0.00786 **
## interaction -0.13334 0.07176 -1.858 0.06427 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.8561 on 260 degrees of freedom
## Multiple R-squared: 0.0481, Adjusted R-squared: 0.03711
## F-statistic: 4.379 on 3 and 260 DF, p-value: 0.004995# ==============================================================================
# 12. Interaction plot
# ==============================================================================
if (requireNamespace("interactions", quietly = TRUE)) {
library(interactions)
interact_plot(
m3,
pred = financial_c,
modx = support_c,
interval = TRUE,
plot.points = FALSE,
main.title = "Buffering Effect of Social Support",
x.label = "Financial Strain (Centered)",
y.label = "Mental Distress"
)
}## Warning: Paket 'interactions' wurde unter R Version 4.5.3 erstellt## Warning: financial_c and support_c are not included in an interaction with one
## another in the model.
Introduction
Student mental well-being is an important topic in social science and public health research. University students often experience academic pressure, uncertainty about the future, and financial challenges. Housing costs represent a substantial part of students’ living expenses and may contribute to economic pressure and reduced well-being. According to the social determinants of health perspective, mental well-being is influenced not only by individual behavior but also by social and economic living conditions (Allen et al., 2014). Previous research has shown that financial difficulties are associated with poorer mental health among students. In particular, subjective financial stress, such as worrying about money or struggling to cover expenses, is consistently linked to lower well-being (McCloud & Bann, 2019). Although housing costs do not directly measure financial strain, they represent an important aspect of students’ financial circumstances and may influence psychological well-being. Social support may help reduce the negative effects of economic stress. Students who feel supported by family, friends, or their wider social network generally report better mental health outcomes (Hefner & Eisenberg, 2009). According to the stress-buffering hypothesis, social support can weaken the impact of stressful experiences by providing emotional and practical resources (Cohen & Wills, 1985). Based on these considerations, this study examines whether housing costs are associated with students’ mental distress and whether social support moderates this relationship.
Literature Review
The social determinants of health framework suggests that mental well-being is shaped by social and economic living conditions, including income, housing, education, employment, and social relationships (Allen et al., 2014; Kirkbride et al., 2024). For university students, financial insecurity may affect several aspects of everyday life, including housing, social participation, and academic performance. Research consistently shows that financial difficulties are associated with poorer mental health among students. McCloud and Bann (2019) found that perceived financial stress and money-related worries are more strongly related to mental health than objective debt levels. Although previous studies have primarily examined subjective financial stress, housing costs constitute an important component of students’ financial circumstances and may therefore serve as an indicator of economic pressure. Social support is another important determinant of students’ mental well-being. Students with stronger social networks generally report lower levels of psychological distress (Hefner & Eisenberg, 2009). According to the stress-buffering hypothesis, social support can reduce the negative effects of stressful experiences by providing emotional support, practical assistance, and a sense of belonging (Cohen & Wills, 1985). Similarly, Åslund et al. (2014) found evidence that social support can buffer the adverse effects of financial stress on psychological well-being. Based on this literature, the following hypotheses were formulated: • H1: Higher housing costs are associated with higher mental distress. • H2: Higher social support is associated with lower mental distress. • H3: Social support moderates the relationship between housing costs and mental distress, such that the association is weaker among students with higher levels of social support.
Methodology
Prior to the main analyses, the dataset was cleaned using listwise deletion of cases with missing values on variables required for the regression models. The final analytical sample consisted of N = 264 participants. The sample was composed of 60.1% female (n = 158), 39.2% male (n = 103), and 0.8% identifying as other (n = 2). Participants were distributed across five study fields: Business, Economics & Law (25.7%), Health & Life Sciences (26.8%), Social Sciences & Humanities (18.4%), STEM & Technology (19.2%), and other fields (10.0%).
Mental distress was measured using the WHO-5 Well-Being Index. The items were recoded into numerical values ranging from 0 to 5 and subsequently reverse-coded (5 − score), such that higher values indicate higher levels of psychological distress (i.e., lower well-being). The five items were then averaged to create a composite mental distress score. Housing costs were used as a proxy for students’ financial situation and were measured using ordinal categories ranging from less than €400 (coded as 1) to more than €1000 (coded as 5). Although housing costs do not directly measure financial strain, they represent an important component of students’ economic circumstances and were treated as a quasi-continuous predictor in the regression analysis. Social support was measured using two indicators: satisfaction with one’s social network and frequency of contact with family members. Both variables were standardized (z-scores) and combined into a composite social support index. To test the stress-buffering hypothesis, both housing costs and social support were mean-centered before computing their interaction term. To examine the hypotheses, three hierarchical ordinary least squares (OLS) regression models were estimated. Model 1 included housing costs as the sole predictor of mental distress. Model 2 added social support. Model 3 included the interaction term between housing costs and social support.
Results
Hierarchical Model Comparison
Three hierarchical OLS regression models were estimated to examine the relationship between housing costs, social support, and mental distress. Model 1 included housing costs as the only predictor of mental distress. This baseline model tested whether students’ housing expenditures were associated with differences in psychological distress. Model 2 added social support to the regression model. The inclusion of social support significantly improved model fit compared to Model 1, F(1, 261) = 7.519, p = .007, indicating that social support explains additional variance in mental distress beyond housing costs alone. Model 3 further included the interaction term between housing costs and social support in order to test the stress-buffering hypothesis. The interaction led to a small increase in explained variance; however, this improvement was not statistically significant at the conventional 5% level, F(1, 260) = 3.453, p = .064. This suggests that the moderating effect of social support is weak and not robustly supported by the data. The overall regression model (Model 3) was statistically significant, F(3, 260) = 4.379, p = .005, explaining 4.81% of the variance in mental distress (R² = .048, Adjusted R² = .037). Although statistically significant, the amount of explained variance was small, indicating that the included predictors account for only a limited proportion of differences in students’ mental distress.
Hypothesis Testing
Hypothesis 1 (H1)
Housing costs were not significantly associated with mental distress, B = -0.046, t(260) = -0.959, p = .338. The negative coefficient indicates that higher housing expenditures were associated with slightly lower levels of distress; however, this effect was very small and not statistically significant. This finding contradicts the initial hypothesis that higher financial burden would be associated with increased mental distress. Therefore, H1 was not supported.
Hypothesis 2 (H2)
Social support showed a statistically significant negative association with mental distress, B = -0.215, t(260) = -2.679, p = .008. This indicates that students who reported higher levels of social support tended to experience lower levels of psychological distress. The effect is consistent with theoretical expectations and prior research on the protective role of social relationships in mental health. Therefore, H2 was supported.
Hypothesis 3 (H3)
The interaction between housing costs and social support was negative, B = -0.133, t(260) = -1.858, p = .064. Although the direction of the effect is consistent with the stress-buffering hypothesis (i.e., social support potentially reducing the impact of financial strain on distress), the interaction did not reach statistical significance. This indicates only weak, trend-level evidence for a buffering effect, which is insufficient to confirm the hypothesis. Therefore, H3 was not supported.
Discussion / Conclusion
This study examined the relationship between housing costs, social support, and mental distress among university students. The findings indicate that social support is an important protective factor for mental well-being, as higher levels of social support were consistently associated with lower mental distress. In contrast, housing costs were not significantly related to mental distress. This suggests that housing expenditures may not be a valid proxy for subjective financial strain, as higher costs may also reflect greater financial resources rather than economic hardship. Furthermore, although the interaction between housing costs and social support was in the expected direction, the data did not provide strong evidence for a buffering effect of social support. Overall, the results highlight the importance of social relationships for student mental health, while financial indicators based solely on housing costs appear insufficient to capture subjective economic stress.
Limitations and Future Research
Several limitations should be considered. First, financial circumstances were measured using housing costs rather than subjective financial stress. Future research should include measures such as perceived financial strain, financial insecurity, or difficulty meeting expenses. Second, housing costs were treated as a quasi-continuous variable, although they were originally ordinal categories. This assumes equal distances between categories, which may not fully reflect reality. Finally, the cross-sectional design limits causal interpretation. Longitudinal studies would allow stronger conclusions about the directionality of relationships between financial conditions, social support, and mental health.
References
Allen, J., Balfour, R., Bell, R., & Marmot, M. (2014). Social determinants of mental health. International Review of Psychiatry, 26(4), 392–407. https://doi.org/10.3109/09540261.2014.928270
Åslund, C., Larm, P., Starrin, B., & Nilsson, K. W. (2014). The buffering effect of tangible social support on financial stress: Influence on psychological well-being and psychosomatic symptoms in a large sample of the adult general population. BMC Public Health, 14, Article 840. https://doi.org/10.1186/1471-2458-14-840
Cohen, S., & Wills, T. A. (1985). Stress, social support, and the buffering hypothesis. Psychological Bulletin, 98(2), 310–357. https://doi.org/10.1037/0033-2909.98.2.310
Hefner, J., & Eisenberg, D. (2009). Social support and mental health among college students. American Journal of Orthopsychiatry, 79(4), 491–499. https://doi.org/10.1037/a0016918
Kirkbride, J. B., Anglin, D. M., Colman, I., Dykxhoorn, J., Jones, P. B., Patalay, P., Pitman, A., Soneson, E., Steare, T., Wright, T., & Griffiths, S. L. (2024). The social determinants of mental health and disorder: Evidence, prevention and recommendations. World Psychiatry, 23(1), 58–90. https://doi.org/10.1002/wps.21160
McCloud, T., & Bann, D. (2019). Financial stress and mental health among higher education students in the UK up to 2018: Rapid review of evidence. Journal of Epidemiology and Community Health, 73(10), 977–984. https://doi.org/10.1136/jech-2019-212154