Page 60 - IJPS-9-1

International Journal of
Population Studies Neighborhood social cohesion and self-reported depression

one-tenth reported SRD. Bivariate analyses showed a
significant difference across age groups with those over UL 1.90 1.03 1.00 1.03 0.80 2.40 2.26 4.78 4.31
65 years of age scoring the highest levels of SRD. There 95% CI
was a significant difference in SRD by country of birth, LL 0.58 0.28 0.96 0.80 0.16 0.17 0.57 2.22 1.28
with those born in the United Kingdom scoring higher OR 1.05 0.54 0.98 0.91 0.35 0.64 1.14 3.26 2.34
(16.31%) than those born in Ireland (10.83%). There was
also a significant difference in SRD across domains of PSS Model 4 p 0.87 0.06 0.03 0.12 0.01 0.51 0.71 0.00 0.00
with lower social support related to depression. Likewise,
those who did not feel safe in their neighborhood were SE 0.30 0.33 0.01 0.06 0.41 0.67 0.35 0.20 0.31 103.29 (9), p=0.001
more likely to report SRD.
B 0.05 −0.62 −0.02 −0.10 −1.04 −0.45 0.13 1.18 0.85 328.05 0.31
3.2. Results of regression predicting SRD
Bivariate correlation between predictors showed the UL 1.86 0.99 0.99 1.02 0.68 2.50 2.21 5.16
strongest relationships between SRD and unemployment 95% CI
(r = −0.54, p = 0.001) and PSN and SRD (r = −0.18, LL 0.58 0.28 0.95 0.79 0.14 0.20 0.57 2.44
p = 0.001), which indicated multicollinearity was within OR 1.04 0.52 0.97 0.90 0.31 0.71 1.12 3.54
acceptable levels (table available on request). The variance
inflation factor values of the independent variables were Model 3 p 0.91 0.05 0.01 0.09 0.00 0.59 0.73 0.00
lower than the common cut-off threshold of 3.0 (Miles
& Shevling, 2001). Table 2 provides hierarchical multiple SE 0.30 0.33 0.01 0.06 0.40 0.65 0.35 0.19 97.76 (8), p=0.001
regression results predicting SRD. In the first model, both
PNS and PSS were significantly associated with SRD. For B 0.03 −0.65 −0.03 −0.11 −1.17 −0.35 0.12 1.27 333.96 0.29
each unit increase in PSS, participants were 1.20 times
less likely to report SRD. Feeling unsafe was related to 95% CI UL 2.01 0.76 1.03 0.95 0.46 3.04 2.93
SRD (OR = 0.22, p = 0.001 0.00) and those who reported LL 0.69 0.24 1.00 0.75 0.11 0.37 0.85
feeling safe were 4.54 times less likely to report SRD.
With the inclusion of risk exposure variables, the model OR 1.17 0.42 1.01 0.84 0.23 1.06 1.58
remained a significant predictor of SRD and accounted for Model 2 B: Standardized coefficients; CI: Confidence interval; LL: Lower limit; OR: Odds ratio; PD: Perceived discrimination; PNS: Perceived neighborhood safety; PSS: Perceived social support;
nearly a quarter of the variance in SRD. The inclusion of p 0.56 0.00 0.07 0.00 0.00 0.92 0.15
unemployment and social housing in the second step and SE 0.27 0.30 0.01 0.06 0.36 0.54 0.32
SRGH in the third step made minimal difference to the 44.83 (7), p=0.001
model. With the inclusion of perceived discrimination in B 0.16 −0.86 0.01 −0.17 −1.48 0.06 0.46 386.54 0.14
the fourth step, PSS no longer significantly predicted SRD.
On the other hand, PNS remained a significant predictor of UL 2.01 0.83 1.02 0.94 0.43
SRD (OR = 0.35, p = 0.01). Inverting this statistic showed 95% CI
that with those reporting feeling safe in their neighborhood LL 0.69 0.27 1.00 0.74 0.11
were 2.85 times less likely to report SRD. In the final two
models, those who had perceived any form of discrimination OR 1.18 0.48 1.01 0.83 0.22
in the past year were significantly more likely to report SRD. Model 1 p 0.54 0.01 0.16 0.00 0.00
We conducted a backward selection, removing the variables
with the lowest predictive power (employment, gender, and SE 0.27 0.29 0.01 0.06 0.35 42.68 (5), p=0.001
housing). We observed no differences in model fit statistics 388.64
and significances of predictor variables. Table 2. Logistic regression predicting self‑reported depression B 0.17 −0.74 0.01 −0.18 −1.53 0.13
The model with interaction effects was a significant
predictor of SRD, R = 0.36, F (1, 10) = 118.39, p = 0.001.
2
Results showed a significant interaction between PNS and
age, although the effect was small (B = 0.07, p = 0.001,
OR = 1.07). Table 3 provides data from the pick-a-point Country of birth (UK/Ireland) SE: Standard error; UL: Upper limit
method. As illustrated by Figure 1, at 65 years, PNS was Employment (Yes/No) Housing (Social/Private)
not significantly associated with SRD (B = 0.11, SE = 0.58, Gender (Female)
p = 0.84). For participants at 45 years old, feeling safe PSS (Scale) PNS (No) SRGH PD (Yes) Model fit
in their neighborhood was associated with reduced Age χ 2 ‑2LL R 2

Volume 9 Issue 1 (2023) 54 https://doi.org/10.36922/ijps.431