Skip to main content

Table 7 Hierarchical linear regression analysis of predictors of physical health (PHQ) score a, b, c, d

From: Quantifying the relationship between gardening and health and well-being in the UK: a survey during the covid-19 pandemic

Predictor variables (reference)

Model 1

Model 2

Model 3

Model 4

B (SE)

p

B (SE)

p

B (SE)

p

B (SE)

p

Constant

41.37 (1.59)

< 0.001

48.34 (3.60)

< 0.001

45.31 (3.46)

< 0.001

45.43 (3.58)

< 0.001

Gender (Female)

       

 Male

-2.36 (1.63)

0.15

-1.64 (1.57)

0.30

-2.85 (1.53)

0.06

-2.69 (1.53)

0.08

Age (18–34)

        

 35–54

-3.55 (1.99)

0.08

-4.48 (2.1)

< 0.05

-6.08 (1.92)

< 0.01

-5.70 (1.97)

< 0.01

 55+

-10.61 (1.86)

< 0.001

-10.85 (2.07)

< 0.001

-12.19 (1.99)

< 0.001

-11.11 (2.22)

< 0.001

IMD quintile (First)

       

 Second

-6.89 (3.14)

< 0.05

-5.21 (3.00)

0.08

-5.13 (3.00)

0.09

 Third

-2.11 (2.96)

0.48

-1.41 (2.81)

0.62

-1.10 (2.82)

0.70

 Fourth

-4.74 (3.01)

0.12

-2.90 (2.86)

0.31

-2.83 (2.86)

0.32

 Fifth

-6.04 (2.93)

< 0.05

-5.11 (2.77)

0.07

-4.94 (2.79)

0.08

Higher education (No)

    

 Yes

-4.33 (1.74)

< 0.05

-4.05 (1.64)

< 0.05

-3.89 (1.65)

< 0.05

Household (Alone)

      

 With partner

-0.29 (2.04)

0.89

-0.79 (1.95)

0.69

-0.56 (1.97)

0.77

 With family

4.64 (2.36)

0.05

4.62 (2.24)

< 0.05

4.83 (2.24)

< 0.05

 Shared accommodation

-5.57 (4.52)

0.22

-5.39 (4.29)

0.21

-6.42 (4.38)

0.14

Obesity (Not obese)

     

 Obese

3.65 (1.84)

< 0.05

3.26 (1.89)

0.09

Long-term conditions (No)

   

 Yes

4.42 (1.35)

< 0.01

4.21 (1.37)

< 0.01

Alcohol consumption

(Low risk)

   

 Increasing risk

4.47 (1.70)

< 0.01

4.43 (1.73)

< 0.05

Weekly gardening time (0 h)

      

 1–5 h

    

4.21 (2.23)

0.06

 6–10 h

    

1.97 (2.63)

0.45

 11 + hours

    

2.89 (2.58)

0.26

Food growing (No)

      

 Yes

    

-4.32 (2.34)

0.07

R2

0.18

0.29

0.38

0.40

Adjusted R2

0.17

0.25

0.33

0.34

  1. a Lower PHQ scores indicate fewer health complaints and thus better health
  2. b Model 1: adjusted for age and gender; Model 2: adjusted for sociodemographic variables; Model 3: adjusted for sociodemographic variables and known health risk and relevant lifestyle factors; Model 4: adjusted for sociodemographic variables, health risk and relevant lifestyle factors and variables related to gardening
  3. c Predictors and coefficients in the table are derived from the best fit models for the outcome based on the Bayesian Information Criterion (BIC) at each stage of the regression. Other explanatory variables tested include household income, caring responsibilities, smoking status, physical activity level, F&V intake, and having an allotment, but these were dropped in the process of improving model fit
  4. d Figures in bold are statistically significant at the 5% level (p < 0.05)