Table 4 Changes in self-reported study park satisfaction among adult residents living in intervention vs. control park neighborhoods.

From: Park use patterns and park satisfaction before and after citywide park renovations in low-income New York City neighborhoods

Study park satisfaction

Intervention park neighborhoods

Control park neighborhoods

Difference-in-differences

Pre-renovationa (n = 545)

Post-renovationa (n = 201)

Change (95% CI)b

Pre-renovationa (n = 345)

Post-renovationa (n = 129)

Change (95% CI)b

Unadjusted DID estimator (95% CI)c

Adjusted DID estimator (95% CI)c,d

p-value for adjusted DID Estimatore

Percent satisfied with overall park quality

34.9 (2.4)

72.5 (2.9)

37.6 (30.2, 45.0)

42.2 (2.9)

41.4 (4.9)

− 0.8 (− 12.0, 10.3)

38.5 (25.0, 51.9)

38.4 (25.2, 51.6)

 < 0.001

Percent satisfied with park facilities

31.3 (2.0)

63.6 (3.6)

32.2 (24.2, 40.3)

37.4 (2.7)

37.1 (4.1)

− 0.3 (− 9.9, 9.3)

32.6 (20.1, 45.1)

33.1 (20.7, 45.6)

 < 0.001

Percent satisfied with playground

37.8 (2.1)

70.3 (3.2)

32.4 (24.9, 40.0)

43.1 (3.9)

41.1 (5.4)

− 2.0 (− 15.0, 11.0)

34.4 (19.4, 49.5)

35.1 (20.1, 50.1)

 < 0.001

Percent satisfied with walking/cycling tracks

26.4 (2.3)

58.7 (3.6)

32.3 (24.0, 40.7)

31.6 (3.8)

37.8 (5.4)

6.2 (− 6.7, 19.0)

26.1 (10.8, 41.5)

27.3 (12.5, 42.1)

0.001

Percent satisfied with the maintenance of the grounds and facilities

39.0 (2.3)

70.6 (3.1)

31.6 (24.0, 39.2)

48.0 (3.1)

38.8 (4.6)

− 9.1 (− 20.1, 1.8)

40.7 (27.4, 54.1)

40.9 (27.7, 54.1)

 < 0.001

  1. All estimates presented in table were pooled across GEE models fit to 25 imputed data sets. All GEE models included variables for time (pre-renovation vs. post-renovation), intervention group (renovated vs. control parks), and an interaction between time and intervention group (the DID estimator).
  2. BMI body mass index, CI confidence interval, DID difference-in-differences, GEE generalized estimating equations, SE standard error.
  3. aUnadjusted percent (SE).
  4. bUnadjusted within-group change (post–pre).
  5. cDifference in change in the given outcome measure in the intervention group minus the control group, as estimated by the interaction term between time * intervention group.
  6. dAdjusted for age group, BMI category, annual household income, public housing, marital status, and children in household.
  7. eCorrected using the Benjamini–Hochberg procedure to control the false discovery rate; significant (p < 0.05) p-values are bolded.
Back to article page