ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared,...

27
Running head: PERCEIVED RESTORATIVE POTENTIAL 1 Appendix A Table 1 Descriptive Statistics for Demographic Variables in Study 1 and Study 2 Study 1 Study 2 Measure n % n % Total Sample 722 281 Gender Male 294 40.7 115 40.9 Female 428 59.3 166 59.1 Not reported 0 0.0 0 0.0 Race/Ethnicity Caucasian/White 554 76.7 224 79.7 African American/Black 65 9.0 21 7.5 Hispanic or Latino 40 5.5 16 5.7 Asian 41 5.7 15 5.3 Middle Eastern 4 0.6 0 0.0 Other 17 2.4 5 1.8 Not reported 1 0.1 0 0.0 Native English speaker Yes 705 97.6 276 98.2 No 17 2.4 5 1.8 Annual income Under $35,000 274 38.0 119 42.3 $35,000 to $49,999 139 19.3 56 19.9 $50,000 to $64,999 87 12.0 36 12.8 $65,000 to $79,999 73 10.1 31 11.0 $80,000 to $94,999 37 5.1 8 2.8 $95,000 to $109,999 36 5.0 14 5.0 $110,000 to $124,999 19 2.6 9 3.2 $125,000 to $139,999 14 1.9 3 1.1 $140,000 to $154,999 5 0.7 3 1.1 $155,000 to $169,999 7 1.0 1 0.4 $170,000 to $184,999 6 0.8 0 0.0 $185,000 to $199,999 3 0.4 0 0.0 $200,000 to $214,999 5 0.7 0 0.0 $215,000 to $229,999 0 0.0 0 0.0 $230,000 to $249,999 4 0.6 0 0.0 More than $250,000 11 1.5 1 0.4 Not reported 2 0.3 0 0.0 Education

Transcript of ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared,...

Page 1: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

Running head: PERCEIVED RESTORATIVE POTENTIAL 1

Appendix A

Table 1

Descriptive Statistics for Demographic Variables in Study 1 and Study 2

Study 1 Study 2Measure n % n %Total Sample 722 281Gender Male 294 40.7 115 40.9 Female 428 59.3 166 59.1 Not reported 0 0.0 0 0.0Race/Ethnicity Caucasian/White 554 76.7 224 79.7 African American/Black 65 9.0 21 7.5 Hispanic or Latino 40 5.5 16 5.7 Asian 41 5.7 15 5.3 Middle Eastern 4 0.6 0 0.0 Other 17 2.4 5 1.8 Not reported 1 0.1 0 0.0Native English speaker Yes 705 97.6 276 98.2 No 17 2.4 5 1.8Annual income Under $35,000 274 38.0 119 42.3 $35,000 to $49,999 139 19.3 56 19.9 $50,000 to $64,999 87 12.0 36 12.8 $65,000 to $79,999 73 10.1 31 11.0 $80,000 to $94,999 37 5.1 8 2.8 $95,000 to $109,999 36 5.0 14 5.0 $110,000 to $124,999 19 2.6 9 3.2 $125,000 to $139,999 14 1.9 3 1.1 $140,000 to $154,999 5 0.7 3 1.1 $155,000 to $169,999 7 1.0 1 0.4 $170,000 to $184,999 6 0.8 0 0.0 $185,000 to $199,999 3 0.4 0 0.0 $200,000 to $214,999 5 0.7 0 0.0 $215,000 to $229,999 0 0.0 0 0.0 $230,000 to $249,999 4 0.6 0 0.0 More than $250,000 11 1.5 1 0.4 Not reported 2 0.3 0 0.0Education < High school diploma 4 0.6 2 0.7 High school diploma 85 11.8 28 10.0 Some college or voc. 228 31.6 105 37.4 2-year college degree 77 10.7 29 10.3 4-year college degree 238 33.0 90 32.0 Post college degree 90 12.5 27 9.6

Page 2: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 2

Appendix B

Image Selection Procedure for Study 1

We selected 60 images from a larger database of 180 images (60 natural, 60 mixed, and

60 built). We sourced images from an online open-source image database (Wikimedia

Commons) and from two professors, one a landscape architect and the other an urban planner.

Our intention was to obtain a final image set for which average visual appeal did not

significantly differ among natural, built, and mixed categories, and for which a range of visual

appeal was represented in each image category. This would allow us to better control for the

potential influence of visual appeal on the relationship between naturalness and perceived

restoration.

To choose the final 60 images used in both experiments, a separate sample of 56

undergraduate students (30 men; Mage = 18.65, SDage = 0.91) rated the 180 images on visual

appeal as part of an online study for which they received course credit. Images were randomly

presented through the web-based survey program Qualtrics. Participants rated how visually

appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all

visually appealing) to 100 (very visually appealing). The values in between the numbered scale

anchors were not visible to participants. We decided to use a large ranged scale to capture more

precisely the potential variability across environments.

We calculated mean visual appeal for each image. Within each environment category

(natural, mixed, built), the 60 images were binned into 10 relatively equal ranges of visual appeal

ratings (average bin sizes: built = 5.28 visual appeal units; mixed = 6.39 visual appeal units;

natural = 5.92 visual appeal units). We created bins to ensure that the final image set included

images ranging from low to high visual appeal within each image category. Depending on the

Page 3: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 3

distribution of visual appeal ratings within each category, each bin contained between 1 and 12

images. To create the final image set for Study 1, two images were randomly selected from each

bin, which allowed the entire range of visual appeal ratings to be represented. The exceptions to

this rule were: 1) we always selected nature images containing people. This is because those

examples were rare but necessary, in order to make comparisons to the mixed and built images

that contained people; and, 2) if a bin only contained one image, then the next highest or lowest

rated image was moved into that bin so that at least two images were represented in every bin.

For the final image set, mean visual appeal still significantly differed among environment

categories, F(2,57) = 4.88, p = .01, but only when comparing natural to built environments, t(59)

= 3.12, d = 0.99, 95% CI on d [0.32, 1.64], p = .008. Natural environments did not differ from

mixed environments, t(59) = 1.45, d = 0.45, 95% CI on d [-0.17, 1.08], p = .32, and mixed

environments did not differ from built environments, t(59) = 1.67, d = 0.53, 95% CI on d [-0.11,

1.16], p = .22 (natural: M = 57.28, SD = 17.96; mixed: M = 49.33, SD = 18.55; built: M = 40.17,

SD = 15.33). Thus, we were able to use images ranging in visual appeal, but we could not

perfectly equate visual appeal across environment categories.

Page 4: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 4

Appendix C

Table 2

Descriptive Statistics for Individual Differences Questionnaire in Study 1 and Study 2

Study 1 Study 2Measure, Scale range range M SD alpha range M SD alphaShyness & Sociability Scales for Adults Sociability, 1 (low) to 5 (high) 1 – 5 2.63 0.88 .83 Shyness, 1 (low) to 5 (high) 1 – 5 3.09 1.00 .85Big Five Extraversion, -3 (low) to 3 (high) -3 – 3 -0.39 1.59 .67Number of hours spent indoors today, 0-24 hours 0 – 24 8.02 4.71Modified Pittsburgh Sleep Quality Index .75 Hours of sleep last night 1 – 15 7.10 1.70 Sleep quality last night, 1 (very poor) to 5 (very good) 1 – 5 3.42

0.99

How rested do you feel?, 1 (not at all) to 5 (very) 1 – 5 3.22 1.18 Average hours of sleep past week 3 – 12 7.07 1.29 Sleep quality past week, 1 (very poor) to 5 (very good) 1 – 5 3.36

0.87

Circumplex Mood Scale Arousal, -40 (low) to 40 (high) -40.00 – 20.00 -12.36 9.31 .63 -31 – 20 -10.79 9.27 .62 Valence, -40 (negative) to 40 (positive) -37.00 – 40.00 9.89 13.03 .86 -40 – 40 12.64 13.92 .89Multidimensional Fatigue Inventory General fatigue, 4 (low) to 20 (high) 4 – 20 11.51 4.00 .79 4 – 20 11.06 3.94 .80 Physical fatigue, 4 (low) to 20 (high) 4 – 20 10.61 4.13 .81 4 – 20 10.23 3.95 .81 Reduced activity, 4 (low) to 20 (high) 4 – 20 10.53 3.80 .78 4 – 20 9.96 3.63 .76 Reduced motivation, 4 (low) to 20 (high) 4 – 20 10.15 3.46 .66 4 – 20 9.50 3.49 .71 Mental fatigue, 4 (low) to 20 (high) 4 – 20 9.48 3.74 .81 4 – 20 8.99 3.59 .81 Total fatigue, 20 (low) to 100 (high) 20 – 98 52.27 15.50 .92 20 – 100 49.75 15.44 .93Perceived Stress Scale, 0 (low) to 40 (high) 0 – 38 15.09 7.89 .90

Page 5: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 7

Appendix D

Excluding Visual Appeal from Linear Mixed Effects Model

Including visual appeal as a predictor of perceived restorative potential may have masked

effects of potentially relevant individual difference variables because of the high correlation

between visual appeal and perceived restorative potential (r = .70). As such, we conducted the

same analysis as reported in the manuscript (Table 2) but excluded visual appeal as a predictor

variable. When excluding visual appeal from the model, all previously reported effects remained

(see Table 3, Model 1, Appendix D). In addition, gender moderated the relationship between

naturalness and perceived restorative potential such that men showed a smaller difference in

perceived restorative potential for natural compared to built environments, relative to women.

Intercept significantly varied by participant, SD = 1.01, 95% CI: [0.95, 1.07], and by image SD =

1.82, 95% CI: [1.53, 2.20]

Linear Mixed Effects Model Predicting Visual Appeal

Because we found in Study 1 that visual appeal partially mediated the relationship

between naturalness and perceived restorative potential, individual differences may affect

perceived restoration through visual appeal. We fit the same linear mixed effects model as

reported in the manuscript (Table 2), except that visual appeal was the outcome variable.

The results were similar to the model reported in the manuscript with the following

exceptions: 1) there were additional main effects of fatigue, population for current location, and

population for longest lived location; 2) as with Model 1 Supplemental, gender moderated the

relationship between naturalness and perceived restorative potential; and 3) the moderating effect

of income was in the opposite direction. Overall, more fatigued individuals reported lower visual

appeal ratings. Participants currently living in more populated areas reported lower visual appeal

Page 6: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 6

ratings, but participants living in more populated areas for the longest reported higher visual

appeal ratings. It is unclear why participant responses would be biased in this way based on

residential history. Finally, there was a slightly stronger relationship between naturalness and

visual appeal when income increased; this effect is opposite to the moderating effect of income

on the relationship between naturalness and perceived restorative potential shown in Table 2 of

the manuscript. Intercept significantly varied by participant, SD = 0.94, 95% CI: [0.89, 1.00] and

by image, SD = 1.56, 95% CI: [1.31, 1.89].

Page 7: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 7

Table 3

Study 1 Fixed Effects Estimates for Multilevel Model Predicting Perceived Restorative Potential

Model 1 – Without Visual Appeal Model 2 – Predicting Visual Appeal95% CI 95% CI

Variable B SE t lower upper B SE t lower upperMain Effects intercept -0.149 0.245 0.317 0.212 naturalness 0.154 0.011 13.43 0.131 0.177 0.128 0.010 12.37 0.108 0.148 people -0.052 0.010 -5.16 -0.071 -0.032 -0.027 0.009 -3.03 -0.045 -0.010 fatigue -0.079 0.049 -1.62 -0.175 0.017 -0.153 0.045 -3.37 -0.242 -0.064 sleep quality 0.072 0.044 1.65 -0.014 0.158 0.066 0.041 1.63 -0.014 0.146 extraversion 0.072 0.054 1.32 -0.035 0.178 0.048 0.050 0.95 -0.051 0.147 pop. current location -0.087 0.060 -1.45 -0.204 0.031 -0.112 0.056 -2.02 -0.221 -0.003 pop. longest location 0.089 0.060 1.49 -0.028 0.206 0.116 0.055 2.09 0.007 0.225 age 0.054 0.042 1.30 -0.028 0.136 -0.005 0.039 -0.13 -0.081 0.071 gender (male) 0.065 0.084 0.78 -0.100 0.231 0.027 0.078 0.34 -0.127 0.181 income -0.020 0.042 -0.48 -0.103 0.062 -0.035 0.039 -0.89 -0.112 0.042 education (no degree) 0.027 0.084 0.32 -0.138 0.192 -0.025 0.078 -0.32 -0.178 0.128

Interactions with naturalness

people 0.007 0.002 4.07 0.004 0.010 0.005 0.002 3.19 0.002 0.008 fatigue -0.029 0.004 -7.02 -0.037 -0.021 -0.020 0.004 -5.41 -0.028 -0.013 sleep quality 0.004 0.004 0.94 -0.004 0.011 -0.003 0.003 -0.98 -0.010 0.003 extraversion -0.032 0.005 -6.85 -0.041 -0.023 -0.026 0.004 -6.25 -0.034 -0.018 pop. current location 0.003 0.005 0.60 -0.007 0.014 0.006 0.005 1.15 -0.004 0.015 pop. longest location -0.027 0.005 -5.03 -0.037 -0.016 -0.016 0.005 -3.35 -0.025 -0.007 age 0.034 0.004 9.58 0.027 0.041 0.033 0.003 10.39 0.027 0.040 gender (male) -0.041 0.007 -5.66 -0.056 -0.027 -0.058 0.007 -8.79 -0.071 -0.045 income -0.011 0.004 -3.03 -0.018 -0.004 0.011 0.003 3.43 0.005 0.017

Page 8: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 8

education (no degree) 0.053 0.007 7.39 0.039 0.068 0.049 0.007 7.45 0.036 0.061

Page 9: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 9

Appendix E

Using mediation package in R to conduct multi-level mediation analyses

We tested if naturalness affects perceived restorative potential indirectly through visual

appeal. To test this mediation model, we used the mediation package in R (Tingley, Yamamoto,

Hirose, Keele, & Imai, 2014), which draws on 1000 bootstrapped samples to calculate parameter

estimates and quasi-Bayesian 95% confidence intervals. For all reported linear mixed effects

models, participant was a random effect. Note that the interpretation of this mediation analysis is

consistent with what we reported in the manuscript. However, this method overestimates the

parameter coefficients because it cannot account for two random effects (participant and image).

Naturalness significantly predicted visual appeal, a = .282, SE = .004, 95% CI

[.275, .290], t(21656) = 73.87, and visual appeal significantly predicted perceived restorative

potential, while controlling for naturalness, b = .661, SE = .006, 95% CI [.650, .673], t(21655) =

111.99. The indirect effect of naturalness on perceived restorative potential was statistically

significant, ab = .187, 95% CI [.181, .190], p < .001. The direct effect of naturalness on

perceived restorative potential remained significant after controlling for visual appeal, c’ = .215,

95% CI [.208, .220], p < .001, and the total effect of naturalness on perceived restorative

potential was c = .402, 95% CI [.394, .410], p < .001. We can conclude that visual appeal

partially mediates the effect between naturalness and perceived restorative potential, accounting

for 46.4% (95 CI [45.2, 48.0]) of the relationship.

We tested an alternative mediation model in which naturalness affects visual appeal

indirectly though perceived restorative potential. Naturalness significantly predicted perceived

restorative potential, a = .401, SE = .004, 95% CI [.393, .410], t(21656) = 96.15, and perceived

restorative potential significantly predicted visual appeal, while controlling for naturalness, b

= .555, SE = .004, 95% CI [.545, .564], t(21655) = 112.00. The indirect effect of naturalness on

Page 10: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 10

visual appeal was statistically significant, ab = .223, 95% CI [.217, .230], p < .001. The direct

effect of naturalness on visual appeal remained significant after controlling for perceived

restorative potential, c’ = .060, 95% CI [.053, .070], p < .001, and the total effect of naturalness

on visual appeal was c = .282, 95% CI [.275, .290], p < .001. We can conclude that perceived

restorative potential partially mediates the effect between naturalness and visual appeal,

accounting for 78.9% (95% CI [76.7, 81.0]) of the relationship.

Page 11: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 11

Appendix F

Table 4 Study 1 Means, Standard Deviations, and Zero-Order Correlations Variable M SD 1 2 3 4 5

1. Age 34.24 12.56 2. Pop. current 1129953 1907067 -.05 [-.12, .03] 3. Pop. longest 1091162 1824342 .04 .72** [-.03, .11] [.69, .76] 4. Income 3.13 2.92 -.03 .05 .06 [-.10, .05] [-.02, .12] [-.02, .13] 5. Fatigue 0.00 1.00 -.08* .01 -.03 -.10** [-.15, -.00] [-.07, .08] [-.10, .04] [-.17, -.03] 6. Wakefulness 0.00 1.00 .00 .06 .06 .11** -.34** [-.07, .08] [-.01, .13] [-.02, .13] [.04, .18] [-.41, -.28]

Note. Correlations are performed on the z-score transformed data. M and SD represent mean and standard deviation, respectively, and were calculated using the untransformed data for ease of interpretation. Fatigue and wakefulness are factor scores. Values in square brackets indicate the 95% confidence interval for each correlation. * indicates p < .05. ** indicates p < .01.Table 5 Study 2 Means, Standard Deviations, and Zero-Order Correlations

Variable M SD 1 2 3

1. Fatigue 49.75 15.44 2. Perceived stress 15.09 7.89 .63** [.56, .70] 3. Valence 12.64 13.92 -.67** -.61** [-.73, -.60] [-.67, -.53] 4. Arousal -10.79 9.27 -.40** -.27** .43** [-.49, -.29] [-.38, -.16] [.33, .52]

Page 12: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 12

Note. Correlations are performed on the z-score transformed data. M and SD represent mean and standard deviation, respectively, and were calculated using the untransformed data for ease of interpretation. Values in square brackets indicate the 95% confidence interval for each correlation. * indicates p < .05. ** indicates p < .01.

Page 13: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 13

Appendix G

Distributions before and after logit transformation

Naturalness – Study 1

Perceived restorative potential – Study 1

Page 14: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 14

Visual appeal – Study 1

Presence of people – Study 1

Page 15: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 15

Perceived restorative potential – Study 2

Page 16: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 16

Appendix H

Table 6

Study 1 Variance Inflation Factors and Tolerance for Mixed Effects Model

Parameter VIF Tolerance

Main effects

naturalness1.68 0.60

visual appeal1.03 0.97

people1.06 0.94

fatigue1.43 0.70

wakefulness1.15 0.87

extraversion1.28 0.78

pop. current residence2.15 0.47

pop. longest residence2.14 0.47

age1.05 0.96

gender1.04 0.97

income1.06 0.94

education1.05 0.96

Interactions with naturalness

visual appeal1.02 0.98

people1.42 0.70

fatigue1.44 0.69

Page 17: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 17

wakefulness1.15 0.87

extraversion1.29 0.78

pop. current residence2.16 0.46

pop. longest residence2.15 0.47

age1.05 0.95

gender1.16 0.86

income1.07 0.93

education1.22 0.82

Note. VIF = Variance inflation factor; tolerance = 1/VIF

Table 2

Study 2 Variance Inflation Factors and Tolerance for Mixed Effects Model

Parameter GVIF df GVIF

adjusted

Tolerance Tolerance

adjusted

Main effects

category 1.00 2 1 1 1

fatigue 2.49 1 1.58 0.4 0.63

perceived stress 2.12 1 1.45 0.47 0.69

valence 2.44 1 1.56 0.41 0.64

Page 18: ars.els-cdn.com · Web viewParticipants rated how visually appealing each environment appeared, using a sliding continuous scale ranging from 0 (not at all visually appealing ) to

PERCEIVED RESTORATIVE POTENTIAL 18

arousal 1.44 1 1.2 0.69 0.83

Interactions with category

fatigue 5.44 2 1.53 0.18 0.65

perceived stress 3.92 2 1.41 0.25 0.71

valence 5.24 2 1.51 0.19 0.66

arousal 1.83 2 1.16 0.55 0.86

Note. GVIF = Generalized variance inflation factor; GVIF adjusted = GVIF^(1/2*df); tolerance = 1/GVIF; tolerance adjusted = 1/ GVIF^(1/2*df)