Background

There have been many iterations of a Connection to Nature (CTN) measure (see Ives et al., 2017, Restall & Conrad, 2015, and Tam 2013 for review). Several of these measures have proposed multi-dimensional structure (Davis, Le, and Coy 2011; Dong et al. 2020; Hatty et al. 2020; Kurth et al. 2020; Meis-Harris, Borg, and Jorgensen 2021; Olivos and Aragonés 2011), however there is little consistency across these dimensional structures. We build upon two recently published measures Meis-Harris, Borg, and Jorgensen (2021), which ground their dimensional structure in a thorough cross-disciplinary review of the nature connectedness literature. We extend their findings by addressing concerns that measures of CTN are vulnerable to self-report biases. In particular, concerns have been raised that people may be more likely to endorse CTN items out of social desirability or may simply not be able to express the abstract concept of connection to nature when asked directly (Gould and Schultz 2021). A possible solution to these issues is proposed by the Disposition to Connect with Nature measure which asks about people’s bonding behavior with nature instead of simply asking about preferences to being in nature (Brügger, Kaiser, and Roczen 2011). We adapted items from the Multi-dimensional CTN scale (Meis-Harris, Borg, and Jorgensen 2021) and the Disposition to Connect with Nature scale (Brügger, Kaiser, and Roczen 2011) as well as including several items that we felt addressed the concerns raised about self-report bias.

Across the three exploratory factor analyses, four factors consistently arose as both the best fitting model and aligning with our theoretical framework. We further tested these four factors in a confirmatory factor analysis with a fourth sample and found adequate fit. The dimensions include:

1. Deep (deeply seeing the self as part of nature)
2. Experiential (spending and enjoying time in nature)
3. Emotional (Emotional desire to connect with nature and protect it.)
4. Presence (engaging mindfully and consciously with nature)

Selecting Initial Items

Search of existing scales

An extensive search of the literature was conducted to source all existing adult measures of CTN. Based on this search, we identified two measures that showed promise for further adaptation: The AIMES Multidimensional CTN Scale (Meis-Harris, Borg, and Jorgensen 2021) and the Disposition to Connect with Nature scale (Brügger, Kaiser, and Roczen 2011). Our search identified that a multi-dimensional structure should be maintained for the measure, however existing multi-dimensional measures showed conflicting dimensional structure. (Hatty et al. (2020) & Meis-Harris, Borg, and Jorgensen (2021) - the later having a more complex and theoretically driven dimensional structure which is why it was chosen to retain for adaptation). The Disposition to Connect with Nature scale addressed issues of self-report bias though given it was written for a European sample the items were not generalizable to other populations. The items from these two scales resulted in a total of 59 items (AIMES = 19 items; Dispositional CTN = 40 items). Items were reworded for generalizability and clarity. Any reference to “connection to nature” was removed to avoid concerns about participants not fully understanding this concept. Items were also removed if they were redundant. This left a total of 39 items adapted from existing measures (AIMES = 17 items; Dispositional CTN = 28). See Table 1 below for all items considered for inclusion in the DEEP Connection to Nature Scale.

Development of new items

We chose to build upon the strong theoretical basis for the dimensional structure of the AIMES with some changes. Meis-Harris, Borg, and Jorgensen (2021) based the dimension of the AIMES (Affection, Identity, Materialism, Experience, & Spirituality) off Ives et al. (2018) which lists five aspects of human-nature relationships that can lead to behavioral change. We agree with that three of these dimensions are part of the construct of CTN (these are also widely agreed to be part of the construct of CTN in the literature):

1. Experiential (time spent in nature)
2. Cognitive (beliefs about the self’s relationship with nature)
3. Emotional (affective relationship with nature).

However, they included two value-based dimensions which, while strongly correlated to CTN, have been defined as separate constructs (Schwartz 1992, 1999; Schultz 2001; Stern and Dietz 1994):

4. Material values towards nature (viewing nature as a resource for human extraction e.g., extrinsic values)
5. Philosophical views of nature (viewing humans as interrelated with nature e.g., universal values)

We chose to remove the the material dimension entirely and adapt the philosophical dimension so it more accurately reflected the individual level of CTN. That is, it measured one’s own beliefs about how, as an individual, they are connected to nature. In line with this re-analysis of the construct of CTN we further adapted the existing 39 items and created an additional 14 items in order to capture the following dimensions of CTN:

1. Experiential
2. Cognitive
3. Emotional
4. Spiritual (formerly philosophical)

Items were created to capture, behavioral, embodied, and experiential relationships with nature (in line with Brügger et al., 2011’s Disposition to CTN) in order to overcome concerns with self-report bias. This resulted in items that asked about conscious attendance to nature. There was also an effort to create items that updated previous philosophical items to more individual relationships with nature which tapped into a more spiritual relationship. This resulted in a total of 58 items.

Table 1. Items considered for inclusion in the DEEP Connection to Nature Scale

Study 1

Exploratory Factor Analysis

Sample

Two general population Prolific samples and one student sample were collected from November 2022 until April 2023 (See Table 2 for details of these samples) Three exploratory factor analyses were conducted on these samples.

Table 2. Sample details across three exploratory factor analyses

Items Removed

Sample 1

Several items were removed in sample 1 before the EFA was conducted. These items were also excluded from samples 2 and 3.

A test for overdetermination revealed that the model may be overdetermined due to high colinearity between items (det <.00001). Inspection of the correlations revealed five items with highly colinearity (see Figure 1). Three of these items were removed .

Figure 1. Correlations of items > .80

The decision was made to remove AIMES_15, AIMES_7, and AIMES_2. However, removing these items did not improve the risk of overdetermination (det <.00001).

Further assumptions were explored to see if other items should be removed. A Bartlett’s test of sphericity found there was adequate shared variance between items (𝜲2 = 17044.9(1225), p <.001).

A Kaiser-Meyer-Olkin test of factor adequacy found that, overall, there was adequate variance within items (MSA = 0.97). However, exploration of individual items identified three items which may not have sufficient variance: DISPO_32, DISPO_38, & DISPO_36 (MSA = 0.89, MSA = 0.82, & MSA = 0.84 respectively). Inspection for ceiling and floor effects confirms this, as well as identifying several other items that may have limited variance (See Figure 2). Several items also appeared to have high positive or negative skew (See Figure 3). Five items were removed due to high skewedness (NEW_9, DISPO_38, DISPO_39, DISPO_6, & DISPO_36). A final check of overdetermination revealed there is still the potential for the fit to be overdetermined (det <.00001).

Figure 2. Box plots depicting ceiling and floor effects of items

Figure 3. Histograms depicting skewedness of items

Sample 2

A test for overdetermination revealed that the model may be overdetermined due to high colinearity between items (det <.00001). Inspection of the correlations no items with highly colinearity. 

Further assumptions were explored to see if other items should be removed. A Bartlett’s test of sphericity found there was adequate shared variance between items (𝜲2 = 8385.44(703), p <.001). A Kaiser-Meyer-Olkin test of factor adequacy found that, overall, there was adequate variance within items (MSA = 0.95). Exploration of individual items showed that no single item was < .80. Inspection of box plots and histograms of distribution showed not items with ceiling/floor effects or high skewedness.

Sample 3

A test for overdetermination revealed that the model may be overdetermined due to high colinearity between items (det <.00001). Inspection of the correlations no items with highly colinearity. 

Further assumptions were explored to see if other items should be removed. A Bartlett’s test of sphericity found there was adequate shared variance between items (𝜲2 = 8353.44(73), p <.001). A Kaiser-Meyer-Olkin test of factor adequacy found that, overall, there was adequate variance within items (MSA = 0.95). Exploration of individual items showed that no single item was < .80. Inspection of box plots and histograms of distribution showed not items with ceiling/floor effects or high skewedness.

Number of Factors

Despite the potential for overdetermination, an EFA was conducted using a principal analysis method of extraction. This method has been found to be robust with models with non-normal data (Briggs & MacCullum, 2003). The Psych package in R was used with a promax rotation and principal analysis factoring method.

The number of factors to be extracted was estimated using both a scree plot and parallel analysis. The screen plot suggested ~3-4 factors, while parallel analysis suggested 6 factors across all three samples. (See Figure 3)

Figure 3. Scree plots and parallel analysis of the number of factors across three samples

Four models were run across all three samples:

• Six factors
• Five factors
• Four factors
• Simplified four factors (removing dual loading and poor loading items from four factor model)

Item Selection Based on Model Fit

The simplified 4-factor model may be still be overdetermined: determinant < 0.00001 (in all 3 samples), even though we have removed items with high covariance.

Tests of shared variance between items and variance within items both suggest the data is suitable for factor analysis.

• Sample 1

  • Bartlett’s test of sphericity showed a 𝜲2 = 3171.39(595), p <.001).

  • Kaiser-Meyer-Olkin test of factor adequacy found that, overall, there was adequate variance within items (MSA = 0.97).

• Sample 2

  • Bartlett’s test of sphericity showed a 𝜲2 = 6674.74(406), p <.001).

  • Kaiser-Meyer-Olkin test of factor adequacy found that, overall, there was adequate variance within items (MSA = 0.95).

• Sample 3

  • Bartlett’s test of sphericity showed a 𝜲2 = 6674.74(406), p <.001).

  • Kaiser-Meyer-Olkin test of factor adequacy found that, overall, there was adequate variance within items (MSA = 0.95).

We calculated several fit indices and found that the simplified four-factor model was an appropriate fit across all three samples. (See Table 4.) While the full four-factor model, shows better fit according to these indices. We chose to keep the simplified factor due to the full four factor model including items that cross loaded. The factors that emerged were:

1. Deep connection to nature - Deeply seeing ones identity as part of nature
   (Note: in the SONA sample, this factor had a lot of “identity” items as well. These “identity” either failed to load on another factor or loaded on the “experience” factor in the Prolific samples.)

2. Experiential connection to nature - Spending time in nature and enjoying spending time in nature

3. Emotional connection to nature - Emotional desire to connect with nature and protect it

4. Presence within nature - Engaging mindfully and consciously with nature

Table 4. Fit Indices for 6, 5, 4, and 4-factor simplified models across three samples

Based on these results we selected 38 items to test in a confirmatory factor analysis. Table 5 displays the loadings of all items across the three samples and details the decisions made to keep or remove those items for the confirmatory factor analysis in the next study. Decisions were made to select items that loaded sufficiently on the four factors across the three samples. Items were retained if:

• They loaded strongly on the identified factor across all three samples

• They loaded strongly on the identified factor across two of the three samples

Table 5. Decisions for item selection for the confirmatory factor analysis

Study 2

Sample

394 undergraduate students participated for course credit in May 2023. Thirty-one participants did not complete the study and twenty-two people failed to pass attention checks so their data was not analysed. The final analysis consisted of 341 participants with ages ranging from 18 - 42 (M = 20.88, SD = 2.7). The same was primarily made up of females (68.3%) and people of Asian ethnicity (42.1%). See the table below for full descriptive statistics. Political Ideology was scored on a 7-point Likert scale with 1 = Most Conservative and 7 = Most Liberal. The mean political ideology was 5.04(1.17), showing that this sample skewed more liberal. This is to be expected with our student sample.

Convergent Validity Measures

Two pre-existing CTN scales were used to identify convergent validity. The Connectedenss to Nature Scale (CNS), (Mayer and Frantz 2004) and the Environmental Identity Scale revised (EID-R), (Clayton et al. 2021). The CNS is a 14-item scale designed to measure the cognitive beliefs about how interneconnected one’s self is to nature. It is measured on a Likert scale ranging from 1(strongly disagree) to 5(strongly agree). The EID-R is a 14-item scale designed to measure how interneconnected one’s self is to nature. It is measured on a Likert scale ranging from 1(not at all true of me) to 7(completely true of me).

Predictive Validity Measures

Pro-environmental behavior

Participants completed the Recurring Pro-environmental Behavior scale (Brick, Sherman, and Kim 2017) which asks people to report how often they engage in various pro-environmental behaviors. This scale includes 20 items measured on a Likert scale ranging from 1(never) to 5(always).

Primal Beliefs

Three subscales of the primals beliefs inventory (Clifton et al. 2019) were used to identify worldviews held by participants. The About Me subscale consists of five questions asking how egocentric ones beliefs are (e.g., Whatever is happening around me often feels related to me or something I’ve done). The Hierarchical subscale consists of five questions exploring beliefs about the hierarchical nature of the world (e.g., Things are rarely equal. Most plants and animals, even people, are better or worse than one another.). The Interconnected subscale consists of four items asking how connected or atomistic one believes the world to be (e.g., Most things are basically unconnected and independent from each other [reversed]). All items are measured on a Likert scale ranging from 1(strongly disagree) to 6(strongly agree).

Covariate measures

Locus of Control

Participants completed a brief version of the Locus of Control Scale (Sapp and Harrod 1993) which includes 9 items revealing how much people perceived they have internal control over events (e.g., To a great extent, my life is controlled by accidental happenings). Measured on a Likert scale from 1(strongly disagree) to 7(strongly agree).

Demographics

They also answered several demographics questions (age, gender, political ideology, and social economic status).

Confirmatory Factor Analysis

Items Removed

Several assumptions need to be met before running a confirmatory factor analysis. We identified that no item had a high inter-correlation > .75. Next, we checked the skewedness and kurtosis of each item to be included in the CFA. The closer to zero for both skewedness and kurtosis, the closer to a normal distribution. Any items whose distributions exhibited skewness and/or kurtosis that exceed general guidelines were considered non-normal removed from the analysis (Hair et al., 2017, p. 61).

For skewedness, using the skewness function from the Moments package in R (Komsta and Novomestky 2022), skewness was calculated for each item. Any item where the number is greater than +1 (right skewed) or lower than -1 (left skewed), this is an indication of a substantially skewed distribution. No items were heavily skewed in this dataset.

For Kurtosis, using the kurtosis function from the Moments package in R, kurtosis was calculated for each item. Any item were the number is greater than +3, the distribution is too peaked. Likewise, a kurtosis of less than -3 indicates a distribution that is too flat. Two items showed high kurtosis: “I have great respect for being alive on this earth and part of the universe” and “Listening to the sounds of nature makes me relax” and were removed from the CFA. Three items were identified as having high kurtosis See Table 6. These items were removed from the CFA

Table 6. High Kurtosis Items

## [1] "No items had high or low kurtosis"

A Mardia Test of multivariate normality from the MVN package in R (Korkmaz et al., 2021) indicated that this data is not normal, suggesting that a rotation-method that is robust to non-normal data should be used (Mardia Skewedness = 7681.36, p < .001 and Mardia Kurtosis = 28.32, p < .001).

Model Fitting

Four factor hierachical Model

Testing a structure which includes an overall Connection to Nature factor and four dimensions:

1. Deep: Deeply seeing the self as part of nature
2. Experiential: Spending more time in nature and enjoying spending time in nature
3. Emotional: Emotional desire to connect with nature and protect it
4. Presence: Engaging mindfully and consciously with nature

Items load onto their dimensions well, with the exception of 1 item (“Thinking of someone carving their initials into a tree makes me cringe”) which loaded onto the Emotion dimension at 0.33.

All dimensions loaded onto the single CTN dimension, but these loaded quite highly (especially Deep and Emotion) which suggests these might be driving the overall CTN score.

Four Factor

Testing a structure which only includes the four dimensions:

1. Deep: Deeply seeing the self as part of nature
2. Experiential: Spending more time in nature and enjoying spending time in nature
3. Emotional: Emotional desire to connect with nature and protect it
4. Presence: Engaging mindfully and consciously with nature

Items load onto their dimensions similarly well to the hierarchical model. Again, 1 item loads weakly: (“Thinking of someone carving their initials into a tree makes me cringe”) which loaded onto the Emotion dimension at 0.33.

Dimensions are intercorrelated reasonable well - not too high as to have multi-collinearity (none above .80)

Single Factor

Testing a structure which includes only a single overall Connection to Nature factor

Loadings are slightly weaker onto a single overall CTN factor. Again, 1 item loads weakly: (“Thinking of someone carving their initials into a tree makes me cringe”) which loaded onto the Emotion dimension at 0.31.

Item Selection Based on Model Fit

Both four-factor models (flat and hierarchical) perform better than the single-factor model. There is no difference between the hierarchical and the flat models.

## 
## Scaled Chi-Squared Difference Test (method = "satorra.bentler.2001")
## 
## lavaan NOTE:
##     The "Chisq" column contains standard test statistics, not the
##     robust test that should be reported per model. A robust difference
##     test is a function of two standard (not robust) statistics.
##  
##                   Df   AIC   BIC   Chisq Chisq diff Df diff Pr(>Chisq)    
## fit_model_hier_4 320 32790 33012  756.71                                  
## fit_model_flat   324 33332 33539 1306.45     388.02       4  < 2.2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

We continued with our validation using the Hierarchical four factor model.

Validation

Convergent Validity

Two pre-existing CTN scales were used to identify convergent validity. The Connectedenss to Nature Scale (CNS, Mayer & Frantz, 2004) and the Environmental Identity Scale revised (EID-R, Clayton et al., 2021). The CNS is a 14-item scale designed to measure the cognitive beliefs about how interneconnect one’s self is to nature. It is measured on a Likert scale ranging from 1(strongly disagree) to 5(strongly agree). The EID-R is a 14-item scale designed to measure how interneconnect one’s self is to nature. It is measured on a Likert scale ranging from 1(not at all true of me) to 7(completely true of me).

We assessed the convergent validity of the new DEEP CTN scale, that is how closely our new measure is related to other established measures of the same construct. We tested this by looking at the zero order correlation between our new measure and its subscales and two established measures: The Connectedness to Nature Scale (CNS - Mayer & Franz, 2004) and the Environmental Identity Scale revised (EID-R - Clayton et al., 2021). A zero-order correlation found moderate-strong correlation between existing CTN scales and the new subscales of the DEEP CTN scale. See figure below:

Predictive Validity

Two measures of predictive validity were used: Pro-environmental behavior & three subscales of the primal beliefs invetory - the world is interconnected, the world is heirarchical, and the world is about me.

Participants completed the Recurring Pro-environmental Behavior scale (Brick et al., 2017) which asks people to report how often they engage in various pro-environmental behaviors. This scale includes 20 items measured on a Likert scale ranging from 1(never) to 5(always).

Three subscales of the primals beliefs inventory (Clifton et al., 2019) were used to identify worldviews held by participants. The About Me subscale consists of five questions asking how egocentric ones beliefs are (e.g., Whatever is happening around me often feels related to me or something I’ve done). The Hierarchical subscale consists of five questions exploring beliefs about the hierarchical nature of the world (e.g., Things are rarely equal. Most plants and animals, even people, are better or worse than one another.). The Interconnected subscale consists of four items asking how connected or atomistic one believes the world to be (e.g., Most things are basically unconnected and independent from each other [reversed]). All items are measured on a Likert scale ranging from 1(strongly disagree) to 6(strongly agree).

We assessed the predictive validity of the new DEEP CTN scale (i.e., how well our measure can predict participant scores on two theoretically related constructs). The “gold standard” for predictive validity of CTN measures is how well it predicts pro-environmental behavior (PEB) (Clayton, 2003; Clayton et al., 2021; Mayer & Frantz, 2004; Tam, 2013). A zero-order correlation found moderate correlations between PEB and the new subscales of the DEEP CTN scale. See figure below:

A linear regression with only the subscales of CTN predicting PEB showed that Deep, Emotional, and Presence were adequate to predcict PEB, but not Experiential.

Total variance explained by the model = 0.28

We also suspected that DEEP CTN and its subscales would predict three types of primal world beliefs (Clifton et al., 2019): That the world is interconnected, that the world is not hierarchical, and that the world is not about me (i.e., less egocentric worldviews). A zero-order correlation found weak - moderate correlations between interconnected beliefs and the new subscales of the DEEP CTN scale. We also found weak correlations between egocentric beliefs and the Deep, Presence, and Overall CTN. Only Deep CTN significantly predicted interconnected primal worldviews. See table below:

Total variance explained by the model = 0.18

Incremental Validity

We explored if the DEEP CTN scale and its facets would perform above and beyond existing CTN scales (i.e., CNS and EID-R) when predicting PEB. A linear regression model was run, which included the dimensions of the DEEP CTN scale and the existing CTN measures (CNS and EIDR) were entered simultaneously to predict PEB. Note that because overall CTN was highly intercorrelated with Deep and Emotional, it was left out and only the four dimensions were included.

We found that the Emotional and Presence subscales remained significant predictors when existing measures of CTN were entered into the model. The Deep subscale was no longer a significant predictor. Both our subscales were of equal predictive value compared to the existing measures.

Total variance explained by the model = 0.31

Covariates

Participants completed a brief version of the Locus of Control Scale (Sapp & Harrod, 1993) which includes 9 items revealing how much people perceived they have internal control over events (e.g., To a great extent, my life is controlled by accidental happenings). Measured on a Likert scale from 1(strongly disagree) to 7(strongly agree). Participants also answered several demographics questions (age, gender, political ideology, and social economic status).

We also tested the robustness of our new DEEP CTN scale, by testing if the DEEP CTN scale and its facets are significant predictors of PEB after controlling for known covariates of these outcomes: political ideology, gender, and socio-economic status (SES). We fist inspected a zero order correlation to identify which covariates should be included. Including any variable that: strongly correlated with PEB AND weakly correlated with the Deep CTN Scale may reduce variance in the dependent variable and therefore enhance the likelihood of seeing significant effects of the DEEP CTN scale. Further, if an additional variable is… strongly correlated with BOTH PEB AND the Deep CTN Scale, its inclusion in the model will address potential mediating effects regarding the relationship between the Deep CTN Scale and a DV of interest. In either case this would suggest it’s important to include as a covariate.

This resulted in including the following covariates: Politics (r = .19** with PEB & r = .18** to .23** with DEEP CTN dimensions) Primal Beliefs - The world is interconnect (r = .21** with PEB & r = .22** to .38** with DEEP CTN dimensions)

A multiple linear regression model where dimensions of the DEEP CTN scale and existing CTN measures (CNS and EIDR) and the covariates (politics & primal belief interconnectedness) were entered simultaneously to predict PEB is reported in the table below.

By adding the covariates, both Emotional and Prescence subscales remained significant predictors of PEB. Adding covariates to the model did not improve the fit of the model above the variables in the incremental validity model (F(2) = 1.731, p = 0.179).

Total variance explained by the model = 0.32

Discussion

The findings from three exploratory factor analysis and one confirmatory factor analysis reveal that the construct of connection to nature can be described by four distinct factors:

1. Deep Identity: Deeply seeing the self as part of nature
2. Experiential Connection to nature: Spending and enjoying time in nature
3. Emotional Connection to nature: Desire to connect with and protect nature
4. Presence within nature: Being deliberately mindful and present within nature

These factors were shown to both relate to and extend beyond two existing measures of connection to nature. The Connectedness to Nature scale (Mayer and Frantz 2004) was most strongly correlated with the deep identity factor. This aligns with the CNS as a cognitive measure of how much ones identity is interrelated with nature. The other three factors also correlated with the CNS, however they correlated only moderately. The revised Environmental Identity scale (Clayton et al. 2021) was most strongly corrleated with both the deep identity and experiential connection factors. This also aligns with past research that has shown the EID-r to be described as a multidimensional scale that describes ones identity and experiences within nature. Both emotional connection and presence within nature also correlated moderately with the EID-r.

Our study also provided evidence for the predictive validity of the scale. We found differential predictive value across the four factors in predicting pro-environmental behavior and primal beliefs that the world is interconnected. All but the experiential connection factor uniquely predict pro-environmental behavior, with deep identity being the strongest predictor. However, when including previous measures of CTN, only emotional connection and presence within nature remained significant predictors of PEB (along both the existing measures). These two factors remained significant even after controlling for participants political ideology and their worldviews. Deep identity was the only factor that uniquely predicted primal beliefs that the world is interconnected suggesting that this factor best represents the more cognitive components of connection to nature.

Limitations

The confirmatory factor analysis and validation of the DEEP CTN Scale was conducted on a student sample which restricts the generalizability of these results. It is evident from the three EFA samples that there are differences in how connection to nature is represented in student populations compared to a more general population sample. For example, many items that are classified as relating to ones identity as connected to nature loaded on the Deep Identity factor within the student sample, while these items failed to load on any factor or loaded on the Experiential connection factor in a general sample. Similarly, many of the spiritual items that were included in the Deep Identity factor loaded strongly in the general population samples but failed to load on any factor in the student sample. One explanation for this is that connection to nature appears to develop over the lifespan. It tends to decline during adolescence and not strengthen again until mid-to-late adulthood (Hughes et al. 2019). It may be that the Deep identity items that showed discrepancies across samples with varying ages picked up on this developmental trajectory. Because of this variation across age-groups we are currently collecting data from a general population to conduct a pre-registered CFA and validation that attempts to replicate the findings in this study.

Future Directions & Conclusion

Based on these preliminary data we conclude that connection to nature is indeed a multi-dimensional construct and these constructs differentially predict important correlates such as pro-environmental behavior and worldviews. Most importantly, our results reveal that experiential connection to nature may be the least important when it comes to promoting PEB. Given that many interventions to promote CTN are focused on getting people “out into nature” it is concerning that these mere exposure may not be the driving factor in promoting PEB. Our results suggests that interventions should instead focus on building an emotional connection with nature and/or encouraging mindful engagement within nature.

References