The Role of Interaction Patterns with Hybrid Electric Vehicle Eco-Features for Drivers’ Eco-Driving Performance

Abstract

Objective:

The objective of the present research was to understand drivers’ interaction patterns with hybrid electric vehicles’ (HEV) eco-features (electric propulsion, regenerative braking, neutral mode) and their relationship to fuel efficiency and driver characteristics (technical system knowledge, eco-driving motivation).

Background:

Eco-driving (driving behaviors performed to achieve higher fuel efficiency) has the potential to reduce CO₂ emissions caused by road vehicles. Eco-driving in HEVs is particularly challenging due to the systems’ dynamic energy flows. As a result, drivers are likely to show diverse eco-driving behaviors, depending on factors like knowledge and motivation. The eco-features represent an interface for the control of the systems’ energy flows.

Method:

A sample of 121 HEV drivers who had constantly logged their fuel consumption prior to the study participated in an online questionnaire.

Results:

Drivers’ interaction patterns with the eco-features were related to fuel efficiency. A common factor was identified in an exploratory factor analysis, characterizing the intensity of actively dealing with electric energy, which was also related to fuel efficiency. Driver characteristics were not related to this factor, yet they were significant predictors of fuel efficiency.

Conclusion:

From the perspective of user–energy interaction, the relationship of the aggregated factor to fuel efficiency emphasizes the central role of drivers’ perception of and interaction with energy conversions in determining HEV eco-driving success.

Application:

To arrive at an in-depth understanding of drivers’ eco-driving behaviors that can guide interface design, authors of future research should be concerned with the psychological processes that underlie drivers’ interaction patterns with eco-features.

Keywords

driver behavior user–energy interaction electric propulsion regenerative braking neutral mode

Introduction

Road transport is one of the major contributors to CO₂ emissions caused by energy infrastructure (Davis, Caldeira, & Matthews, 2010). Consequently, energy-efficient vehicles, such as hybrid electric vehicles (HEVs), that can reduce road transport CO₂ emissions (Bitsche & Gutmann, 2004) have become increasingly widespread (see, e.g., Dijk, Orsato, & Kemp, 2013). Yet actual energy efficiency is still associated with the driver: Vehicle maintenance and route selection (strategic or tactical eco-driving decisions) constitute significant influences on drivers’ ecological footprints (Sivak & Schoettle, 2012). Likewise, operational eco-driving (Sivak & Schoettle, 2012), comprising all driving behaviors performed to achieve higher energy efficiency (see, e.g., Barkenbus, 2010), vastly influences the real-world energy efficiency of road vehicles (e.g., af Wåhlberg, 2007; Bingham, Walsh, & Carroll, 2012), with the potential for long-term fuel savings of on average 10% (cf. Barkenbus, 2010). Therefore, applying a human factors and ergonomics perspective for understanding patterns in energy-efficient driving behavior is fundamental to improving the sustainability of road transport.

Which eco-driving behavior yields optimal energy efficiency particularly depends on drivetrain characteristics, for example, the availability of regenerative braking (Cocron et al., 2011; Kuriyama, Yamamoto, & Miyatake, 2010). When compared to conventional combustion vehicles, HEVs contain several additional drivetrain features, such as electric propulsion, regenerative braking, and neutral mode (i.e., a driving mode in which the HEV is coasting without energy from the motor/engine and without regenerative braking). These key drivetrain features (hereafter labeled eco-features) enable the driver to control the system’s energy flows and thus constitute an HEV’s specific user-energy interface. Understanding how driver interaction with eco-features is related to fuel efficiency therefore constitutes a key contribution to the field of user–energy interaction.

Due to the comparatively complex energy flows in the HEV (McIlroy, Stanton, Harvey, & Robertson, 2013; e.g., bidirectional energy flow), drivers who generally use different sets of ecodriving strategies can have similar energy efficiencies (Franke, Arend, McIlroy, & Stanton, 2016a). However, specific eco-driving strategies that include the usage of eco-features (i.e., interaction patterns with eco-features) tend to be associated with increased or decreased fuel efficiency (Franke et al., 2016a). An important question in this respect is what determines the drivers’ selection of specific interaction patterns. Among other factors focusing on the role of driver characteristics in eco-driving, technical system knowledge and eco-driving motivation (see e.g., Franke et al., 2016a; Franke, Arend, McIlroy, & Stanton, 2016b; Stillwater & Kurani, 2013) have been examined, in particular for HEV drivers. Both tend to be related to the selection of specific eco-driving strategies in addition to fuel efficiency (Franke et al., 2016a). Consequently, investigating the role of technical system knowledge and eco-driving motivation is considered a relevant topic for understanding driver interaction with eco-features.

The objective of the present research was to examine (a) which interaction patterns with eco-features are related to energy efficiency (i.e., eco-driving success) and how technical system knowledge and eco-driving motivation are related to (b) interindividual differences in those interaction patterns and (c) eco-driving success.

Background

Driver interaction patterns with eco-features: A control-theoretic approach

In general, for energy-efficient driving with HEVs, drivers have to constantly adapt their behavior, thereby utilizing the eco-features in a certain way. Control-theoretic models constitute a fruitful framework for modeling such continuous interactions (Fuller, 2011; Summala, 2007) and are therefore particularly suited to model drivers’ interaction patterns with eco-features, embedded in the general (eco-)driving task (see Franke et al., 2016a). Control-theoretic models propose that drivers constantly monitor system and environmental variables and control/adapt their driving behavior by comparing an actual state with a target state (e.g., Fuller, 2005, 2011). Both constitute levels of task difficulty (e.g., a car driving with a certain speed [actual state] is slowed down [controlled] until the preferred speed level [target state] is reached). Transferred to the scenario of a driver interaction with an eco-feature, the driver will control the system in order to achieve the eco-feature state perceived as energy efficient (e.g., in HEV driving with the combustion engine [actual state], the driver slowly reduces pressure on the gas pedal [control] until the combustion engine is shut off and solely the electric propulsion moves the wheels [target state]).

In control-theoretic models, the individual target state as well as general differences in (eco-)driving behavior are assumed to be dependent on the driver (see, e.g., Davidse, Hagenzieker, van Wolffelaar, & Brouwer, 2009; Fuller, 2005, 2011). There will be differences between drivers regarding factors such as their motivation (e.g., the importance of the primary goal of driving safely relative to the secondary goal of fuel efficiency, or varying goal priorities due to traffic/roadway demands; e.g., Dogan, Steg, & Delhomme, 2011; Fuller, 2007) or knowledge (e.g., Franke et al., 2016a). Consequently, both are hypothesized to relate to drivers’ interaction patterns with HEV eco-features.

Energy flows related to HEVs’ eco-features

By controlling the vehicle, drivers influence the energy flows associated with the eco-features. First, electric propulsion (i.e., the driving state when the wheels are moved by the electric motor) allows the conversion of electric into kinetic energy via the wheels. Second, with regenerative braking (i.e., the driving state of energy recovery while the car is decelerating), kinetic energy of the wheels can be converted back into electric energy. Third, the neutral mode (i.e., the driving state of coasting created by the kinetic energy already possessed by the wheels without drivetrain propulsion or employment of regenerative braking) is, compared with other eco-features, associated with maximum utilization of the kinetic energy built up, thus avoiding conversion. Using this perspective of a relationship between driver behavior and energy flows may also contribute to a better understanding of general patterns of user–energy interaction.

Present Research

The objective of the present research was to advance the understanding of user–energy interaction by examining (a) which interaction patterns with eco-features are related to energy efficiency (i.e., eco-driving success) and how technical system knowledge and eco-driving motivation are related to (b) interindividual differences in those interaction patterns and (c) eco-driving success. To study drivers with day-to-day experience of HEVs’ eco-features, a sample of 121 HEV drivers who had constantly logged their fuel efficiency was recruited.

Research Questions and Hypotheses

Table 1 provides an overview of the research questions (Q) and hypotheses. When the direction of effects could be deduced from literature or theory, hypotheses were directional; when not, hypotheses were nondirectional. Directional hypotheses regarding Q2, Q3, and Q4 are based on results from Franke et al. (2016a).

Table 1:

Research Questions and Hypotheses in the Present Study

Research Question	Hypothesis
1: How do drivers perceive the energy efficiency of HEV eco-features?	1: Perceived energy efficiency of each eco-feature is related to technical system knowledge.
2: How do interaction pattern dimensions relate to fuel efficiency?	2a: Frequency of using the electric propulsion is related to lower fuel efficiency.
	2b: Frequency of regenerative braking is related to fuel efficiency.
	2c: Acceleration without versus with electric propulsion is related to lower fuel efficiency.
	2d: Deceleration with neutral mode versus regenerative braking is related to fuel efficiency.
3: How do drivers’ technical system knowledge and eco-driving motivation relate to interaction patterns dimensions?	3a: Technical system knowledge is related to the frequency of using regenerative braking and to deceleration with neutral mode versus regenerative braking.
	3b: Higher technical system knowledge is related to acceleration without the electric propulsion.
	3c: Eco-driving motivation is related to frequency of using regenerative braking and to deceleration with neutral mode versus regenerative braking.
	3d: Higher eco-driving motivation is related to higher frequency of using the electric propulsion and acceleration with the electric propulsion.
4: How do drivers’ technical system knowledge and eco-driving motivation relate to energy efficiency?	4a: Higher technical system knowledge is related to higher fuel efficiency.
	4b: Higher eco-driving motivation is related to higher fuel efficiency.
	4c: The relationships of technical system knowledge and eco-driving motivation to fuel efficiency weaken when the relationships of interaction patterns to fuel efficiency are taken into account.

Note. Directional hypotheses are Hypotheses 2a, 2c, 2d, 3b, 3d, 4a, 4b, and 4c. Nondirectional hypotheses are Hypotheses 1, 2b, 2d, 3a, and 3c.

Based on the control-theoretic framework introduced in the section Background, drivers control each eco-feature to arrive at the target state they perceive efficient. Q1 is therefore concerned with how energy efficient HEV drivers perceive the eco-features.

Typical interaction patterns with HEV eco-features were deduced from an interview study documented in Franke et al. (2016a). These patterns were conceptualized as dimensions (e.g., each deceleration with eco-features can be classified between entirely with neutral mode and entirely with regenerative braking). To examine a broad range of interaction patterns, two dimensions representing general (i.e., a behavioral tendency to use an eco-feature) as well as two representing specific (i.e., a characteristic driving maneuver) interaction patterns were assessed. General interaction patterns referred to the frequency of using electric propulsion and regenerative braking, and specific ones related to acceleration without versus with utilization of electric propulsion and deceleration with neutral mode versus with regenerative braking.

In addition to these factors, the relationships of technical system knowledge and eco-driving motivation to interaction patterns with eco-features (Q3), as well as to fuel efficiency (Q4), were also examined. Hypothesis 4c concerns the role of technical system knowledge and eco-driving motivation in determining fuel efficiency. It is expected that their positive influence on fuel efficiency arises as a result of differences in eco-driving behavior (including differences in drivers’ interaction patterns with eco-features). This is the case when (a) both technical system knowledge and eco-driving motivation relate to interaction patterns and (b) their relationship to fuel efficiency weakens when the influence of interaction patterns on fuel efficiency is also taken into account.

Method

Participants

To assess hypotheses with 80% power of detecting bivariate correlations |r| ≥ .30 and R² ≥ .10 in the expected multiple linear regression models with up to six predictors (for Hypothesis 4c), a sample size of at least 130 HEV drivers was required (a priori power analyses calculated with G-Power; Faul, Erdfelder, Buchner, & Lang; 2009). To achieve this sample size, drivers of the Toyota Prius 2, Prius 3, and Prius c were recruited. To reach a sample of experienced drivers who constantly logged their fuel consumption data, recruitment focused on drivers from the online database spritmonitor.de. Only drivers who had logged their fuel data at least once within 6 months before the initial contact were recruited (most drivers, however, logged their fuel consumption regularly on a weekly basis). Drivers who met these criteria were contacted from October 2015 to December 2015 with an invitation to participate in an online questionnaire. The study was run from December 2015 to January 2016.

The final sample consisted of 121 HEV drivers (93% male; M_age = 48 years, SD_age = 11). Of the total number of participants, 20 had previously participated in an interview-study documented in Franke et al. (2016a). Drivers had a mean driven distance of 62,392 km (SD = 52,324; range = 5,600–250,000), or 38,769 miles (SD = 32,513; range = 3,480–155,343), with HEVs in general corresponding to a mean driving experience of 3.3 years (SD = 2.3; range = 0.4–16). At the time of data collection, 42% of the drivers drove a Prius 3, 30% a Prius c, and 28% a Prius 2 as their primary HEV. With this current HEV, drivers had a mean driven distance of 57,482 km (SD = 48,791; range = 4,000–230,000), or 35,718 miles (SD = 30,317; range = 2,485–142,915).

Scales and Measures

The same rating scale was used for all items and all scales described next (6-point Likert scale ranging from completely disagree = 1 to completely agree = 6). Scales, instructions, and items are included in the appendix.

Perceived efficiency of eco-features

Efficiency evaluation of each eco-feature was assessed based upon two facets: items representing the perceived efficiency of the usage of each eco-feature and items focusing on the perceived efficiency of the energy dynamics incurred in utilizing the respective eco-feature. Items were deduced from driver statements collected via a previous interview-study (Franke et al., 2016a). Scale reliabilities (Cronbach’s alpha) were sufficient to excellent (electric propulsion, α = .83; regenerative braking, α = .79; neutral mode, α = .91).

Usage dimensions of eco-features

As stated in the section Present Research, interaction patterns can be conceptualized as eco-driving strategy dimensions. In the questionnaire, participants rated a number of eco-driving strategy dimensions, all assessed with two-item scales. As general usage dimensions, the frequency of using electric propulsion (usage of electric propulsion, α = .75) and the frequency of using regenerative braking (usage of regenerative braking, α = .91) were assessed. As specific usage dimensions, acceleration without versus with utilization of the electric propulsion (acceleration, α = .75) and, similarly, deceleration with neutral mode versus with regenerative braking (deceleration, α = .84) were assessed.

Knowledge and motivation

Technical system knowledge (knowledge; three items, α = .84) and eco-driving motivation (motivation; two items, α = .81) were both assessed using scales developed in previous research (Franke et al., 2016a, 2016b). The knowledge scale (adapted from Franke, Rauh, Günther, Trantow, & Krems, 2015) was developed to represent technical knowledge concerning the HEV and its system dynamics. The motivation scale was designed to represent behavioral intention (i.e., intention to perform energy efficient driving behavior; e.g., Ajzen, 1991; Lauper, Moser, Fischer, Matthies, & Kaufmann-Hayoz, 2015).

Fuel efficiency

In order to assess the fuel consumption of the drivers as accurately as possible, four indicators were taken into account (see appendix): Drivers estimated their average fuel consumption between April and September 2015 (“estimated”). Their fuel consumption for the same period was also calculated from logger data (“logged”): At the end of the questionnaire, drivers were asked to provide a data file with their logged fuel consumption (users of the spritmonitor.de website can download their individual fuel logs). Logger data were available from 100 participants (18 drivers decided not to provide logger data; three provided data files that did not contain fueling events between April and September 2015). The strong relationship (r = .86) between estimated and logged fuel consumption indicated that participants’ estimations were sufficiently accurate.

To arrive at a reliable indicator of fuel efficiency, the following steps were performed (parallel to the rationale of Franke et al., 2016a). To account for potential differences in the trip profiles of the drivers (e.g., some participants may drive primarily on urban vs. rural routes), two indicators were used: Drivers estimated their fuel consumption for an urban and a rural route —both with standardized conditions (e.g., weather, other traffic). Furthermore, in order to remove mean fuel consumption differences between HEV models, the indicators were standardized based on the respective distribution parameters of the whole fleet (taken from spritmonitor.de). Results were inverted so that higher values corresponded to higher fuel efficiency. This procedure was performed for both indicators, resulting in the fuel efficiency indicators (FEI): FEI.urban and FEI.rural. Since both correlated strongly (r = .64), they were aggregated to one factor (FEI.all; α = .78; n = 120 due to missing data for one participant), which was used for all following analyses.

Results

Statistical analyses were based on descriptive statistics, correlation coefficients, and linear regression models. Effect sizes were interpreted according to Cohen (1992). Outliers more than three standard deviations from the scale mean (i.e., |z| > 3) were identified for the variables evaluation of regenerative braking (z = −3.25, thus n = 120), motivation (z = −3.12), and FEI.all (z = −3.11) and were subsequently excluded. P values were two tailed for nondirectional and one tailed for directional hypotheses (Kimmel, 1957). The significance level used was α = .05.

Efficiency Evaluation of HEV Eco-Features’ Energy Efficiency (Q1)

In reference to Q1, Figure 1 depicts key parameters of the three efficiency evaluation distributions (median, quartiles, ranges, and frequency distributions). Regenerative braking (M = 3.65, SD = 0.66) and neutral mode (M = 3.68, SD = 1.55) were, on average, perceived as slightly more energy efficient than electric propulsion (M = 3.33, SD = 1.09). However, all mean evaluations lay between 3 and 4 (i.e., slight disagreement and slight agreement with a positive evaluation). Moreover, the efficiency evaluation of electric propulsion and neutral mode showed fairly strong interindividual variance.

Figure 1.

Box plots and histograms depicting the efficiency evaluations of the eco-features.

A correlational analysis was performed to further assess Q1 and Hypothesis 1 (see Table 2). Assumptions were tested according to Field (2009). All variables except three were non-normally distributed. Spearman rank correlation coefficients (ρ) were therefore also computed. Regarding the relationships of knowledge to efficiency evaluations, none were significant and effect sizes were, at the most, small (Hypothesis 1 rejected). All correlations between different efficiency evaluations were moderate to strong and significant (p < .001). Efficiency evaluation of electric propulsion was positively related to regenerative braking and negatively to neutral mode. The relationship between the efficiency evaluation of regenerative braking and neutral mode was negative.

Table 2:

Correlation Matrix of Efficiency Evaluations, Knowledge, Motivation, and FEI.all

Variable	1	2	3	4	5	6
1. Efficiency evaluation of electric propulsion	—	.48***	−.41***	−.12	.17	−.18*
2. Efficiency evaluation of regenerative braking	.43***	—	−.42***	−.02	.14	−.18
3. Efficiency evaluation of neutral mode	−.40***	−.39***	—	.14	.06	.15
4. Knowledge	−.12	.02	.13	—	−.12	.23*
5. Motivation	.13	.09	.06	−.12	—	.19*
6. FEI.all	−.21*	−.12	.15	.17	.22*	—

Note. Coefficients above the diagonal are r; coefficients below the diagonal are ρ. Correlations of two normal distributed variables are interpreted based on r, otherwise ρ. Here, efficiency evaluation of electric propulsion and FEI.all were normally distributed. FEI.all = fuel efficiency indicator.

p < .05. **p < .01. ***p < .001 (all p values were two tailed).

When considered together (Q1), these results imply that the majority of drivers perceived the energy efficiency of regenerative braking as rather mediocre, whereas electric propulsion and neutral mode showed large variance, thereby implying that despite day-to-day interaction, drivers’ perception of the energy efficiency of eco-features was considerably diverse. Yet, which eco-feature was perceived efficient to what degree was highly interrelated. This finding can be considered an indication for a common underlying pattern: Perceiving electric propulsion as efficient involves perceiving regenerative braking as more efficient and neutral mode as less efficient and vice versa.

Relation of Usage Dimensions of Eco-Features to Fuel Efficiency (Q2)

To assess Q2, the four usage dimensions of eco-features were correlated with FEI.all (see Table 3). Higher usage of electric propulsion and regenerative braking were both related to lower fuel efficiency (small albeit significant effect; Hypotheses 2a and 2b supported). Accelerating with higher utilization of electric propulsion was moderately to strongly and significantly related to lower fuel efficiency (Hypothesis 2c supported). Finally, decelerating with regenerative braking opposed to with neutral mode when possible was also slightly related to lower fuel efficiency (Hypothesis 2d supported).

Table 3:

Correlation Matrix of Usage Dimensions of Eco-Features and FEI.all

Variable	1	2	3	4	5	6
1. Usage of electric propulsion	—	.50***	.42***	.39***	−.22*	.70***
2. Usage of regenerative braking	.52***	—	.43***	.79***	−.19*	.86***
3. Acceleration	.44***	.42***	—	.39***	−.41***	.69***
4. Deceleration	.37***	.73***	.33***	—	−.24**	.83***
5. FEI.all	−.24**	−.18*	−.40***	−.18*	—	−.31**
6. Intensity of actively dealing with electric energy	.74***	.83***	.72***	.74***	−.29**	—

Note. Coefficients above the diagonal are r; coefficients below the diagonal are ρ. Correlations of two normal distributed variables are interpreted based on r, otherwise ρ. Here, acceleration and FEI.all were normally distributed. FEI.all = fuel efficiency indicator.

p < .05. **p < .01. ***p < .001 (all p values were two tailed).

All interrelations between usage dimensions were positive, moderate to strong, and significant (p < .001). All variables thus share common variance and could probably be aggregated to one factor. To assess this possibility, we conducted a principal factor analysis with the usage dimensions. A one-factor solution was proposed by both Kaiser and scree plot criterion. This factor (eigenvalue = 2.07) explained 52% of the total variance. All factor loadings were higher than .52 and can therefore be considered significant (according to Stevens, 2012). Consequently, based on the one-factor solution, a mean score (reliability, α = .79) of the four interaction dimensions was computed, termed intensity of actively dealing with electric energy. Drivers scoring higher on this factor use the electric propulsion and regenerative braking more often but neutral mode less frequently, whereas drivers having smaller values prefer applying the neutral mode more often and electric propulsion as well as regenerative braking less frequently.

To ultimately answer Q2, a simple linear regression with the intensity of actively dealing with electric energy as predictor, and FEI.all as criterion, was performed (combining Hypotheses 2a through 2d, Hypothesis 2all resulted—the aggregated factor is related to lower fuel efficiency). Again, assumptions for (simple and multiple) regression analyses were tested according to Field (2009). Results were computed without and with exclusion of outliers/influential cases. Exclusion yielded similar parameter estimates but lower p values (due to exclusion of residual outliers), and consequently only the conservative solution (without exclusion) is reported in the present research. First, one univariate outlier on the factor was excluded (z₃₆ = −3.17), thus n = 120 for all following analyses. Intensively dealing with electric energy was negatively, moderately, and significantly related to FEI.all (β = −.31, p = .001, N = 118) and accounted for 9% (R²_adj, p = .001) of its variance, thereby supporting Hypothesis 2all.

Relation of Knowledge and Motivation to Interaction Dimensions (Q3)

To examine Q3, correlations and p values were calculated. Knowledge was not related (all effect sizes were less than small; Hypotheses 3a and 3b rejected) to usage of electric propulsion (ρ = −.04, p = .325), usage of regenerative braking (ρ = .00, p = .976), acceleration (ρ = .06, p = .771), and deceleration (ρ = −.08, p = .362). Motivation was positively associated with acceleration (ρ = .16, p = .043) and nonsignificantly but positively related to usage of electric propulsion (ρ = .13, p = .073), usage of regenerative braking (ρ = .11, p = .232), and deceleration (ρ = .11, p = .225). All effect sizes were small. These results partially support Hypothesis 3c; Hypothesis 3d was rejected.

Furthermore, a multiple regression analysis was performed with knowledge and motivation as predictors of the aggregated factor. Transferring hypotheses from Q3, knowledge was expected to relate negatively, and motivation to relate positively, to the intensity of actively dealing with electric energy (Hypothesis 3all). The regression model was nonsignificant (R²_adj = .00, p = .316, N = 117). The same applied to the effects of the predictors knowledge (β = −.05, p = .290) and motivation (β = .13, p = .089). Although directions were as hypothesized, the effects were only small, and thus Hypothesis 3all was rejected.

Relation of Knowledge and Motivation to Fuel Efficiency (Q4)

For assessing all hypotheses of Q4, a multiple stepwise regression analysis with knowledge and motivation (Step 1) and, additionally, the aggregated factor (Step 2) as predictors of FEI.all was performed. As depicted in Table 4, knowledge and motivation had small to moderate positive relationships to fuel efficiency (Hypotheses 4a and 4b supported). When incorporated into the model, the aggregated factor had a medium negative effect on FEI.all. The effects of knowledge and motivation on fuel efficiency did not change by taking into account this additional influence (Hypothesis 4c rejected).

Table 4:

Stepwise Regression With Knowledge, Motivation, and the Aggregated Factor (Intensity of Actively Dealing With Electric Energy) as Predictors of FEI.all

Step	Predictor	Statistic	Step 1	Step 2	R ² _adj	p	R ² _change	p
Step 1	Knowledge	B	.25	.24	.09	.002
		β	.25	.24
		p	.007	.008
Step 1	Motivation	B	.25	.29
		β	.24	.28
		p	.006	.001
Step 2	Aggregated factor	B		−.33	.19	< .001	.11	< .001
		β		−.33
		p		<.001

Note. N = 117. B is the regression slope of the respective predictor; p values refer to the effects in the row/column before.

To receive an indication of the practical significance of the effects on eco-driving, the decrease in fuel consumption, as indicated by regression slopes, was calculated by retransformation of the FEI.all indicator into l/100 km and gallons per 100 miles. For all three HEV types, a one-scale-point increase in knowledge was associated with a decrease of 0.1 l/100 km, or 0.04 gallons per 100 miles. One scale point more in motivation was related to reduced fuel consumption of 0.1 (Prius 2, Prius c) to 0.2 (Prius 3) l/100 km, or 0.04 to 0.09 gallons per 100 miles. Furthermore, fuel savings of 0.1 (Prius c) to 0.2 (Prius 2, Prius 3) l/100 km, again, 0.04 to 0.09 gallons per 100 miles, were associated with a one-scale-point change toward less intensively dealing with electric energy.

In particular, the aggregated factor represents large potential for reducing fuel consumption, because 88% of the drivers had values larger than the midpoint of the scale (with change toward the lower end being associated with lower fuel consumption). A considerably lower intensity in actively dealing with electric energy (e.g., three scale points) could thus result in fuel savings of roughly 0.4 to 0.5 l/100 km, or 0.17 to 0.21 gallons per 100 miles, which corresponds to M = 11% (SD = 2%; range = 8%–17%), for those drivers.

Discussion

Summary of Results

The objective of the present research was to advance understanding of the role of driver interaction with HEV eco-features in determining eco-driving success. On average, drivers perceived eco-features neither absolutely energy efficient nor inefficient. In fact, there were large interindividual differences in the perceived energy efficiency of electric propulsion and neutral mode, indicating that despite day-to-day experience, drivers do not develop a common perception of the energy efficiency of HEV eco-features. A driver’s level of technical system knowledge was not significantly related to the perceived energy efficiency of eco-features (Hypothesis 1 rejected). More intense utilization of electric propulsion (in general and for acceleration) and regenerative braking (in general and for deceleration, instead of neutral mode) were related to lower fuel efficiency (Hypotheses 2a, 2b, 2c, and 2d supported). Eco-feature strategy dimensions were aggregated to one factor representing the intensity of actively dealing with electric energy. This factor was moderately related to lower fuel efficiency (Hypothesis 2all supported). In addition, technical system knowledge was related neither to usage dimensions of eco-features nor to the aggregated factor (Hypotheses 3a and 3b rejected). Eco-driving motivation was not correlated to this factor but related to acceleration with utilization of the electric propulsion (Hypothesis 3c rejected, Hypothesis 3d partly supported). Finally, technical system knowledge and eco-driving motivation both had positive effects on fuel efficiency (Hypotheses 4a and 4b confirmed). These effects remained constant when the influence of the aggregated factor on fuel efficiency was taken into account (Hypothesis 4c rejected).

Due to the statistical analyses finally performed, we reached 92% (for a two-tailed test; respectively, 96% for a one-tailed test) power for bivariate correlation coefficients |r| ≥ .30 and 85% power of finding a significant regression model with three predictors and R² ≥ .10. Thus, the probability that expected effects were not identified here is relatively low (Faul, Erdfelder, Lang, & Buchner, 2007).

Theoretical Implications

The present results contribute to the understanding of user–energy interaction patterns: The high interrelations between perceived energy efficiency (Q1) of each eco-feature (i.e., high efficiency evaluation of electric propulsion and regenerative braking is paired to low efficiency evaluation of neutral mode and vice versa) illustrate the same tendency as the interrelations between usage dimensions of eco-features (Q2). Both patterns support the existence of the factor representing the intensity of actively dealing with electric energy. Applying the perspective of energy conversions (similar to Franke et al., 2016a), this factor may be interpreted as follows: Actively dealing with electric energy implies a high degree of energy conversions (active utilization [conversion of electric into kinetic energy] and regeneration [conversion of kinetic into electric energy]), whereas dealing less intensively with electric energy is associated with fewer energy conversions (kinetic energy has to be built up but is then conserved by usage of the neutral mode). These results can hence be seen as indicating something like an “energy conversion fallacy” (i.e., a considerable share of users neglect/underrate the losses incurred in energy conversions in estimating fuel efficiency and overuse eco-features related to a high degree of conversions). A tentative theorizing interpretation of this pattern following the prospect theory (Kahneman & Tversky, 1979, 1984; Tversky & Kahneman, 1992) could be that drivers might interact with the system in order to achieve higher energy efficiency based on one primary reference energy (here, either electric or kinetic energy). Further studies are necessary to examine these assumptions.

Results from Q2 indicate that eco-feature strategies relate to fuel efficiency and thus constitute driving behavior facets relevant for eco-driving success. This finding can therefore be regarded as support for the conceptualization of driver interaction with eco-features based on control-theoretic notions.

Furthermore, technical system knowledge and eco-driving motivation related to fuel efficiency (Q4), which supports the assumption that both have to be included as driver characteristics when conceptualizing eco-driving in HEVs based on a control-theoretic approach (see Franke et al., 2016a). Note that the effects found in the present sample were somewhat smaller than those found by Franke et al. (2016a) for a sample of HEV drivers with above-average fuel efficiency.

It should be noted, however, that knowledge did not yield significant relationships to perceived efficiency or usage dimensions of eco-features, and motivation was only partly related to the usage dimensions (all except one effect were small and nonsignificant; Q3). In addition to this, the effect of both on fuel efficiency remained constant when the effect of the aggregated factor on fuel efficiency was also taken into account. Whereas the direction of the small (although nonsignificant) effects found for eco-driving motivation in Q3 are in line with the present hypotheses, the assumption that knowledge relates to usage dimensions (and thus, the selection of eco-driving strategies related to eco-features; cf. Franke et al., 2016a) was not supported. Two possible interpretations for this pattern would be the following: First, the positive effect of knowledge on fuel efficiency could originate from eco-driving strategy aspects not related to general usage or nonusage of eco-features. For instance, former results indicate that speed choices (e.g., variable vs. constant speed) relate to both knowledge and fuel efficiency (Franke et al., 2016a). Second, due to the complexity that HEV eco-features introduce to the drivetrain (e.g., McIlroy et al., 2013), aspects of technical system knowledge could be directly reflected on the operational level of driving (Michon, 1985), in the specific implementation of eco-driving (similar to procedural knowledge; see, e.g., Anderson, 1982; McIlroy & Stanton, 2015). If so, hypothesized differences in eco-driving behavior that depend on a driver’s individual knowledge might be identified only by methods providing precise descriptions or observations of the concrete strategy implementations (e.g., assessed in interview, field, or driving simulator studies; Stanton et al., 2013; for an example, see Pampel, Jamson, Hibberd, & Barnard, 2015). However, further research is needed to reach a clear understanding of this pattern in our results.

In summary, it has to be pointed out that some findings in the present study were not in full accordance with findings in our previous study (Franke et al., 2016a). These differences could be due to sample characteristics, statistical factors (Schönbrodt & Perugini, 2013), or methodological factors. Further studies are needed for clarification. In addition, future research should address the influence of roadway and traffic demands on eco-driving behavior (e.g., driver’s strategy selection in high-density traffic situations) as well as the factors mediating the relationship between knowledge and fuel efficiency.

Practical Implications

The results from Q4 imply a possibility to gain approximately 11% fuel efficiency for many drivers through modifying interaction patterns toward a higher focus on conserving kinetic energy, and avoiding actively intensifying the conversion of kinetic into electric energy, and vice versa. From the perspective of green ergonomics (Hanson, 2013; Thatcher, 2013), modifying driver behavior via the design of specific system feedback is a particularly fruitful approach to enhance the sustainability of road transport (see, e.g., for eco-driving, Gilman, Keskinarkaus, Tamminen, Pirttikangas, & Röning, 2015; Jamson, Hibberd, & Merat, 2015). Instead of magnifying the perceived efficiency of energy conversions (e.g., through arrows exclusively indicating utilization and regeneration of electric energy), the energy interface should visualize the maintenance and loss (e.g., indicated by the size of arrows or by branched arrows with one branch implying losses) of kinetic energy. The potential of this approach, however, for day-to-day driving and possible negative side effects on safety (Kircher, Fors, & Ahlstrom, 2014) must be examined first. Furthermore, usage of the neutral mode should be encouraged (see Q2), for example, by increasing its accessibility (e.g., as gas pedal position).

Conclusion

The present research provided first quantitative insights into drivers’ interaction with key HEV eco-features. Fuel efficiency, and thus the ultimate sustainability effect of HEVs, was shown to be related to interaction patterns with HEV eco-features. Furthermore, the present research provided first evidence of a general pattern of user–energy interaction that can be condensed into one dimension that classifies drivers based upon their strategy regarding energy conversions or reference energy. The link of this dimension to eco-driving success can be considered as an indication for the practical significance of understanding driver interaction with eco-features from the perspective of user–energy interaction.

Key Points

Drivers’ interaction patterns with hybrid electric vehicle (HEV) eco-features were related to eco-driving success (fuel efficiency).

An aggregated factor that represented a general pattern of interacting with HEV eco-features (intensity of actively dealing with electric energy) was related to eco-driving success, indicating a fuel saving potential of about 11% (0.4–0.5 l/100 km, or 0.17–0.21 gallons per 100 miles) for many drivers.

Technical system knowledge and eco-driving motivation were not related to interaction patterns with HEV eco-features but constituted significant predictors of higher fuel efficiency.

The results emphasize the importance of understanding user–energy interaction in HEVs for achieving an optimal sustainability effect (e.g., by changing system feedback).

Footnotes

Appendix

Translated Items of All Scales Used in the Present Research

Scale Label	Item Text
Perceived efficiency of electric propulsion	Scale instruction: How do you evaluate the efficiency of the electric propulsion for the average fuel consumption on a longer route? Items: The more one drives actively electric, the better it is for the average fuel consumption on the route. The less one forces the electric propulsion, the more fuel efficient the vehicle. Routes I drive electric are virtually gratis. Routes I drive electric also cost fuel at the end of the day. To drive fuel efficiently, the electric propulsion should be fully exploited.
Perceived efficiency of regenerative braking	Scale instruction: How do you evaluate the efficiency of regenerative braking for the average fuel consumption on a longer route? Items: The more one recuperates, the better it is for the average fuel consumption on the route. The less one deploys regenerative braking in decelerations, the more fuel efficient. Through regenerative braking, I basically receive energy as a gift. Despite regenerative braking, a large amount of kinetic energy is lost in deceleration. To drive fuel efficiently, regenerative braking should be fully exploited.
Perceived efficiency of neutral mode	Scale instruction: How do you evaluate the efficiency of the neutral mode for the average fuel consumption on a longer route? Items: The more one rolls in neutral mode, the better it is for the average fuel consumption on the route. The less one deploys the neutral mode, the more fuel efficient the vehicle. Neutral rolling is a waste of energy because I do not recuperate it. To drive fuel efficiently, regenerative braking should be fully exploited.
Usage dimensions of eco-features	Description of the driving situation (the same for all dimensions): You are alone on the road with your HEV. The weather is sunny and calm/windless with 20°C. The car (the engine/motor) is already warm. You are familiar with the route. The terrain is relatively flat. There is little traffic on the route. You are on a longer route (>20 km) in an urban area (main roads and byroads; with speed limits of 30, 50, and 70 km/h; traffic lights; crossings). Scale instruction (the same for all dimensions): When I am on the road with my HEV, I regularly utilize the following strategies (i.e., when I am on the road with my HEV, I try to . . .)
Intensity of using electric propulsion	Items: . . . drive electric as much as possible. . . . drive actively electric as little as possible.
Intensity of using regenerative braking	Items: . . . recuperate as often as possible. . . . recuperate as seldom as possible.
Acceleration	Items: . . . accelerate after traffic lights with the electric propulsion for as long as possible. . . . to deploy as little electric energy as possible in acceleration after traffic lights.
Deceleration	Items: . . . decelerate without regenerative braking (in the neutral mode) when possible. . . . decelerate by using regenerative braking when possible.
Technical system knowledge	Items: I am familiar with the propulsion technology of hybrid cars (e.g., types and functionality of electric motors). I am familiar with concepts like energy density and energy conversion efficiency. I have an idea about how regenerative braking technically works.
Eco-driving motivation	Items: I am always striving to drive fuel efficiently. I often try to drive as efficiently as possible.
Fuel efficiency	Items for estimated fuel consumption: What was the average fuel consumption you achieved with your currently most-used HEV in the previous summertime (beginning of April to end of September 2015)? Items for FEI.urban (driving situation the same as for interaction dimensions): What would be your typical fuel consumption under these conditions? Items FEI.rural (driving situation the same as for interaction dimensions, except for the last point, which was changed accordingly): You are on a longer route (>20 km) in a rural area with country roads and small villages. What would your typical fuel consumption be under these conditions?

Note. HEV = hybrid electric vehicle; FEI = fuel efficiency indicator.

Acknowledgements

We thank Josef Krems for providing parts of the research infrastructure for the present study.

Matthias G. Arend is a research intern at the Department of Cognitive and Engineering Psychology at Technische Universitat Chemnitz.

Thomas Franke is an appointed professor of engineering psychology and cognitive ergonomics at the University of Lübeck. He received his PhD in psychology from the Department of Cognitive and Engineering Psychology at Technische Universität Chemnitz in 2014.

References

af Wåhlberg

A. E.

(2007). Long-term effects of training in economical driving: Fuel consumption, accidents, driver acceleration behavior and technical feedback. International Journal of Industrial Ergonomics, 37, 333–343. doi:10.1016/j.ergon.2006.12.003

Ajzen

(1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179–211. doi:10.1016/0749-5978(91)90020-T

Anderson

J. R.

(1982). Acquisition of cognitive skill. Psychological Review, 89, 369–406. doi:10.1037/0033-295X.89.4.369

Barkenbus

J. N.

(2010). Eco-driving: An overlooked climate change initiative. Energy Policy, 38, 762–769. doi:10.1016/j.enpol.2009.10.021

Bingham

Walsh

Carroll

(2012). Impact of driving characteristics on electric vehicle energy consumption and range. IET Intelligent Transport Systems, 6, 29–35. doi:10.1049/iet-its.2010.0137

Bitsche

Gutmann

(2004). Systems for hybrid cars. Journal of Power Sources, 127, 8–15. doi:10.1016/j.jpowsour.2003.09.003

Cocron

Bühler

Neumann

Franke

Krems

J. F.

Schwalm

Keinath

(2011). Methods of evaluating electric vehicles from a user’s perspective: The MINI E field trial in Berlin. IET Intelligent Transport Systems, 5, 127–133. doi:10.1049/iet-its.2010.0126

Cohen

(1992). A power primer. Quantitative Methods in Psychology, 112, 155–159. doi:10.1037/0033-2909.112.1.155

Davidse

R. J.

Hagenzieker

M. P.

van Wolffelaar

P. C.

Brouwer

W. H.

(2009). Effects of in-car support on mental workload and driving performance of older drivers. Human Factors, 51, 463–476. doi:10.1177/0018720809344977

10.

Davis

S. J.

Caldeira

Matthews

H. D.

(2010). Future CO₂ emissions and climate change from existing energy infrastructure. Science, 329, 1330–1333. doi:10.1126/science.1188566

11.

Dijk

Orsato

R. J.

Kemp

(2013). The emergence of an electric mobility trajectory. Energy Policy, 52, 135–145. doi:10.1016/j.enpol.2012.04.024

12.

Dogan

Steg

Delhomme

(2011). The influence of multiple goals on driving behavior: The case of safety, time saving, and fuel saving. Accident Analysis & Prevention, 43, 1635–1643. doi:10.1016/j.aap.2011.03.002

13.

Faul

Erdfelder

Buchner

Lang

A.-G.

(2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41, 1149–1160. doi:10.3758/BRM.41.4.1149

14.

Faul

Erdfelder

Lang

A.-G.

Buchner

(2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175–191. doi:10.3758/BF03193146

15.

Field

A. P.

(2009). Discovering statistics using SPSS. London, UK: Sage.

16.

Franke

Arend

M. G.

McIlroy

R. C.

Stanton

N. A.

(2016a). Ecodriving in hybrid electric vehicles: Exploring challenges for user–energy interaction. Applied Ergonomics, 55, 33–45. doi:10.1016/j.apergo.2016.01.007

17.

Franke

Arend

M. G.

McIlroy

R. C.

Stanton

N. A.

(2016b). What drives ecodriving? Hybrid electric vehicle drivers’ goals and motivations to perform energy efficient driving behaviors. In Stanton

N.A.

Landry

Di Bucchianico

Vallicelli

(Eds.), Advances in human aspects of transportation (pp. 451–461). London, UK: Springer. doi:10.1007/978-3-319-41682-3_38

18.

Franke

Rauh

Günther

Trantow

Krems

J. F.

(2015). Which factors can protect against range stress in everyday usage of battery electric vehicles? Towards enhancing sustainability of electric mobility systems. Human Factors, 58, 13–26. doi:10.1177/0018720815614702

19.

Fuller

(2005). Towards a general theory of driver behaviour. Accident Analysis & Prevention, 37, 461–472. doi:10.1016/j.aap.2004.11.003

20.

Fuller

(2007). Motivational determinants of control in driving task. In Cacciabue

P.C.

(Ed.), Modelling driver behaviour in automotive environments: Critical issues in driver interactions with intelligent transport systems, (pp. 165–188). London, UK: Springer. doi:10.1007/978-1-84628-618-6_10

21.

Fuller

(2011). Driver control theory: From task difficulty homeostasis to risk allostasis. In Porter

B. E.

(Ed.), Handbook of traffic psychology (pp. 13–26). Amsterdam, Netherlands: Elsevier. doi:10.1016/B978-0-12-381984-0.10002-5

22.

Gilman

Keskinarkaus

Tamminen

Pirttikangas

Röning

(2015). Personalised assistance for fuel-efficient driving. Transportation Research Part C, 58, 681–705. doi:10.1016/j.trc.2015.02.007

23.

Hanson

M. A.

(2013). Green ergonomics: Challenges and opportunities. Ergonomics, 56, 399–408. doi:10.1080/00140139.2012.751457

24.

Jamson

A. H.

Hibberd

D. L.

Merat

(2015). Interface design considerations for an in-vehicle eco-driving assistance system. Transportation Research Part C: Emerging Technologies, 58, 642–656. doi:10.1016/j.trc.2014.12.008

25.

Kahneman

Tversky

(1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263–292. doi:10.2307/1914185

26.

Kahneman

Tversky

(1984). Choices, values, and frames. American Psychologist, 39, 341–350. doi:10.1037/0003-066X.39.4.341

27.

Kimmel

H. D.

(1957). Three criteria for the use of one-tailed tests. Psychological Bulletin, 54, 351–353. doi:10.1037/h0046737

28.

Kircher

Fors

Ahlstrom

(2014). Continuous versus intermittent presentation of visual eco-driving advice. Transportation Research Part F: Traffic Psychology and Behaviour, 24, 27–38. doi:10.1016/j.trf.2014.02.007

29.

Kuriyama

Yamamoto

Miyatake

(2010). Theoretical study on eco-driving technique for an electric vehicle with dynamic programming. In Proceedings of the 2010 International Conference on Electrical Machines and Systems (ICEMS) (pp. 2026–2030). New York, NY: IEEE.

30.

Lauper

Moser

Fischer

Matthies

Kaufmann-Hayoz

(2015). Psychological predictors of eco-driving: A longitudinal study. Transportation Research Part F: Traffic Psychology and Behaviour, 33, 27–37. doi:10.1016/j.trf.2015.06.005

31.

McIlroy

R. C.

Stanton

N. A.

(2015). A decision ladder analysis of eco-driving: The first step towards fuel-efficient driving behaviour. Ergonomics, 58, 866–882. doi:10.1080/00140139.2014.997807

32.

McIlroy

R. C.

Stanton

N. A.

Harvey

Robertson

(2013). Sustainability, transport and design: Reviewing the prospects for safely encouraging eco-driving. In Proceedings of the 5th International Conference on Automotive User Interfaces and Interactive Vehicular Applications (AutomotiveUI ’13) (pp. 278–284). doi:10.1145/2516540.2516578

33.

Michon

J. A.

(1985). A critical view of driver behavior models: What do we know, what should we do? In Evans

Schwing

R. C.

(Eds.), Human behavior and traffic safety (pp. 485–520). New York, NY: Plenum.

34.

Pampel

S. M.

Jamson

S. L.

Hibberd

D. L.

Barnard

(2015). How I reduce fuel consumption: An experimental study on mental models of eco-driving. Transportation Research Part C: Emerging Technologies, 58, 669–680. doi:10.1016/j.trc.2015.02.005

35.

Schönbrodt

F. D.

Perugini

(2013). At what sample size do correlations stabilize? Journal of Research in Personality, 47, 609–612. doi:10.1016/j.jrp.2013.05.009

36.

Sivak

Schoettle

(2012). Eco-driving: Strategic, tactical, and operational decisions of the driver that influence vehicle fuel economy. Transport Policy, 22, 96–99. doi:10.1016/j.tranpol.2012.05.010

37.

Stanton

N. A.

Salmon

P. M.

Rafferty

L. A.

Walker

G. H.

Baber

Jenkins

D. P.

(2013). Human factors methods: A practical guide for engineering and design. Farnham, UK: Ashgate.

38.

Stevens

J. P.

(2012). Exploratory and confirmatory factor analysis. In Applied multivariate statistics for the social sciences (pp. 325–394). New York: Routledge.

39.

Stillwater

Kurani

K. S.

(2013). Drivers discuss ecodriving feedback: Goal setting, framing, and anchoring motivate new behaviors. Transportation Research Part F: Traffic Psychology and Behaviour, 19, 85–96. doi:10.1016/j.trf.2013.03.007

40.

Summala

(2007). Towards understanding motivational and emotional factors in driver behaviour: Comfort through satisficing. In Cacciabue

P. C.

(Ed.), Modelling driver behaviour in automotive environments (pp. 189–207). London, UK: Springer. doi:10.1007/978-1-84628-618-6_11

41.

Thatcher

(2013). Green ergonomics: Definition and scope. Ergonomics, 56, 389–398. doi:10.1080/00140139.2012.718371

42.

Tversky

Kahneman

(1992). Advances in prospect theory: Cumulative representation of uncertainty. Journal of Risk and Uncertainty, 5, 297–323. doi:10.1007/BF00122574