Discussion of the findings

This section’s primary focus is on comparing the findings of the empirical research to those made by prior studies on the topic. Not all of these reflections can be

considered fully accurate because there can be differences in the data gathering instruments, the survey and the question formats and the interpretation of the results by different authors.

In the literature review it was established that the public has safety concerns over AVs, and much of these are caused by the fear that the system cannot drive as well as humans can (Rödel et al 2014; Johnsen et al 2017; Pew Research Center 2017;

Deloitte 2018; Lienert 2018b). Fluctuation has been a trend in perception of AV safety as 2018 was the year of highly publicized AV road accidents. It is therefore surprising that in the survey conducted as part of this thesis work, confidence in AV safety was deemed very high. Only 17.3 of the respondents thought that AVs will be worse drivers than humans, and 20.7 percent thought AVs will be less safe than regular vehicles. Explaining the reason behind these results is much more difficult.

It is possible that since the accidents in 2018 were not domestic, they had less of an effect on Finnish and European respondents than they did on the Americans. It could also be that since the respondent group had highly favorable views towards new technologies in general, they may trust that technological development of autonomous vehicles will eventually sort out most safety related concerns.

Neither explanatory or descriptive analysis found notable differences between the genders in acceptance of autonomous vehicles (Appendix 7.3). This is somewhat surprising as prior studies had made clearer observations that men approved of AVs better than women do (Payre et al 2014; Honenberger et al 2016; Johnsen et al 2017). There is no clear indicator in the survey results which could explain this deviation from the findings made by other studies. It can simply be a coincidence.

Preference to keep manual controls in AVs seems to have remained relatively consistent over the years. In the survey 90.3 percent of the respondents wanted AVs to keep the controls for manual drive while in an earlier study by Schoettle and Sivak (2015) this preference was at 96.2 percent. A Cox automotive (2018) poll found this percentage to be at 84 percent, despite the fact that it was conducted right in the heels of the first lethal car accident involving an AV.

Ward et al (2017) knowledge of AV technology can significantly influence the likelihood to purchase an AV. Statistical significance of news for WTP could not be verified in the logistic regression analysis, but a cross tabulation reveals clearer results (Appendix 7.4). 13.5 percent of the respondents who follow AV related news at least somewhat actively answered that they will “highly likely” own an AV one day while none of the non-followers answered the same. This information is aligned with Rogers’ (2003) theory on innovation decision process, and how individuals who are considering adoption are also actively seeking information about it.

Choi and Ji (2015) and Hohenberger et al (2016) found that perceived usefulness and anxiety are strong predictors of intentions to use AVs. In the multiple linear regression analysis perceived usefulness was the 3rd highest indicator of intentions to use while anxiety was the 2nd. Therefore, the causal relationship that these two variables have with intentions to use is beginning to seem solid, while other variables of the car technology acceptance model need further testing.

Table 20. Comparison of intentions to use

Source Probability to use AV

Nordhoff et al 2018 90.8 %

Ellis et al 2016 75.0 %

Payre et al 2014 68.0 %

This study 2019 67.7 %

Gartner Inc 2017 45.0 %

When asked whether the respondents would see themselves taking a ride in an AV, 67.7 percent answered “somewhat likely” or higher. Table 20 shows that in prior research probability to use has varied between 45.0 to 90.8 percent. There are more notable polls that have measured low intentions to use AVs, but their results are not explicitly comparable to other studies listed in Table 20 due to different format of the questions (Schoettle & Sivak 2014; Bansal & Kockelman 2017; Daziano et al 2017).

While certain repeated polls have witnessed a downward trend in intentions to use AVs partly due to traffic accidents involving AVs in 2018, overall it seems that intentions to use AVs have remained at a high level (Cox Automotive 2018).

Table 21. Comparison of mean WTP

Source Inflation adjusted mean WTP in EUR

Bansal et al 2016 6 656 EUR

Bansal and Kockelman 2017 5 265 EUR

Daziano et al 2017 4 400 EUR

This study 2019 3592 EUR

Schoettle and Sivak 2014 1000 - 1500 EUR*

IHS Markit 2017 915 EUR**

*Estimation, clear mean WTP not reported, **Mean WTP for highest region, Germany

Table 21 shows that willingness to pay for AVs was among the respondents was at a moderate level when compared to prior research. The variation in the results is quite extreme and it is therefore difficult to establish a common average or a rule-of-thumb for WTP. If only the region of the studies is considered, the American studies (Bansal et al 2016; Bansal & Kockelman 2017) have observed a higher level of WTP than the studies conducted in Europe and Australia. Most of the studies that are listed in Table 21 also observed that more than half of the respondents would not pay anything at all for AV technology. This study however made a different conclusion as only about a fourth of the respondents were non-payers.

It could be expected that measuring willingness to pay with a quantitative survey would be inaccurate unless it was supplemented by a much wider range of supportive questions. These could include such variables as how much the respondents usually pay for car technologies; do they purchase their vehicles new or second-hand and how many people use the same vehicle. This could have helped to profile the respondents better in order to evaluate more clearly what factors affect their responses on WTP. The reason why WTP was not measured this exhaustively

in this study was because it would have taken too much space from other important questions. It seems likely however that if AV academia wishes to establish some level of consensus on willingness to pay, the research needs to begin using a more varied line of questioning, and more background variables need to be controlled for than just the bare minimum.