4.1. Explorative assessment to the effect of PC, PU and MB on likelihood of use
4.1.1. The survey and sample design
The Delft University of Technology and Deloitte Consulting joined forces to conduct the explorative survey. The survey required Dutch insurance policy holders to evaluate eight categories of mobile insurance on four generic constructs. These generic constructs are directly derived from the constructed hypotheses from chapter three. The questionnaire also included additional items on request of Deloitte which are outside the scope of this thesis
Population & sample design
Given the research aim and questions, the population of interest for this explorative survey is the entire group of Dutch (consumer) insurance policy holders. Consequently the unit of analysis is the Dutch insurance policy holder.
43 With Dutch insurance policy holders the focus is set on consumers, disregarding business related insurances. A broad range of consumers fit this definition, covering almost the entire Dutch adult population. Therefore, the group of Dutch (consumer) insurance policy holders is defined in this study as the entire adult population of the Netherlands (>18 year). This comes down to a population of more than 13 million people (CBS, 2014a).
Survey pretest
In order to assess the comprehensiveness and quality, the survey is pretested within the organization of Deloitte with approx. 4 people. After several adjustments and retesting, the survey is tested on its comprehensiveness with a three random participants. Hereafter, the survey was launched and distributed through multiple channels.
An unavoidable downside of the explorative survey is its length. In order to decrease the duration, the questionnaire mainly involves multiple-choice questions and no mandatory open questions were included. The final survey, including generic and additional questions has a duration of 20 to 25 minutes.
Data collection procedure
Due to practical reasons it is not possible to include the entire population of interest in this study. The aim was to conduct a maximum number of valid responses within the time constraints of this research.
Eventually the explorative survey obtained 137 valid responses, corresponding with a margin of error (MOE) of 8,37% at a 95% confidence interval. The MOE is the amount of random sampling error in the survey's result; the smaller the sample the larger margin of error (Kotz, Read, Balakrishnan, Vidakovic,
& Johnson, 2004)6. Most studies consider a maximum MOE of 5% as acceptable, highlighting the disclaimer to interpret the survey results with care (Simon, 2006).
Respondents were approached through multiple channels to participate in this survey. By means of active data collection, hard-copy surveys were conducted on the street and in the train (offline).
Passive data collection is applied by inviting respondents for an electronic survey (online).
Passive data collection (online):
Electronic surveys were composed with Surveymonkey.com, collector.tudelft.nl and Surveygizmo.com.
Finally the latter was selected based on its comprehensiveness and usability.
The link to the electronic survey is distributed through social networks of Deloitte, study associations Curius and relatives (5-11-2014). The link to the electronic survey was accessible between 5-11-2014 and 10-11-2014.
Approximately 2.100 leaflets with a link to the electronic survey were distributed to the following postal codes: 1021CW, 1021VK, 1102LA, 1102KA, 1102HE, 1102HH, 2625LK, 2625LE, 4707KP, 4707KJ, 4707KE (6-11-2014). Approximately 1300 leaflets are distributed in the area of Amsterdam south-east (postal code 1102xx), a relative low educated area (CBS, 2013). The other 900 leaflets were distributed in middle class areas in Amsterdam north, Delft and Roosendaal. This distribution seems to cover the population as presented in 4.1.2.
Active data collection (offline):
6Formula for the margin of error (MOE) with: p = probability of proportion of population (0,5), n = sample size (137) and N = population size (13.386.487):
MOE = 1,96 √(𝑝(1 − 𝑝)
𝑛 ) ∗ √(𝑁 − 𝑛 𝑁 − 1)
44
For the hard-copy survey, SurveyGizmo’s hard-copy feature is used. Herewith, a fully identic hard-copy version of the electronic survey is downloaded.
Hardcopy surveys were conducted at commuters in train station Bijlmer Arena, Amsterdam (6-11-2014). Due to the duration of the survey, little commuters were willing to participate.
Hardcopy surveys were, with approval of the Dutch railways, conducted at commuters on board of the train, railway line Amsterdam – Groningen (7-11-2014). This railway line is chosen because the interval time between most stations exceeds the duration of the survey.
In order to stimulate the survey response, cinema ticket and ITunes credits were allot among the participants. In order to avoid missing values, all questions in the electronic questionnaire were mandatory, except for irrelevant practical questions. Respondent were unable to proceed the questionnaire with unfilled questions, resulting in zero missing values. Missing values and illegal codes in the hard-copy questionnaires were immediately followed up with the respondent in order to fully complete the questionnaire. Incomplete questionnaires are disqualified from the dataset, 6 in total.
Survey completion
Eventually, 75 surveys were completed online and 62 were completed offline. As aforementioned respondents were unable to submit an incomplete survey online. Thus no online results had missing values, however the dropout rate is rather high. From the 178 respondents, who commenced the online questionnaire 42% submitted it. It has to be mentioned that the fill-in rate of the dropouts is rather low, thus respondents quieted the questionnaire quite fast.
The dropout rate in the offline questionnaire was rather low; 9,7%. An argument for this relative low dropout rate could be the feeling of urgency created by inviting them personally or their time availability since these questionnaires were mainly filled in the train.
Representativeness
The representativeness of the explorative assessment is tested with background questions on gender, age and education. Gender: 55% of the sample was male compared with 49% in the population (CBS, 2014a). The other way around, 45% was female compared with 51% in the population. The difference on gender between the sample and the population is minimal and therefore representative. Figure 4.1 provides the distribution of age and educational level over the sample and population (CBS, 2014a).
Figure 4.1: Explorative survey representation (age & education)
The sample shows a clear overrepresentation of low age and high educated (academic) people. Due to this overrepresentation the outcomes of this survey cannot blind be generalized to the population.
0%
5%
10%
15%
20%
25%
30%
35%
40%
Sample [%] Dutch population [%]
0%
10%
20%
30%
40%
50%
60%
45