We collected nominal, ordinal and metric data using different types of single- & multiple-item scales to judge preferences, usage behavior, and some demographic characteristics of the members of SNWs as well as their willingness to pay for the service.
ANALYSIS RESULTS
Demographics
The sample can be characterized using frequency distribution for each variable of interest. For this particular marketing research project, the demographic data collected included country (nationality), gender and age. As such, the following graphs encapsulate the demographic composition of all respondents.
As indicated in the chart above, the distribution between the male and the female respondents is reflective of the actual world population with 43% of men and 57% for women.
A descriptive statistics was applied to the variable age. The results could be found in the table and the graph below. The average age of the respondents is 23,23 years, with the minimum age of 18 and the maximum age of 34. It can be seen from the Box Plot below, that 50% of the respondents are between 20 and 25 years old.
The nationalities involved are Slovak, Hungarian, Austrian, US American. Those countries represented by few respondents are group in the category labeled Other.
General Important Findings
- Which SNWs are people registered with?
With the Multiple Response tool we could identify that following frequencies regarding the registration with a particular SNW occure: StudiVZ (20,6%), Facebook (34,9%), MySpace (16,3%), Xing (9,1%), IWIW (12,7%) and 6,3% of the respondents stated also another SNW, differnt from those they could choose as the predefined answer. The high representation of the IWIW (originates in Hungary) may be explained by the significant number of Hungarian respondents to the questionnaire (28,12%). Thie distribution of results above highly corresponds with the distribution of the respondents favorite SNWs.
Furthermore, the frequency table also indicates that on average each repondent is registered cca. to 2 different SNWs.
- How much are people willing to pay?
A very important finding for our futher analyses was that at the moment none of the SNW is charging any fee for the services offered. This raises the question whether the users would be willing to pay or not.
In our online questionnaire, we asked several seperate questions regarding willingness of users to pay. In the first, users were asked about the maximum price they would pay per month. The average price per month people would be willing to pay for the membership to their current SNW is 1,07 Euros, whereas the maximum sum given was 8 Euros. 59,8% of the respondents would not be willing to pay anything.
In another question, users were asked to indicate a fair price that SNWs could charge them. A total of 85,9% of respondents would pay between 0 and 2 Euros monthly.
In a third question, users were asked to indicate the maximum difference they would pay for using their favorite SNW. Given the range from 0 to 3+ Euros, the significant majority of 65,6% stated that they would not accept paying more for their favorite SNW. These findings indicate that the demand for SNWs is elastic, because users would switch from their favorite SNW to another due to a small difference in the price level.
Fee per month willing to pay for membership to the favorite SNW that is considered to be a fair price
Maximum price willing to pay per months for membership to the current SNW
Maximum difference willing to pay if the fee for using the favorite SNW higher then for all the others
- What is the percentage of time spent on SNW out of total time on Internet?
The Box Plot below shows that the average person spends 21,81% of his/her total internet time on the SNW. Most respondents actively spend there up to half an hour per day.
Time on SNW/total time on internet (%)
Usefullness Assesment
Factor analysis
In one of the questions of our research, we wanted to find out how useful people the different sources of information on SNWs find. We wanted to correlate the results of this question with their willingness to pay. In the question we asked the respondents to rate the usefulness of 5 different fields in which we can find information on SNWs. We measured the results with a likert scale with 5 different answers ranging from “very useful” to “not useful at all”. Our hypothesis was that the higher the value of information received, the higher the users’ willingness to pay a membership fee.
We conducted a factor analysis for this question in order to reduce the number of variables of the different information usage. At first we looked at the Keiser-Meyer-Olkin Measure and the Bartlett's test of sphericity, both of which indicted that the factor analysis would be useful with our data.
As we can see from the analysis table and from the scree plot above there are two components that have an Eigenvalues higher than 1, they explain ca. 68,5% of the variability in the original 5 variables. We can therefore considerably reduce the complexity of the data set by using only these components.
(Questionnaire #12)
Based on the rotated component matrix above, we have grouped the variables that influence the assessment of the usefulness of information from SNWs into 2 factors.
The variables “Gaining access to information about courses“, “Finding and/or creating a study group“ and “Getting information about job/work opportunities“ were formed into one group that we named the “usefulness of information from a practical aspect”. The variables “Maintaining relationships with friends“ as well as “Finding information about upcoming events“ formed a group, we will from now on refer to as the “usefulness of information from a social aspect”. With the factor analysis we also created two new variables in our data table. These “factor scores” represent our two new groups: practical aspect and social aspect. To make it easier we use these new variables for our correlation analysis later on.
Preference Assessment
Factor Analysis
The variable representing reasons why the SNW is considered to be the favorite one, was also regrouped using the factor analysis.
(Questionnaire #13)
The Scree plot also shows an Eigenvalue greater than 1. After rotating the component matrix using VARIMAX we were able to extract 2 factors. “Many friends registered“ as well as „Being used to the SNW and knowing it best“ was comprised into the factor dependence. Lets call the following 3 variables “Being easier to use“, “Preference of the applications“ and “Access to more valuable information“ convenience.
HYPOTHESIS TESTING
- Does the size of the network influence the user’s willingness to pay?
Hypothesis: The larger the network size, the higher the willingness to pay for a SNW.
Null-hypothesis: There is no correlation between the network size and the amount a SNW user is willing to pay.
The following analyses were developed based on the data collected regarding the level of willingness to pay when compared to the size of the network that a user belongs to. We created the following groupings of the data, in order to simplify the analysis and calculations:
(NetworkGroups): 1.00 = 1 network
2.00 = 2 networks
3.00 = 3+ networks
(MaxPayGroups): 1.00 = 0 Euros
2.00 = 1-2 Euro
3.00 = 3 and more Euros
This data includes one missing value, which constitutes for only 0.8% of the cases.
As our data is not metric, but rather ordinal, it did not fulfill the requirements for using the Pearson Correlation. Therefore, we decided to run the Spearman´s Rank Correlation test.
From the Spearman Rho test, we can conclude that there is not a significant correlation between the number of networks that a SNW user belongs to and the maximum that a user is willing to pay. However, there is a slight tendency (.204) to support our hypothesis, but this tendency is not large enough to draw a favorable conclusion.
Our second variable to measure the network size was the number of contacts our respondents have on their profile.
We also regrouped the variable for the maximum price the respondents would be willing to pay monthly for a SNW:
(How Many Contacts): 1.00 = 1 - 40 contacts
2.00 = 41- 80 contacts
3.00 = 81 - 160 contacts
4.00 = 161 - 309 contacts
5.00 = 310 + contacts
With the above mentioned grouped willingness to pay and the grouped contacts we also did a Spearman correlation:
We have to state once again that the Spearman Rho value shows that there is no correlation between the variables. Therefore the Null-Hypothesis is approved.
- Does the degree of satisfaction with the applications offered lead to a higher willingness to pay?
Hypothesis: The more aligned with the users’ interests the applications are, the higher is the willingness to pay.
Null hypothesis: there is no significant relationship between the willingness to pay for applications offered on an SNW and the user’s satisfaction with the applications.
Initially, we tested the correlation between the extent that users said they agreed or disagreed that they like the applications offered by the SNW they are registered for and the amount that the user would be willing to pay per application. To simplify the analysis, the amount willing to pay was grouped into three categories:
- Not Willing (0 Euros)
- Willing to Pay Small Amount (0.1 Euros – 2 Euros)
- Willing to Pay Large Amount (2.1 Euros and more Euros)
According to the output, there is a very weak correlation between the amount users are willing to pay and the extent to which they like the applications. (The correlation appears negative in this table due to the ordering of the variables. See below figure, which illustrates the relationship.)
Since there is not a strong correlation between these variables, we must accept the null hypothesis, that there is no relationship between a user’s willingness to pay and their feelings towards the offered applications.
When looking at the means of the different categories of satisfaction (1 being the Highest) represented by the dots, we can state that, even though the correlation is weak and not significant, a slight tendency still exists.
Though there is not a relationship between the amount of the willingness to pay and the offered applications, we found some interesting results in a cross-tabulation between the same variables. There are 89 respondents who are unwilling to pay regardless if they like SNW applications. However, of the remaining 37 respondents who were willing to pay for applications added, only 3 of them are completely dissatisfied with the offered applications.
The reason that there is not a linear correlation between the payment and applications is because the people who are the most highly satisfied are not willing to pay in the High category (3), but rather in the Low category (2), so this threw off the correlation. There is, as shown in this cross-tabulation, a relationship between willing and not willing to pay and the degree of satisfaction.
- Does higher usage intensity positively affect the willingness to pay?
Hypothesis: The higher the usage intensity, the higher is the willingness to pay.
Null Hypothesis: There is no correlation between the usage intensity and the level of willingness to pay.
To test the usage intensity we asked our respondents the following questions:
- How many messages do you send per day?
- How often do you log on to your profile?
- What percentage of your total time spent on the internet do you spend on your favourite SNW?
- Estimate the average time per day in minutes that you actively use you favourite SNW.
We tested the all the responses for these questions with a Spearman’s rank correlation between each variable and the variable” level of willingness to pay”. To run the test we grouped the percentage of time spent on SNWs with regard to the total time spent on the internet in four quartiles.
The results showed that for there is a weak positive correlation between the variables “Messages sent per day”, “Percentage of total time spent on the internet spent on a SNW?” and “Average time per day usage of favourite SNW in minutes” and the level of willingness to pay” with significance level below 0,3.
Between the variable “How many times per day do you log on to your profile?” and the level of willingness to pay there is a middle correlation with a significance level of 0,320.
With the low correlations for the variables testing the usage intensity against the willingness to pay we have to accept the Null-hypothesis.
- Do the different sources of information available and their percieved value on the SNWs influence the users’ willingness to pay?
Hypothesis: The higher the value of information perceived, the higher the willingness to pay.
Null Hypothesis: There is no correlation between the value of percieved information and payment willingness.
We have already conducted a factor analysis about the question concerning the usefulnes of different sources of information (See Facor analysis earlier in this report). For the correlation analysis we are using the two new variables that we have created in the factor analysis. In the correlation analysis we are using Bivariate Correlations procedure as were working with scale variables. We correlated our two new variables: social aspect and the practical aspect with the willingness of users to pay.
As we see from the two tables above, we can identify a positive correlation in both cases. As the significance level is very small the correlation is significant. This means that we have confirmed our hypothesis. We have also produced two scatter plots for the two tables to identify outlier variables and to see if there is a possibility to improve this result but we did not find any significant outlier variables. If we compare the two variables: social aspect and practical aspect we see that the practical aspect is more strongly correlated to the willingness to pay (with 0,3) than the social aspect (0,231). This means that people who find practical information on SNWs useful would be more willing to pay a monthly fee than those who use these sites to maintain their social relationships.
- What is the relationship between willingness to pay and the dependence on a certain SNW?
Hypothesis: The more dependent a user is on a certain SNW, the more likely they are willing to pay.
Null Hypothesis: There was no significant difference between people willing to pay and those not willing to pay for the services of a SNW concerning the dependence on this particular SNW.
First, we recoded the variable fair price the respondents will be willing to pay as follows:
Willingness to Pay: Not willing = 0 Euros (61 cases)
Willing = 1 Euro and more (67 cases)
As a result of the factor analysis conducted earlier (see part SNW Preference), the original variables: “Many friends registered“ as well as „Being used to the SNW and knowing it best“ were replaced by the factor called dependence (on the particular SNW). We used the factor scores which replaced the original variables for the analysis below.
To verify the hypothesis we used Mann-Whitney test as there were only two independent samples (willing/not willing to pay):
We can see that p > 0,05 (p=0,449) and therefore we do not reject the null hypothesis. There is no significant difference in the dependence on the particular SNW between the 2 groups defined as willing and not willing to pay.
- How would the introduction of a mandatory fee for SNW affect student’s choice to remain registered?
Hypothesis: If a mandatory fee was charged to students for registration on a SNW, then students would choose to remain registered.
Null Hypothesis: There is no significant relationship between the introduction of a mandatory fee and the users´ decision to remain registered.
In the Crosstabs below, we compared the respondents’ likelihood to remain registered despite a mandatory fee with the maximum price they would be willing to pay for remaining registered. From the findings, we can deduce that the majority of respondents who answered that they would not be willing to pay any type of fee also answered that they would not remain registered. 35 out of a total 76 people who would refuse to pay for SNW services also said that they would definitely consider de-registering. 25 of those who would refuse to pay said that they would probably consider de-registering. In total, 60/76 people who would not pay would, at some level, consider de-registration from their account.
This Crosstabulation also shows that of the total 127 respondents to this question, 87 (42 + 45) people said that they would definitely or probably consider de-registering, while only 40 (20+20) would consider remaining registered. This result is highly significant, because it reveals the opposite of what we hypothesized, therefore, we must accept the null hypothesis that there is no direct correlation between the price one is willing to pay and likelihood to remain registered. When faced with a fee, 2.175 times more people would consider de-registering than remaining registered.
CONCLUSIONS AND RECOMMENDATIONS
In conclusion, we found that there are no strong correlations between usage intensity, dependence, satisfaction with applications, network size, and perception of information with the level of willingness to pay for an SNW. Though there are some slight tendencies, none of them are highly significant. Overall, people are simply unwilling to pay for the services offered by an SNW.
We recommend that the firms who own and operate the Social Networking Webistes not assess a monthly fee for registration at this time. Since SNWs are fairly new to users, this could be a possible reason for their unwillingness to pay. If the operators of these sites needed extra revenue, we would recommend that they either wait until a later time to assess a fee (when perhaps the user’s level of attechment to the site increases) or find other means of funding (such as increased advertising).
LIMITATIONS
In the course of the SPSS session we have realized that our questionnaire was developed without strong statistical knowledge. Members of our team were only familiar with the basic theorie in statisctics, so when developing our questionnaire, we were limited to the kinds of questions and scales that we could apply.
The questionnaire did not allow us to collect the type of data needed to run the more sophisticated analyses. Though we had metric data, for example on the maximum payment question, the categories were not sensitive enough to reflect a correlation in the SPSS analysis. Though some questions included in the questionnaire were originally constructed to yield metric data (such as a percentage answer), in order to analyze the data in SPSS, we had to group answers, which turned the data into categorical.
In the questionnaire we have also included a question (number 21) asking if the users are currently paying for the services of a SNW. However, all the respondents answered with no. Due to this fact the 2 branching questions related to the question (21) were irrelevant. As a consequence, we could analyse the willingness to pay only among the people who are currently not charged any fee by the SNW providers.