5.2 The questionnaire and the structural model
5.2.4 Generalizability of the findings
exceed the value of their respective VE values, which are the values on the diagonal, written in bold. According to Janssens et al. (2008) this means that this study holds discriminant validity:
Table 9. Discriminant validity
Up To Date Charity Sub Norm
Hygiene Less Religious Focus
Up To Date 0,5744
Charity 0,073441 0,7774
Sub Norm Motivational 0,075625 0,009604 0,5648 Sub Norm Hygiene 0,093025 0,002025 0,0112 0,499806489
Less Religious Focus 0,025281 0,04 0,1142 0,044944 0,500198419
Based on the methods from Janssens et al. (2008) it can be stated that the results provided by the
structural model, can be considered as valid, when focusing on the construct validity.
The target was to reach people from Copenhagen, since this is the diocese that is the focus point for this thesis. To avoid misunderstandings, the different districts of the diocese were used as options to make it simple for people to fill in where they live. Yet, to increase the numbers of respondents, data from people living in Greater Copenhagen were included in the analysis as well.
This means that the people living in the diocese of Helsingør are included to the study, but since that diocese has the second lowest membership rate in Denmark, only surpassed by Copenhagen, it seemed appropriate to include these people as well27
When looking at the distribution of the respondents, measured by age group, the findings of this study do not represent each age group equally. Among the 30 people, who distributed the questionnaire in their network, at least one representative from each of the five age groups was included, in order for the questionnaire to be spread into each age group. Still, the majority of respondents were from the age group of 20-25 and 26-30. These held 45% and 43% of the respondents from the entire respondent group respectively. This means that the findings primarily are reflecting the opinion held by these two age groups. The level of representativeness for the study, when it comes to explaining the general opinion held by people from 18-20 and from 30-40 is therefore low, and so is the generalizability.
The distribution among the genders showed that 66 % of the respondents were females. It is usual that women find greater interest in filling out questionnaires than men do. Yet, it might also cause the results to represent the general opinion held by women of the target age groups, than the opinion of the men.
Further, the educational level and the daily occupation of the respondents were also two aspects that seemed to lower the generalizability level of the study.
Since the focus groups only consisted of highly educated people and students, it felt necessary to include an item that could illustrate the level of education of the respondents and an item describing their current situation. It was found that 66% of the respondents either have a bachelor’s – or master’s degree. Further, nearly 64% of the respondents are students. This means that the majority of the respondents from the questionnaire resembles with the respondents from the focus group. The findings of the study are therefore primarily generalizable for people who hold long, higher educational degrees or students in general.
27 http://miliki.dk/kirke/statistik/folkekirkens-medlemstal/ - visited 04.04.2013
All of the above mentioned demographical factors have a collective impact on the generalizability level of this study, and it is now clear that the findings are not perfectly generalizable for the entire target group of light users from age 18-40. However, according to Kerlinger & Lee (2000) the question a researcher should ask himself, is not whether the findings are generalizable or not. It is rather a question of: “How much can we generalize the results of the study?” (Kerlinger & Lee, 2000, 474 – (underline is to show that those words are written in italic type, in their book)).
Therefore, a solution, based on the findings from this study, is most likely to appeal to female light users from the age of 20-30, who either are students or have completed a long and higher education. Further, the findings are based on the responses from Evangelical-Lutheran church members. This should therefore also be kept in mind, if the findings are applied to other religious settings.
5.2.5 Hypotheses testing
Table 10 and 11 are showing the AMOS output presenting the regression weights and the confidence intervals for the three dependent variables. The results from these two tables are used to test the different kinds of hypotheses:
1. How much impact each individual variable has on the dependent variable it is connected with.
2. How much better each individual variable is at explaining the dependent variable it is connected with, when comparing it to the variables explaining that same dependent variable.
The data used here, are the unstandardized estimates.
Table 10. Unstandardized regression weights
Table. 11. Confidence intervals
Construct influencing others Confidence interval HygieneFactors <-- LessReligiousFocus
1,031 - (2*0,116) 0,799
1,031 + (2*0,116) 1,263
HygieneFactors <-- SubNormHygiene
0,1 - (2*0,08) -0,06
MotivationalFactors <-- SubNormMotivational
0,032 - (2*0,054) -0,076
0,032 + (2*0,054) 0,14
MotivationalFactors <-- Charity
0,304 - (2*0,06) 0,184
0,304 + (2*0,06) 0,424
MotivationalFactors <-- UpToDate
0,788 - (2*0,099) 0,59
0,788 + (2*0,099) 0,986
LikelyToGoToChurch <-- MotivationalFactors
0,423 - (2*0,091) 0,241
0,423 + (2*0,091) 0,605
LikelyToGoToChurch <-- HygieneFactors
(-)0,161 - (2*0,054) -0,269
(-) 0,161 + (2*0,054) -0,053
Hypothesis Estimate S.E. C.R. P
HygeineFactors <--- LessReligiousFocus 1,031 ,116 8,916 ***
HygeineFactors <--- SubNormHygeine ,100 ,080 1,258 ,208
MotivationalFactors <--- SubNormMotivational ,032 ,054 ,588 ,557
MotivationalFactors <--- Charity ,304 ,060 5,035 ***
MotivationalFactors <--- UpToDate ,788 ,099 7,969 ***
LikelyToGoToChurch <--- MotivationalFactors ,423 ,091 4,651 ***
LikelyToGoToChurch <--- HygeineFactors -,161 ,054 -2,992 ,003
When looking at the independent variables explaining “Motivational Factors” in Table 10, it is clear that both “Charity” and “Up To Date” has a significant impact on “Motivational Factors” (P <
0,001). This means that H1 and H2 are confirmed. “Sub Norm Motivational”, which was the last independent variable that was tested to have a connection with “Motivational Factors”, did not have a significant influence (P = 0,557). Therefore H3 is rejected.
According to the confidence intervals in Table 11, it can be seen that none of the three constructs, explaining “Motivational Factors”, overlap each other. Further, the values of “Up To Date” are greater than the values of “Charity”. Therefore “Up To Date” explains “Motivational Factors”
significantly better than “Charity” does, and H4 can be confirmed. Further, the values of “Charity”
are greater than the values of “Sub Norm Motivational”. This means that “Charity” explains
“Motivational Factors” significantly better than “Sub Norm Motivational” does. H5 can therefore also be confirmed.
When it comes to the construct of “Hygiene Factors” and the influence that “Sub Norm Hygiene”
and “Less Religious Focus” have on this dependent variable, only the latter proved to have a significant influence. Table 10 shows P-values P = 0,208 and P < 0,001 respectively, and therefore H6 is rejected and H7 is confirmed.
When looking at the confidence intervals for these two constructs in Table 11, they do not overlap each other. Further the values of “Less Religious Focus” is greater that the values of “Sub Norm Hygiene”. Therefore “Less Religious Focus” explains “Hygiene Factors” significantly better than
“Sub Norm Hygiene” does. H8 is therefore confirmed.
The two hypotheses regarding the influence that “Motivational Factors” and “Hygiene Factors”
had on “Likely To Go To Church” are both significant. With P-values in Table 10 on P < 0,001 and P = 0,003 respectively H9 and H10 are confirmed.
When looking at the confidence intervals for these two constructs in Table 11, they do not overlap. Further, the values for “Motivational Factors” are greater than the values for “Hygiene Factors”. This means that “Motivational Factors” explains “Likely To Go To Church” significantly better than “Hygiene Factors” does. H11 is therefore confirmed.
Table 12 illustrates the confirmation and rejection of the hypotheses.
Table 12. Hypotheses results