This research project pursues the qualitative research approach and applies different methods. The two research methods, induction and deduction, appear to be extremely interesting.
Scientific findings will be gained by induction for one part of this research work and by deduction for the other part. The actual implementations will be described in Sections 3.2.1 to 3.2.4. [73]
Overall, Structural Equation Modeling appears highly suitable for the visualization of the priority list of the individual design elements. SEM is a very precise statistical model in which several values can be compared by either deduction or induction.
3.1.1 Induction
Induction represents a bottom-up method where the findings of one particular participant are kept hold of and combined with each other. It is possible to combine several data and individual decisions. [73]
Figure 18
Method of induction (Bottom Up) [73]
This model (Figure 18) is based on the expositions stated by Balzert, Schröder and Schäfer and was adapted by the author. [73] It shows a possible presentation of the results.
Individual participant n Individual participant m Comparison between
This approach is likewise considered for all the other test persons in each possible com-bination. Are there any specific patterns revealing similar decisions, for example, with test persons of the same gender and the selection of the better coloring concept of the reference portal?
At this time, there is no prognosis on the result to be expected as possible results are absolutely open and a hypothesis cannot be proposed.
3.1.2 Deduction
Deduction is a theoretically oriented research approach that requires conditions that have to be specified and have to be determined as directly designed influencing factors. [74]
Several “IF components” are specified from which the effective quality components
“THEN components” develop.
Figure 19
Method of deduction (Top down) [73]
Figure 19 shows a different possible presentation of the results for the evaluation and development of a new set of rules. Both methods will be incorporated and used for the evaluation and assessment of the results.
This method is excellently qualified for scientific research with the presentation of hypothesis, prognosis. The objective may be to confirm theories. [74]
Confirmation or disproof
of the hypothesis Confirmation or disproof of the hypothesis
In the center there are the relations of the discoveries as an explorative demonstration of the cause–effect relations that, in the course of this theory development, lead to the hypothesis. [74]
The examination setup applied here [74] is to lead to knowledge production of the hypothesis:
The participants find the usability of Portal B more user-friendly and more struc-tured than Portal A.
This hypothesis may be confirmed or disproved. It is possible too that some individual results will support or contradict the results found in some areas of the hypothesis. This will be incorporated as a top-down method.
3.1.3 Structural Equation Modeling (SEM) [75]
The results to be assessed and measured have to be presented in a construct of theoretical and intellectual nature. This is the way to enable the results of the research work to be measured. This process is called operationalization.
To visualize the construct for operationalization a model needs to be developed. In this research project, the presentation is based on the Structural Equation Model.
Therefore, it is necessary to define the following elements as parts of the Structural Equa-tion Model [76]:
– Indicator (Item)
The variables concerned are single observed ones that blend into the factor. The measurable results are incorporated and condensed. These are a part of the latent variable. [76] (δ1–δ7)
Latent variable (Factor)
The factors blend into the observed variables. They are gained from the indicators.
The individual factors blend into the measurement model as variables. These fac-tors may be either independently latent (exogenous) or dependently latent (endogenous). Frequently, there is no measurement procedure that imple-ments an objective factor analysis.[78] (
ξ
1–ξ
3)Measurement model
In this model the results from the latent variables blend in. Connections are mo-deled between the indicators and the latent variables. [77] At this stage a vague statement is made as an interpretation of the covariance (which random variables go together with which other random variables). [79] (η1
)
Figure 20
Path diagram for SEM [80]
Figure 20 shows the presentation of the results in the SEM.
The latent variables of the design factors, structural factors as well as advertising-related factors constitute a nonstandardized measure of association. Stochastics call the latter co-variance [78] and describes the monotonic link of two random variables with a joint pro- bability distribution. [79]
λn designates the relations and the influence of the latent exogenous variables on the respective factor. Within the factors, the variables obtain a distinction of significance for the user.
γn shows the influence of the respective factor onto the usability quality of Healthcare portals.
Because of the calculations, the weighting of the indicators may represent a comparable measure. The assessment of the results may be used for the hypothesis-based research work.
3.1.4 Alternative statistical data analyses
The analysis and evaluation of the data obtained need to be based on an appropriate statistical procedure. In the process, it must be observed that bivariate or multivariate analysis methods are employed. Because both the inductive and deductive research methods are applied, the dependencies between the participants and the chosen variables or the interdependencies (mutual dependencies) need to be investigated. [81]
3.1.4.1Causal–analytical and descriptive research design
The causal–analytical and descriptive research design is suitable for the hypothesis- oriented research methods. In this case, in advance, there is already a model-like idea existing about the confirmation or refutation of the hypothesis. It is necessary to work out the assumed dependencies between the indicators (variables) and the factors and to pre-dict them. The descriptive method must be added to the descriptive statistics. [81]
3.1.4.2Explorative research design
The research part of the combination of assumed dependencies as well as personal data of the test persons may be characterized by the explorative research design. Dependent and independent variables cannot be differentiated and there is no way to make a state-ment on the possible analysis result in advance. Despite the problematic statestate-ment of a prognosis, the direction for analysis should be specified. This procedure is counted as belonging to the evaluating statistics. [81]
The analysis methods described are used in each of the evaluation areas. In doing so, the dependencies as well as the interdependencies, are analyzed and possible patterns of the cluster of people are determined. These represent the basis for new findings to be derived.