Factors and Cluster Analysis

Factors and Cluster Analysis

Student’s Name

Institution

Factors and Cluster analysis

Introduction

Factors analysis is a technique that allows the reduction of a bigger amount of correlated variables to a smaller amount of latent dimensions.The purpose of reflective and formative factors in desertification research is to help students in improving their result analysis skills. The main difference between reflective and formative factors originates from the direction flow of casualty between the latent flow and its indicators. In the reflective factors, constructs of their models become the common effect on all indicators. Any modification that takes place on the latent variable also affects all the items (Romesburg, 2004). In formative factors, the models adopt a different measurement to justify the constructs-items interrelationship as shown in the flow chart below. Here, the model displays a set of dissimilar causes with each cause representing a small portion of the whole construct (Romesburg, 2004).In addition, the other different between the two factors can be shown in the behavior of their models particularly on the impacts of changes made on their indicators. In reflective model, eliminating one or more indicators does not affect latent variable (Romesburg, 2004) On the other hand, changing or removing one indicator will result in diminishing the scale of validity in the formative model.

Similarities between Explanatory Factor Analysis and Confirmatory Factor Analysis

Both exploratory factors analysis and confirmatory factor analysis derive their explanation from a common factor model. The common factor model explains that each observable response is partly influenced by some underlying common factors and partly some unique factors. In common factor model, the force of the link between each factor and each measure fluctuates. They fluctuate in such a way that a given factor can influence some measures in a very powerful way as compared to other factors (Dwyer, Gill & Seetaram, 2012).The two-factor analyzes are carried out by examining the correlations patterns found in between the observed measures. Highly correlated measures, whether negative or positive have the likelihood of being influenced by the similar factors, while measures that are uncorrelated are influenced by different factors.

Differences between Exploratory and Confirmatory Factors Analysis

The differences between these two types of factors analysis can be explained based on the roles they play. Factor analysis always tries to explain the nature of the constructs that influence a group of possible responses. On the other hand, confirmatory factors of analysis are used when testing whether a precise set of constructs is manipulating the outcome in a particular way. The difference is also seen in terms of their objectives. The goals of explanatory factors analysis are to find out the number of similar factors influencing the measures. They can as well as be used in testing the power of relationships existing between the factors and observable measure (Romesburg, 2004). Explanatory factors are performed using seven steps while confirmatory factors are performed using six steps.

Similarities between Exploratory and Confirmatory Cluster Analysis

Exploratory and confirmatory analysis use variables that containing dissimilar measurable levels. The variables exploratory analyses are incompatible with each other hence must be transformed to cluster analysis (Wang, 2009).

Differences between Exploratory and Confirmatory Cluster Analysis

In explanatory cluster analysis, the amount of clusters is not known while the quantity of cluster is known in confirmatory cluster analysis. Therefore, cluster’s number identification is needed in explanatory cluster analysis and the number of clusters does not need to be identified in confirmatory cluster analysis.In explanatory clusters analysis, the cluster characteristics are unknown. Examples of characteristics here may include the center of clusters. On the other hand, the characteristics of confirmatory clusters analysis are partially known (Wang, 2009). This, therefore, means that getting a precise interpretation is not easy and hence there is a need for interpretation in explanatory cluster analysis. Conversely, cluster analysis is easy hence does not call for interpretation. Fit to data in explanatory cluster analysis is always maximized while it may be poor in confirmatory cluster analysis.

Uses and Misuses of Explanatory Factors Analysis and Confirmatory Analysis

Uses

The primary use of factors analysis is to find out if the small quantity of underlying factors can explain the results obtained from multiples tests.Exploratory factor analyses are used in various ways, for instance, when identifying the most important features in classifying a grouped data. They are also used in demonstrating the directions of measurable scale. They also helped in determining clusters of items hanging in questionnaire, and lastly they are used in determining the nature of constructs responses in a particular area (Wang, 2009).Confirmatory factors analyses are on the hand used in determining the validation of models. They also help in testing the relationship between the different models. This type of factors analysis is used in testing the correlation between sets of data. Confirmatory factors analysis helps in matching the capability of different model to account for the similar set of data.Cluster analysis is used to bring together cases instead of variables; therefore, it can be used in population segment. Here, the cluster analysis acts as the only target group. In confirmatory factors analysis, there may be a poor result in a fit to data (Wang, 2009).Confirmatory factors analysis methods are not found in standard software like SSPS that offers limited rudimentary confirmatory analysis. In SSPS, all the initial values must be freed for estimation. This is what limits the usage of confirmatory factors analysis. Here, fixing some parameters is not possible.

Misuses

Most researchers have studied and found that formative and confirmatory factors analysis structural effects of both may sometimes create many problems for the researchers. This problem may arise because of the nature of items that are contained in their models, which always vary with the outcome variable. The reverse is true. This makes researchers to be careful in the type of model they choose for their studies.

Relative Importance of Explorative Factor Analysis and Confirmatory Factor Analysis

Because there are many controversies surrounding the choice of analysis to use for a specific study, it is mostly advisable to assess their advantages and disadvantages before deciding on the one to choice. Selection of a model without previously analyzing the one that suits the better result may result into damage of the result of the study. So it is always important to take into account that the model that fits your study because each model has its degree of importance. In any case one is interested in using Exploratory Factor Analysis and he lacks basic theories underlying the constructs, this model can provide that opportunity. This problem can also be solved using Confirmatory Factor Analysis models.In any case a Confirmatory Factor Analysis has been performed and it provides no significance, it is normally allowed to follow and correct the inconsistency in Confirmatory Factor Analysis. This is done using Exploratory Factor Analysis. This gives an opportunity of testing modifications one wants to make in a new model on a new data.When using different sets of data, it is always reasonable to use an Exploratory Factor Analysis in generating theories pertaining the constructs that are underlying the measures. A Confirmatory Factors Analysis can then follow this. In this case, one is merely fitting the data and not doing the actual theoretical testing of constructs (Romesburg, 2004).This is only possible where the results of Explorative Factor Analysis are put directly in Confirmatory Factor Analysis on the similar data. The well-accepted procedure here is to perform half of Exploratory Factor Analysis on a data. After that, it is followed by testing the generality of the extracted factors with a Confirmatory Factor Analysis on the other half of the data.Both Exploratory Cluster Analysis and Confirmatory Cluster Analysis are useful because they can help in grouping the clustered data instead of variables (Dwyer, Gill & Seetaram, 2012). This enables the segmentation of the data being analyzed.The interventions needed in both cluster analysis seems to be very clear. This is possible because each sub-cluster will only be interpreted on its own as opposed to the whole cluster. As a result, it makes general intervention very simple because adjustment is done on individual clusters.

What Factor Analysis vs. Clusters Analysis Does and cannot Do

Factors analysis can be used in the interpretation of a set of items in a questionnaire. In a case where the researcher has got different categories of models contained in the questionnaire, he can use factors analysis in testing different hypotheses contained in those models (Dwyer, Gill & Seetaram, 2012). This is made possible because confirmatory factor analysis provides an alternative hypothesis that ensures there is a match of items in the models and confirms that these categories of models matches with the variance used in the research. Alternatively, factors analysis cannot be used in SPSS systems. Factors analysis of numerous small studies cannot be used to represent the result of a bid single one.Relative Factor Analysis can be used to derive a theory of a contract underlying the particular measures. Confirmatory Factor Analysis can follow this. For this to be obtained, the test must be done using different sets of data. In case someone get a significant fit while performing a Confirmatory Factor Analysis, it is advisable to follow the inconsistency between the data using Reflective Factor Analysis. In general, factor analysis is used in exploring the sets of data that help in identifying hypothesis. They are also to reduce many variables to a small and a manageable manner (Dwyer, Gill & Seetaram, 2012).Cluster analysis is usually used after factor analysis and is used in identifying the groupings. It depends on discriminate analysis to confirm if the groupings are arranged statistically between the models and also used in identifying the significance of variable between the groups. Cluster analysis cannot be used to test the goodness of fit of models or it cannot be used in knowing the important of a model.

References

Dwyer, L., Gill, A., & Seetaram, N. (2012). Handbook of research methods in tourism: Quantitative and qualitative approaches. Cheltenham: Edward Elgar.

Romesburg, H. C. (2004). Cluster analysis for researchers. North Carolina: Lulu Press.

Wang, Y. (2009). Statistical techniques for network security: Modern statistically-based intrusion detection and protection. Hershey: Information Science Reference.