K-mean is used for numerical value only. I don't know the DIDS scale but if outputs are qualitative like a survey. You will need to convert it into numeric variables (perhaps percentages). If you obtain percentages, you won't need to use z scores. However, to avoid multicolineraity effect when you will run your clustering analysis, you will need to introduce n variables minus 1.

https://www.researchgate.net/post/For_a_K-mean_cluster_analysis_when_variables_are_percentages_have_we_to_use_n-1_variables

Thank you for the input. It is a quantitative study where we need to do hierachal cluster analysis and then K means

Hello dear Rabeeya Farooq ,

Cluster analysis is a statistical technique used to identify groups, or clusters, within a dataset based on similarities between observations. In the context of identity statuses using the Dimensions of Identity Development Scale (DIDS) by Koen Luyckx, cluster analysis can help identify distinct patterns or profiles of identity development among individuals.

Here are the steps to perform cluster analysis and then use K-means clustering to create identity statuses from DIDS scores:

1. Data Preparation: Collect data using the DIDS scale, which typically includes responses to various items related to identity development. Ensure that the data is cleaned and formatted properly before proceeding with the analysis.

2. Standardization: Before conducting cluster analysis, it's essential to standardize the variables (DIDS scores) to ensure that they are on the same scale. Standardization involves converting the raw scores into z-scores, which represent the number of standard deviations away from the mean. This step is crucial because it prevents variables with larger scales from dominating the clustering process.

3. Cluster Analysis: Once the data is standardized, you can perform cluster analysis using a method such as K-means clustering. K-means clustering aims to partition the observations into a pre-specified number of clusters (identity statuses) based on the similarity of their DIDS scores. The algorithm iteratively assigns each observation to the nearest cluster centroid (mean) and updates the centroids until convergence.

4. Determining the Number of Clusters**: Before applying K-means, it's essential to determine the optimal number of clusters. This can be done using techniques such as the elbow method, silhouette method, or hierarchical clustering. These methods help identify the number of clusters that best capture the underlying structure of the data.

5. Interpreting the Clusters: Once the clusters are generated, it's crucial to interpret them to understand the distinct identity statuses they represent. This involves examining the mean DIDS scores within each cluster and identifying the key characteristics or patterns associated with each status.

6. Validation and Interpretation: After identifying the clusters, it's essential to validate them using external criteria or theoretical frameworks related to identity development. This ensures that the clusters are meaningful and interpretable in the context of identity theory.

7. Reporting and Publication: Finally, document the results of the cluster analysis, including the method used, the number of clusters identified, and the characteristics of each cluster. Consider publishing your findings in peer-reviewed journals to contribute to the scientific understanding of identity development.

Good luck and Happy clustering !

Samawel JABALLI

can you tell me how the validation part is done?