Group members
- Ma Jian Ting 33306101
2) Nur Shafeena Binte Samaruddeen (33246804)
3) Desmond Tan Soon Hong 33092919
Objective:
The objective of this research was to do a cluster analysis on the data which was gathered via facebook comments. The data was to be coded according to a list of different behavioural questions, a total of nine behavioural questions and then further differentiated to the market segments accordingly. After which, the aim was to recommend specific product and services to the people that were from these different segments.
Method
Coding Standard
The research was conducted based on a total of 304 comments that were retrieved from 10 random articles on The New York Times facebook page. We then coded the data using a coding scheme and recorded it into an excel file. There were nine behavioural questions as shown below which were used to judgmentally classify the data:
| The New York Times |
Scales (1 to 7) |
| 1.Clarity of ideas | 1 clear ideas ………………………………………..…… 7 confused ideas |
| 2. Level of emotion | 1 low emotion ………………………………….……… 7 highly emotional |
| 3. Level of objectivity | 1 very objective ………………………………………… 7 very subjective |
| 4. Past perspective | 1 a lot of focus on the past ………………………… 7 no focus on the past |
| 5. Now perspective | 1 focus on now ………………………………………… 7 no focus on now |
| 6. Future perspective | 1 focus on the future ………………………………… 7 no focus on the future |
| 7. Level focus on personalities | 1 strong focus on people …………………… ……… 7 no focus on people |
| 8. Level of criticism of corporates/business | 1 highly critical of business ………………………… 7 no criticism of business |
| 9. Level of criticism of the government/public service departments | 1 highly critical of governments/public service …………… 7 no criticism of governments/public service |
For the first behavioural question which is on the clarity of ideas, our coding standard was based on the facts and recommendations provided in the comments. If none of these two were mentioned at all, we gave a rating of confused ideas which was a 6-7. If either one of it was mentioned, we rated it on an average clarity of 4-5, and if both points were strongly elaborated on we gave the comment a rating of clear ideas ranging between 1-3.
The Second behavioral question was on the level of emotion which we coded based on the expressiveness in the comments, typing style of the people for example, if they used punctuations such as exclamation marks to show their emotions such as excitement and anger. In addition, we looked out for emoji’s to assist us in our coding standard as it is one way people tend to show their emotions these days. Those comments that included the above mentioned styles were given a high emotion rating between 4-7 and if the comments did not have any of such styles we rated them as low level of emotion ranging from 1-3.
The third behavioral question was on the level objectivity in the data, for which our coding standards were based on if the comments given were neutral, bias or unbiased. If comments were relied on personal experiences, we deemed them as subjective and hence were given a rating ranging between 5-7. Those comments that were neutral, we gave it a rating of 4. Then comments that were unbiased and had not been influenced by opinions of others or personal feelings were rated between 1-3.
The fourth behavioral question was on the past perspective which we coded based on the comments which were specifically on history, for instance some comments mentioned about the previous government that ran the country etc. If any of these comments contained this, we given the rating between 1-3 which mostly focus on the past and those unlikely to contain the above, were given the rating between 4-7 as people were not mentioning past affairs.
The fifth behavioral question is with regards to now or current affairs. The coding for this was done based on if comments are discussed or mentioned present or current related affairs, for example comments that elaborated more on present related topics were given a rating between 1-3 and for those there were irrelevant to this standard were given a rating of 4-7 which has no focus on “now”.
The sixth behavioral question based on future perspective. The coding for this was done based on planning, suggestions, concern and recommendation that was found in the comments. Basically, we rated comments that showed some form of suggestion, recommendations and concern for their future between the range 1-3 and those comments that did not mentioned any of the above were rated between 4-7.
The seventh behavioral question was regards to level focus on personalities. We coded this questions based on factors such as characteristics of people, welfare of people and general discussion with regards to “people”. Comments as such, were given a rating of 1-3 as these comments had strong focus on people and their personality traits. However, comments without any of the above factors were given a rating between 4-7.
The eighth and ninth behavioral questions is regards to the level of criticism of corporates or business and government or public service department. We coded these questions based on comments that criticized, condemned, were spreading hatred and where filled with sarcasm. Comments as suchs, were given a rating between 1-3 and as for comments did not have any of the above were given a rating between 4-7.
Coding of the data was done by two people, one acted as the maker; the person who coded the data and the other as the checker; the person who verified the coding. This was to ensure that we got coding done as accurately as possible. It is always safer to have more than one person code the data, this method was also adopted and proven by Jennifer Fereday and Eimear Muir-Cochrane who used the assistance of their manager to help code the data and the results were then compared (Fereday & Cochrane, 2006).
Cluster Analysis – Two step cluster
After all the data had been coded, we then proceeded to perform the cluster analysis which was done using the SPSS system. Cluster analysis is a technique of classification for forming comparable groups within the data sets (Fred H, David C, 1987). This is a technique that has been very well applied in different disciplines for its partitioning ability. For an instance, we can relate it to Bhatnagar and Ghose who applied a latent class modeling approach to segment Web shoppers based on demographics and benefit sought (Bhatnagar, Ghose, 2004). Therefore, this method was considered suitable in forming groups depending on the similarity of their attitudinal variable and demographic. We selected a segmentation method that incorporated two-step analysis, which is a statical program in SPSS 19.0, this is suggested as appropriate in clustering huge data sets that has a mixture of attributes (Norusis, 2003). This technique is based on a distance measure that allows data with continuous and categorical attributes to be clustered. This is taken from a probabilistic model where the distance between two clusters is equivalent to the decrease in log-likelihood function due to a result of merging (Chiu, Fang, Chen, Wang, Jeris, 2001).
This research uses data collected from facebook comments and is coded according to a specific coding standard as we had mentioned above. From our analysis we got a total of four clusters out of which we had selected three clusters; Uncertainty, Future Thinkers and Here and Now that we had further segmented. All three clusters demonstrate different behavioural habits in terms of the nine behavioural related questions that were coded. Each of the cluster showed both similarities and difference in opinions. For example, all three clusters seem to criticize the government or public services, however they also showed difference amongst themselves when it came to their level of emotions as well as on the topics related to past, present and the future.
Limitations:
Sample size is too small. (304 comments only)
Based on our research objectives, a larger sample size could have generated more accurate results. A large sample size for quantitative studies is critical and important as compared to qualitative studies. Some of the articles which we found on New York Times have limited comments which we can’t make use of. A lack of data is a significant obstacle in our research. Time is a limitation for us as we have a deadline to meet and this research is highly time consuming process.
Segments development:
From the SPSS analysis we gathered a total of 4 clusters, out of which we have segmented 3 clusters. Cluster number 1 is called Uncertainty, the analysis shows that people from this cluster did not provide clear ideas on the subject. They were also highly emotional and were being subjective by basing their comments on personal experiences and views. People from the Uncertainty cluster were more focused on the past, for example they only mentioned facts relating to history and were least focused on the present and future. In addition, this cluster did not seem to have any focus on the people, however, they were highly critical of the government or public service.
We named cluster number 2 as the Future Thinkers due to the SPSS analysis showing that the people had low focus on the past and instead were highly focused on the present moments as well as the future. This was visible in their comments where they provided suggestions, were focused and concerned with their future affairs. They have a strong focus on people and their personalities compared to the Uncertainty cluster which had zero focus on people. However, it seems that this cluster does not have a great clarity of ideas and also held high levels of emotions. To add on, the analysis shows that the Future Thinkers are also very critical of the government or public service based on their comments.
Lastly, our third cluster which was determined based on the SPSS analysis results is called Here and Now cluster. This cluster is very confused in their ideas, they have high levels of emotions and are extremely subjective. People in the Here and Now cluster showed no attention on the past and the future, but they were immensely focused on the present affairs only. In view of this analysis, this cluster has also been highly critical of the government or public service as well.
Product and services recommended for segments:
Uncertainty – suggest something to deal with high level of emotion, how to gain clarity of ideas, ways to have a broader mindset and to understand the government better. How ? example, programs? Talks? > back this up with reference.
Future thinkers- Something to gain clarity of ideas, they should spread knowledge speak up provide the service.
Here and now- Gain knowledge of the past, focus on the future> go for talks. Watch the news, something with regards to the government.
|
Initial Cluster Centers |
||||
|
Cluster |
||||
|
1 |
2 |
3 |
4 |
|
| Q1 |
6 |
3 |
7 |
6 |
| Q2 |
7 |
7 |
7 |
5 |
| Q3 |
2 |
1 |
7 |
7 |
| Q4 |
1 |
5 |
5 |
7 |
| Q5 |
7 |
1 |
1 |
5 |
| Q6 |
7 |
1 |
7 |
1 |
| Q7 |
7 |
1 |
7 |
5 |
| Q8 |
7 |
5 |
5 |
7 |
| Q9 |
1 |
1 |
1 |
1 |
Appendices: SPSS data
Referencing
http://journals.sagepub.com/doi/pdf/10.1177/160940690600500107
https://www-sciencedirect-com.libproxy.murdoch.edu.au/science/article/pii/S0148296302003570
Prentice-Hall, Upper Saddle River, NJ (2003)
https://dl.acm.org/citation.cfm?doid=502512.502549
The post The objective of this research was to do a cluster analysis appeared first on My Assignment Tutor.
-
- Assignment status: Resolved by our Writing Team
- Source@PrimeWritersBay.com
GET THIS PAPER COMPLETED FOR YOU FROM THE WRITING EXPERTS CLICK HERE TO ORDER 100% ORIGINAL PAPERS AT PrimeWritersBay.com
Comments
Post a Comment