Fortune 500 Corporate Blogs: Research Background and Methodology

The Premise

My thesis was born from the idea (right or wrong) that writing blogs leads to good things and that big companies should publish blogs to communicate and connect with customers and other stakeholders. I wanted to measure the effects of corporate blogs using the sentiment of posts in memes that resulted from corporate blog posts.

Hypotheses

The hypotheses tested were as follows:

  • The overall sentiment found in corporate blog memes would be bias towards positive sentiment. It was.
  • The sentiment of the first entries in memes would show a positive correlation with the average sentiment of later posts. It didn’t.
  • The sentiment of the most connected posts (those containing the most incoming links) would show a positive correlation with the average sentiment of the other posts. There was a moderate correlation, but its significance was questionable.

Identifying Corporate Blogs

Blogs were identified by looking at all Fortune 500 company websites and performing consistent Google searches to capture any affiliated blogs, sponsored sites or otherwise related projects involving blogs. Selection criteria was developed in order to create consistent rules around which blogs were included as objectively as possible.

Collecting Blog Posts and Memes

All corporate blog entries published in July and August 2008 were recorded and run through Blogpulse in order to find other blog entries that linked back to them. In other words, corporate blog posts were used as the starting point for memes.

There are undoubtedly limitations to this method of viewing, finding and recording blog memes and even more specifically related to the tools used and the way data was recorded. Some of the overall limitations of the study are posted and were analyzed in greater detail in my thesis paper.

As I’ve tried to say in many places on this blog, there are undoubtedly more limitations than it’s relevant or possible to mention. As a result, all statistical evidence presented on this blog should be viewed with the limitations in mind.

Assigning Sentiment

All consumer blog entries were machine-processed for sentiment using Sentimine, a social media sentiment tool created by the Parnassus Group. As blog posts were topic-agnostic, a custom configuration of Sentimine was used to derive sentiment at the document level rather than specific entities mentioned in posts such as people or products.

Related Pages