Open Source Intelligence Analysis – Demographics

Getting good demographics can help you to quickly understand the context of messages that circulate through different websites. The easiest method for large websites is to look them up on http://www.quantcast.com/

For instance, we an obvious pattern that hispanics and blacks tend to visit conspiracy websites much more often than whites. We also see that many of the sites tend to have older visitors with higher incomes. The exception are data driven websites like Wikileaks, which tend to be lower income but highly educated viewers who are mostly white or asian.

If you can find facebook groups for websites like this, you can cross-check some of the basic information by looking at user photos, names, and ages (keep in mind that facebook users tend to be younger than average):

To get a quick introduction to the character of a website, simply do an imagesearch of it on google, e.g.: site:http://forum.prisonplanet.com/index.php

Search through the websites looking for mentions of states and/or cities using google, e.g.: texas site:infowars.com (don’t add a space between the search command and the website, so use site:websitehere.com). Look for introduction threads or user profiles that list locations. Twitter accounts can also assist in this process.

With this information you can cross-correlate the cities members live in to get an idea of their general make-up, and how it compares to other demographic sources.

If there are a lot of unique images on the website, use google’s image search function to look around for other websites with the same images, which will expand your understanding of the psychographics of the users by finding similar sites and images.

For more google search ideas, look at “How to solve impossible problems: Daniel Russell’s awesome Google search technique”:

http://www.johntedesco.net/blog/2012/06/21/how-to-solve-impossible-problems-daniel-russells-awesome-google-search-techniques/

If you want to map out keywords and connections, use a graph similar to this:

http://www.touchgraph.com/seo

A basic search gives us something like this:

Which shows us that we can also harvest data from youtube and amazon, as well as the smaller linked websites.

Now we have the basic demographics, we look for commonalities. Search through abstracts of psychology journals using pubmed.gov or google scholar, looking for keywords related to conspiracy theories, demographic information and psychology journals.

We end up with some curious things like this:

http://heb.sagepub.com/content/32/4/474.short

This article examines the endorsement of conspiracy beliefs about birth control (e.g., the belief that birth control is a form of Black genocide) and their association with contraceptive attitudes and behavior among African Americans. The authors conducted a telephone survey with a random sample of 500 African Americans (aged 15-44). Many respondents endorsed birth control conspiracy beliefs, including conspiracy beliefs about Black genocide and the safety of contraceptive methods. Stronger conspiracy beliefs predicted more negative attitudes toward contraceptives. In addition, men with stronger contraceptive safety conspiracy beliefs were less likely to be currently using any birth control. Among current birth control users, women with stronger contraceptive safety conspiracy beliefs were less likely to be using contraceptive methods that must be obtained from a health care provider. Results suggest that conspiracy beliefs are a barrier to pregnancy prevention. Findings point to the need for addressing conspiracy beliefs in public health practice.

And:

http://onlinelibrary.wiley.com/doi/10.1111/0162-895X.00160/abstract

This study used canonical correlation to examine the relationship of 11 individual difference variables to two measures of beliefs in conspiracies. Undergraduates were administered a questionnaire that included these two measures (beliefs in specific conspiracies and attitudes toward the existence of conspiracies) and scales assessing the 11 variables. High levels of anomie, authoritarianism, and powerlessness, along with a low level of self-esteem, were related to beliefs in specific conspiracies, whereas high levels of external locus of control and hostility, along with a low level of trust, were related to attitudes toward the existence of conspiracies in general. These findings support the idea that beliefs in conspiracies are related to feelings of alienation, powerlessness, hostility, and being disadvantaged. There was no support for the idea that people believe in conspiracies because they provide simplified explanations of complex events.

And:

http://www.jstor.org/discover/10.2307/3791630?uid=2&uid=4&sid=21101379127527

From this information we can break them into traditional psychographics using stock models:

Now you can create a database that can be used for advanced analytic operations, using R, excel, SAS or a programming language like Python. R tends to be more effective for smaller sets less than 2GB because of it’s memory usage, but it has nearly all statistical functions anyone has thought to use which makes it very useful for experimental projects. SAS is commercial software that is mainly effective for large data sets. Excel is a decent entry level solution. Python is not quite as flexible as R yet, but it’s modules are improving and it can be interfaced with R.