The tools and methods used
In the making of and gathering of data there have been used several different kinds of tools and methods. And other tools have been used to visually map our controversy, to show different aspects of it. A brief explanation is provided in the figure below, and a more elaborate one in the text paragraphs below the picture.
Google overview
We kicked this project off with some rather basic google searches relating to our subject, in order to get at least some preliminary knowledge of the subject and controversy and to find some actors from which we could begin the Hyperlink crawl.
for the more general understanding we googled simple terms and concepts such as: "Intelligence" "IQ" "what is IQ" "Flynn effect" "g factor" While our searches for the Hyphe in chronological order were: “IQ AND “depends on”” “IQ AND dependency” “IQ AND genetics” “IQ AND environment” “IQ AND race” “IQ AND upbringing” To avoid the google ads we manualy imported the url of the first couple of google pages to the Hyphe tool. |
Hyphe
Hyphe is a webcrawler used to make a hyperlink map. A hyperlink map shows which websites link to what websites via aptly named hyperlinks. This is important, because linking to another site shows an acknowledgement of that site. By then making a network of hyperlinking in our controversy, we’re able to see who the main actors are, measured by who people acknowledge through hyperlinks. in pratice what we have done is that we have started on google to search for our controversy. We have started out searching for what IQ depends on as written about in the google overview header.
We then started crawling the relevant websites. Crawling is the main function in Hyphe, where it takes all the links from a given website and store them for future use. We have crawled with the setting “depth of 1” meaning that the tool takes the links on the frontpage, and then take those sites one layer deeper and store them as well, a process called scraping. We do however have some exceptions to this depth of one crawling. As crawling certain “hub” sites with a depth of 1 can overload the tool with links. We therefore limited certain sites to a "depth of 0", where the tool only stores the links on the given frontpage. After a while going through new websites found via the Hyphe tool and crawling the relevant ones, we start to see a network, where sites link to each other. We have chosen to only look at sites that have been linked to 2 or more times by the network actors. We have done this because of two things. First these have a lower probability to be ads, or different unique things to one website (like linking to the authors personal facebook page, which can be relevant, but typically is nothing but advertisement for a specific group). The other, perhaps better, argument to be made is: Thoes sites that have fewer than 2 references, are not seen as a relevant actors by the network because they don’t get acknowledged by it. Even if these sites have good arguments that could help solve the controversy, it is of the time of mapping irrelevant if these actors aren’t acknowledge. Once we start to gain the same websites and ads, or when the site that show up are wholly apart from the topic or controversy, only then will the network seem to be "emptied". This doesn’t however show that the network of the controversy is truly done, it just shows that our initial keywords have played out their use. Through the searches we have made it is possible to go on using other words, and gain a different network. On the top of this site you can find the searches we have made chronologically. The evolution of the searches are done from what we found on the google overview sites, and what words the appearing actors used in their working in the debate, here for example genetics. With the network in hand we started to tag its sites, to be able to distinguish different layers and aspects of our hyperlink map. Our tags can be found on the bottom of the page. Our tags have been made on the account of the actors, and what words they use. |
Lippmannian device
Through the hyperlink network we identified 5 actors of particular importance, these we made a wordcloud analysis of, by using the software Lippmannian device.
Lippmannian device searches the sites we have deemed relevant. It searches them one at a time for a list of words we provide. These words have been obtained by looking at the sites from the hyphe search, and finding defining or repeated words as for instance the word “race” which is occurring on almost all sites. After a thorough search the following list of words is obtained (“” indicating whole expression searches for): black white asian hispanic race nazi facism “political correctness” statist immigration pseudo semit iq racism school flynn fact realism intelligence expertise anthropology fallacy genetics brainwashing conspiracy rich elite Lynn nutrition parent families “anti science” Watson education academic socio twin This list is then used to make a wordcloud through Lippmannian device and Wordle. Lippmannian device scrapes google, for hits on the words on the sites we have provided. After this is done, we gain a frequency count in which the words is used, and use this to make a wordcloud in the program Wordle. The wordcloud accounts for the frequency, where the words that have been used the most are the biggest, and vice versa. |
scopus
Scopus is a large, searchable database for academic texts with the additional function to download parts and partial overviews of texts based on search criteria. for instance we searched for articles including “IQ AND Cause OR dependency OR “depends on”” in either the name of the given article, keywords or somewhere in its abstracts.
The capital words is known as boolean operators, and are ways to specify a search: using AND narrows the search to only include what is on the opposite side of the AND and using OR allows the words on the opposite sides of the operator to be used interchangeably. We further refined the search to texts made after 1999 to show the buildup of the controversy over time. But still keeping the search to what is relevant for the current debate climate. With this search we got 1168 results. divided into the following subject areas A quick glance at the subject list holds shows some surprising results, what does engineering have to do with our IQ controversy. The answer is nothing, as it so happens they are all fields that use the same abbreviation IQ, but for something else entirely. Engineering seems to appear because of mentions of an ID-IQ service commonly used in that field.
To avoid mapping an unrelated field we refined our Scopus search further by excluding engineering tags for the archive.
With this small limit we downloaded the the archive as a CSV file, asking the program to include author keywords for us. Author keywords are words the author of the texts finds relevant to their publishing, which we in turn will use for categorizing the articles clustered around a given keyword. Furthermore including the reference list of the articles allows the creation of a map of their mutual references in the visualisation tool Gephi. To directly import the CSV file to gephi will however not show the interconnectedness of articles and references. But running it through the medialab tool “refrencescape” will connect these and download a new file, that will let us move on to the next tool Gephi. |
GEPHI
The visualization tool Gephi is used to illustrate both the Hyphe and the Scopus network, but not at the same time. Gephi allows us to illustrate the networks by different tags and layers, and in different ways, by varying size and color of the actors. For example it allows us to see both genetics and environment and where they place themselves in the network. It also allows us to use an algorithm called “ForceAtlas 2” which provide a gravitational field, so that only the nodes who “stick” together through links or references stay close in our network. It also stretches the network out and apart so we’re able to see groups of actors, which in the program is called nodes.
In the scopus network we have used its modularity function, which calculates and colors actor groups via calculated closeness the the rest of the nodes in the network. Through this we can see 4 big groups, and a scattered group in between them all. These are 4 different views on IQ and are coloured as such. The group in between are the fundamental studies that everybody have to acknowledge and comment on, these we manually give a fifth color to distinguish them. |
tags
The tags used to describe websites in the hyperlink network was tagged by the following criteria:
Genetic Influence: Explicitly states that genetics have an influence on intelligence.
Environment: Explicitly states that environment have an influence on intelligence.
Reference: Site is linked to by another site as reference to back up an argument.
News: Is the website a published media source.
Blog: Is the website a blog.
Country: Write what country the author is from if available.
Governmental Authority: Governmental website.
Statistic data: Site uses arguments based on statistical modelling and research.
Anti-science: Site describes opposing views as against science, with the term anti-science.
Against IQ testing: This site express itself against IQ and the testing of it.
No race: Site argues for that different human races do not exist.
Education: Site argues for education as a main contributor to IQ
Genetic Influence: Explicitly states that genetics have an influence on intelligence.
Environment: Explicitly states that environment have an influence on intelligence.
Reference: Site is linked to by another site as reference to back up an argument.
News: Is the website a published media source.
Blog: Is the website a blog.
Country: Write what country the author is from if available.
Governmental Authority: Governmental website.
Statistic data: Site uses arguments based on statistical modelling and research.
Anti-science: Site describes opposing views as against science, with the term anti-science.
Against IQ testing: This site express itself against IQ and the testing of it.
No race: Site argues for that different human races do not exist.
Education: Site argues for education as a main contributor to IQ