So I developed an algorithm to do some visual stuff in 3D space and tested it out on participants as you do. I have some good quatitative results but I should really triangulate my results with some qualitative ones which I collected through post-experimental semi-structured interviews.
I have been going through these interviews and coding, recoding and drawing networks to estabilish how some codes are related to others. It is not as complicated as grounded theory but maybe it is a form of latent content analysis. It just seems like this form of analysis would have been established elsewhere like in the social sciences or psychology. I am not certain what it is called. I have looked it up on google and some books but the descriptions don't seem to quite fit.
Can someone in this forum with the know-how help me find a name to the types of analysis I am doing? I can provide more details on the analysis I did. Just ask the questions.
What exactly are your interviews trying to get at? Can you explain a little more about your questioning and what you are hoping to find out, as this will help you choose your method of analysis?
Meanwhile try -
Tashakkori A & Teddlie Charles (1998) Mixed Methodology Applied Social Research Methods series. Vol 46,. Sage
Fairclough, N. (2003) Analysing Discourse. Textual Analysis for Social Research. Routledge.
This is a brilliant book.
Tashakkori A & Teddlie Charles (1998) Mixed Methodology Applied Social Research Methods series. Vol 46,. Sage
I have not read the other one you mentioned. I am trying to look at positive and negative statements made about the algorithm I am evaluating and then trying to find out the reasons behind why the participants thought this. For instance if someone says "I think your thesis sucks!", then I would ask in the interviews "Why" and they would continue to say "Because xx doesn't really represent yy since zz is missing"... So I am looking for all the reasons why the participants thought my thesis sucks or my thesis rocks. I am building a cause-effect network using Alias.ti. Any idea what this method might be called?
sorry for delayed response whattodonext - been busy working for once!
By the sounds of this you may not need discourse analysis, but I am by no means an expert. Discourse analysis is great for investigating assumptions, the values and beliefs behind certain statements i.e what is perhaps being implied rather than overtly stated. For your anaylsis it may produce data that is 'richer' than you require if you don't need to get really into syntax and linguistic structure from your interviewes responses. Also there are many different types of DA and you will be opening a can of worms when it comes to writing up a section on methods & methodology. Mmmmm leave this with me, I am going to have a think about it. I am tempted to say go for a content analysis, but I am going to ask one of my colleagues who is designing some AI & find out what methods they are using to unpack their interviews.
Isn't content analysis simply looking for some words and counting the number of it? I thought what I needed was content analysis but I realised that I wanted slightly richer meaning to the data. Not as rich as Discourse Analysis but more than simply a count of a specific word.
So far what I have done is something like coding the interviews. So for instance if one participant says "The fruit looks like an apple", I would code that as "AppleLook". Then if another participant says "The fruit is an apple since it is round and red", I would attach three codes: "AppleLook", "ShapeRound" and "ColourRed". Then I would put in network association saying "AppleLook" is caused by "ShapeRound" and "ColourRed". Then at the end of it all, I would write it all up with sample quotes.
I am sure someone else probably did something similar. My use of qualitative analysis is to look for reasons why people got a certain opinion of my algorithm so this seems to work.
Have a look at http://academic.csuohio.edu/kneuendorf/content/
It has a good explanation of content analysis and a bibliography which might help.
Content analysis is a bit more than just counting language units.
The fact that you are encoding & decoding means you are already making inferences & value judgements, so you have to set these inferences in context witthin the espistemology or world view of you & your research.
I hope that helps?
I did a content analysis years ago for my BA project, so I can't offer you any more informed advice I'm afraid than I have already; as I have limited knowledge and expereience + wouldn't want to misinform you. I guess you could try speaking with a Communications or Media / Cultural Studies research student at your uni? This is traditionally a media studies type mthod used for decoding TV & newpaper language.
Good luck
I have spoken with the AI gang here, Few questions - what is the reason for your analysis? Are you using the data to make improvements to the algorithm?
If it is just looking at user aceptability, then apparently there are established methods for this.
How have you done the interviews> Have you vidoe-taped them?
I did use their comments to design my next experiments... and I did video tape their performance but not the interviews... Interviews were only audio taped.
Masters Degrees
Search For Masters DegreesPostgraduateForum Is a trading name of FindAUniversity Ltd
FindAUniversity Ltd, 77 Sidney St, Sheffield, S1 4RG, UK. Tel +44 (0) 114 268 4940 Fax: +44 (0) 114 268 5766