29th January 2011 at 4:30 am #3758Taiwo OlaiyaMember
I need urgent assistance on how hypotheses can be tested in qualitative research29th January 2011 at 11:40 am #3761Eugenio De GregorioMember
you can find a very useful tool for the verification of hypotheses in qualitative research into the software ATLAS.ti.
It does implement the “Query tool” as a function for the verification of association between codes and groups of codes into single documents or sets of document. You can find an overview and a description in this videotutorial http://www.youtube.com/watch?v=C11CHbuo5ME
Hope it helps!
Eugenio30th January 2011 at 2:11 am #3760Jon WillisMember
Qualitative research is not really designed for hypothesis testing. It is designed to explore an area that can later be hypothesis tested with a quantitative approach (or alternately, it can be used to explore the meanings of quantitative answers in more detail). You will need to convert your qualitative data into categorical data that can be quantified (usually through coding), then use hypothesis test the categorical data.1st February 2011 at 1:27 am #3759AnonymousInactive
While most qualitative researchers don’t like to say ‘test hypotheses’ we all do it in a kind of way. For example, we have a research question (like ‘how do students experience their first year in sociology’). It is impossible not to have at least a feeling/hunch/vague idea about some aspect of the answer. (For example, based on your own experiences you might think that students probably come in with little idea about what the nature of that discipline) So we put a question into the interview around their previous knowledge not because it is logically related to the research question so much as because we think (hypothesise) that it is related.
Answering that question about prior views is a kind of ‘hypothesis test’ that will probably need some counting, though typically we would not have big enough numbers to warrant any kind of probability stats. It’s the sort of thing we might report in numbers (“thirty seven of our fifty-four students made it clear that they did not have a firm idea of what sociology were as they began the course”). Or we might turn the numbers into a vaguer term (like “over two thirds”). If the point became important for our overall argument, we might expand on it, typically by offering some quotes or examples. (“One student noted wryly that she arrived in the first lecture with a feeling that she would be learning about disease, and only after some weeks realised that she had mistaken ‘sociology’ for ‘pathology”). The format for reporting is very different from a classic ‘hypothesis test’, but the underlying action is not!
And as we analyse the data, we are constantly forming hunches that we test out. For example, listening to an interview, we note that the young woman seems to have enjoyed the emphasis give to gender in the first year classes, and think “but the young man in the previous interview complained about too much boring feminist crap. Maybe gender makes a difference to their experience –I’d expect females to enjoy the way sociology has been so influenced by feminist thought over the last decades” We’d ‘test’ this perhaps by doing a bit of simple counting , but more importantly by looking at our data form males and females and seeing what was actually said. Maybe a the blokes complain about feminism, but it’s in the context of talking about the lecturer, and when they actually discuss the topics in the course, both males and females seemed to enjoy the topics on gender and sexuality more than any other topics. So our hunch was right in one way wrong in another.
We would also look hard for the ‘deviant cases’ that seem odd. (Maybe there’s a woman who complained long and hard about the emphasis on gender, and said her favourite topic was methods? What do we make of this? Can we explain her? Does she suggest some new possibility for answering our question about gender influencing experience? “Hey! She’s the oldest person in the group! Does age make a difference to how they experience the course?”)
There’s no right way to set up your hypotheses or to test them, because they are specific to your field and topic. That doesn’t mean that we don’t use them. We should be constantly testing ideas against our data, and unsparing in the rigour of our tests.
There is no ‘right’ way, but I do have a ‘favourite’ way of my own. I was taught by two wonderful researchers (David Hickman and Lyn Richards) that it is very useful to play the classic ‘null hypothesis’ game with qualitative data. You have a pre-existing feeling about the answer to the research question, or during analysis you get a hunch about the data. Write it down and then make it an opposite- a ‘null hypothesis’ (The classic null hypothesis statement is “there will be no difference between control and experimental groups”.) So the idea about previous knowledge becomes “I expect students to say they had a clear idea of the nature of sociology before they began the course” and the idea about gender that evolved during analysis becomes “gender makes no difference to their views”. Then ask yourself what data you need to examine to find evidence that will support or disprove the null hypothesis. Look for that data and be very strict with yourself about what you accept as support or disproof.
What you then while looking will depend a lot on how you are working. If you are coding your data, you may have some categories that you can search (for example you coded all the text about ‘the lecturer’ and all the text about ‘course topics’). You need then to find which quotes come from males and which from females. If you are using software such as Atlas ti or NVivo to manage your data, you will find that there are tools for this. If you have no coding, you will need to go through your data looking for all the relevant comments. If your data are not textual, you will work in a different way. There is no simple recipe for ‘hypothesis testing’ with qualitative material.
The human mind is an odd beast. I find time after time that when I force myself to say what I think (hypothesis) then state its opposite ‘null hypothesis’ and seek material relating to the null, stuff I had ignored while focussed on the original implicit hypothesis confronts me. As I deal with that ‘stuff’, my analysis becomes richer and more nuanced. I might or might not change my mind about the original hypothesis. What matters is that I am clearer about what I think and why.
Put aside all the epistemological qualms you might have about positivism. (I share them). Self- consciously using null hypotheses as a temporary device for dealing with qualitative data is not about becoming a positivist. It’s a trick that forces you to look at your data from more than one angle. You are not suddenly becoming a quantitative researcher who believes in ‘objectivity, you are simply looking carefully and systematically at your data and taking your own subjectivity into account as you do so. When you reach a conclusion (either the small conclusions that ‘the data are like this’ or the bigger conclusion that ‘the answer to my research question is this’) you will be in a good position to offer a convincing narrative
Hope this small rant is helpful.28th September 2017 at 1:07 am #107050Marlo ManahanParticipant
Can you give me an example of Hypothesis in Qualitative Research?28th August 2018 at 1:18 am #110478jessicapattersonParticipant
The human mind is an odd beast. I find time after time that when I force myself to say what I think (hypothesis) then state its opposite ‘null hypothesis’ and seek material relating to the null, stuff I had ignored while focussed on the original implicit hypothesis confronts me. As I deal with that ‘stuff’, my analysis becomes richer and more nuanced. I might or might not change my mind about the original hypothesis.
- The forum ‘Default Forum’ is closed to new topics and replies.