Christopher Stevenson

Christopher Stevenson

Chris is an epidemiologist and biostatistician with a particular interest in evidence based health policy. He is a passionate teacher with a commitment to blending teaching with cutting edge research

Location Deakin University, Geelong, Australia

Activity

  • John
    You have identified a key issue for any researcher. It is always a challenge to focus on the question and develop the most appropriate response using the most appropriate tools rather than to just reach for what is already familiar and shape the question to suit your preferred tools. I'm glad we have been able to broaden your horizons and I look forward...

  • @ChrisRawson @allanbeer
    Hi Chris, Allan
    Chris is right - it is to do with the fact that the CI tells you about the mean and not the range of values in the population. Take the January data, for example. The sample mean is 48.2 and the sample standard deviation is 38.19. The 95% confidence interval is (41.7 to 54.7) (slightly different to your values Allan...

  • Hi Graham. The categories themselves have non-numerical descriptors, but we associate numerical data with them - usually in the form of counts. It is these counts which are the quantitative data. These counts can then be put in the form of a table and quantitative statistical analyses done on the table. Similarly continuous data can be classified by a...

  • This is a really good point Geraldine. When you start thinking deeply about your research question, then you will often find that a different approach to the one you started with is necessary. It's important to do this sort of thinking as you develop your research question and be flexible regarding your methods.

  • This is a good start to the problem Alan. One more thing you might like to think focus on the difference is between your two areas. One area had twice the number of households willing to share rides compared to the other. This might lead you to look at what the characteristics of the second area are and what extra barriers to ride sharing it might contain....

  • Hi Janelle
    Hopefully the topics we cover in week 2 will help point you towards how you would analyse your data. Please let us know in the discussions.

  • Hi Graham
    This is an area where it is easy to confuse the two concepts of qualitative and quantitative. We often want to get a measure of the strength of peoples opinions - opinion polls are an obvious example of this. The scale you describe is an attempt to derive a quantitative measure of the subjective opinion. The end result is a number which you can do...

  • Hi Graham.
    This is a good summary of how the three methods often work together. You use field and qualitative research to help you design an effective controlled research phase.

  • Thanks for this post Graham. Often when you read the results of a research project it sounds like a straightforward process - first you decide your question, then you decide your methods, then you collect your data... and so on. However, it rarely happens this way. As you discovered, nearly every researcher will spend sometimes considerable time going back and...

  • Hi Sue
    You have discovered one of the big benefits of spending time developing a research question in the sort of specific detail we talk about in this step. It is often the case that we only get a clear idea of the shape of our research problem and the methods we will need to use when we break our question down into details such as what we are going to...

  • Hi Ian. The aim of this question is to get you to apply the concepts in week 1 to a real research issue in an area that you are familiar with. We won't actually be expecting you to collect data as part of this course. The main thing is to practice the concepts we have studied so far by designing a research study that interests you.

  • Hi Ian. The paper you refer to does define reliability as being both "The extent to which results are consistent over time and an accurate representation of the total population under study". However, their discussion focuses on just the first aspect and this is the way I have always used the term reliability. External validity is the broader concept of...

  • This is a really good idea Leanne. One of the issues for Miranda and her team was knowing how much of the changes were the result of their intervention and how much would have occurred anyway. Choosing a comparison shop without the intervention where you could collect the same data would give you this information. Unfortunately it wasn't practical in Miranda's...

  • This is a really good approach Linda. You started with a broad question (what is the effect of nurses eating healthy meals). Now you are moving to the specifics of how you define a 'healthy' meal and how you might encourage nurses to eat a healthy meal. You need to move to the level of specific, measurable detail before you can run a successful quantitative...

  • Hi Linda. I think it is interesting that you have chosen to focus on nurse productivity as an outcome of healthy meals. I can see how a hospital administrator might be more interested in an intervention that increased productivity as well as improving nurses' health. However, I think you might have to be more specific about how productivity has increased. The...

  • That certainly sounds like an interesting study Carolyn. Have you thought about what you might measure and how you might collect data to answer this question?

  • You make a good point Mariana about how research usually works. You don't often start with a single question, do a data collection and come up with your final answer. Mostly you start with a more general data collection and use this to narrow your focus. After you have analysed your data, you will come up with appropriate strategies. The you will move to a...

  • You raise an important point about designing research. If you want to know how effective something is, then you first need a baseline measurement. You can't tell how much your programs of physical activity have changed fitness unless you first measure how fit your school group is to start with.

  • You are right Linda in saying that numbers are only as good as the way they were obtained. For example, you could measure smoking rates with a carefully designed population survey, or you could go outside and count the number of people you see smoking. Either way you get a count of the number of smokers, but only one of these counts is likely to mean anything....

  • You make a good point here Patrick. We want to know two things - will the sales of healthy foods go up and will the sales of unhealthy food go down. We don't just want people to eat more of the healthy foods - we also want them to eat less of the unhealthy foods.

  • You make a good point Morgan. One good way to use data is to help prioritise in an objective way. In your case it's about deciding on the relative burden of two different health issues in an objective way.

  • You make a good point Stan. Numbers have a vital place in developing evidence to support decision making, but they are not the only form of evidence - as any of you who have done our companion course on Why Experience Matters would know.