150 likes | 168 Views
“Developing Research Questions”. Can you develop a research project?. Up until now you have may have been comfortable with analysing and interpreting the results from a given dataset. …. In your statistics class you were provided with data from an existing experiment.
E N D
Can you develop a research project? • Up until now you have may have been comfortable with analysing and interpreting the results from a given dataset. …. In your statistics class you were provided with data from an existing experiment. • Now it’s over to you, to design and collect your own data. • I hope this video helps you with the task of developing a research question.
Q4 What would be your response variable? • The response variable (also known as the dependant variable) is the outcome or measured variable. In this case and commonly in agriculture the response variable is a continuous variable. In a small poultry feeding experiment the three diets could be the treatments. The three diets are called the explanatory variable. • Q5 How would you collect the data on each chick? • I’d have to weigh them individually after a few weeks of growth. In fact I’d probably weigh them weekly.
Q6 How many chicks would you need? • We can to do a sample size analysis to determine the number in each treatment. If one chick is an experimental unit I will have to think about independence of the measurements and whether to use individual chicks or pens of chicks as the experimental unit. • Q7: Would they be in pens or cages? How would you arrange the chicks in the experiment? • I would have to know what the experimental unit was – ie , is it a chick or a group of chicks. Maybe a group of chicks could be experimental unit. • I would then have to consider replication of these experimental units and looking at the mean weight of the replicates. In this way I could determine how many chicks and pens I need.
Q8: Would you be ready to describe the statistical question? • Once I have the design chosen I would be ready to describe the null hypotheses. The null hypothesis, is the hypothesis of no difference. It is written in terms of the weight gain for the three diets. • In other words I am saying that under the null the mean chick weight gain is the same for the 3 diets. • Q9: Can you describe the alternate hypothesis? • The alternate hypothesis is that there is a statistically significant difference (say at alpha=0.05) between the weight gains of one or more of the diets.
Q10: Do you have a type of design in mind? • With this experiment I would design and analyse it as an analysis of variance (ANOVA). This is because I have three categorical explanatory variables and the response variable or dependent variable is continuous.
Q11 How does this diagram help you? • This diagram shows a research cycle. The first thing I have to do is to work out my research question Q12 How do you start developing a new research idea? • There are 2 key ways to start with developing a research idea. • a) people, and people include those from industry, scientists and research students, and it also includes • b) the scientific literature Q13 What about your contacts and experience? Through my personal experience I might be aware of an industry problem or from discussions with friends or colleagues. Some practical problems are mentioned at conferences and scientific meetings
Q16 What is it you are looking for? • We are looking for gaps in the knowledge. So when I look into the scientific journals. I can use industry reports and look for discussion points or comments in the introductions of papers. I look for gaps and mentions of what else we need to find out. • Q17 How wide do you look for questions? • I also don’t forget the technical literature, the press and conferences, and chats in the tea room.
Q18 What do you do to focus this • I read and focus a little more with this extra information and plan some simple research questions/ that’s my starting point. Q19 Do you link this to the scientific method • If I follow the principles of experimental design I can examine cause and effect, which is the powerful aspect of experimentation. One or more treatment may cause different effects on the response variable. In an experiment you need at least two treatments for a comparison.
Q20 Why is planning so important? • They say time taken to plan and develop the research questions and get appropriate methods really is the best way to approach a new project. • You even have to plan the actual analysis at the outset. • The design is linked to the analysis. Q21 Where do you get extra help ? • I make sure that and I check my design with a statistical consultant. • Many universities have dedicated statistical consultants to go to get advice at the starts of the project.
Key messages • What is a practical scientific research question • Where are the gaps in the knowledge • Get ideas from people • Get ideas from literature • Get ideas from conferences and the press • Develop your statistical hypothesis • Think it through and keep it simple • Plan the design before you start