Composing a hypothesis


If you plan to test a hypothesis in your research, you will need to identify measurable variables and distinguish between the independent and dependent variable. You will then need to design a test to see whether the hypothesis is true (i.e. proven) or not (i.e. disproved). For example, you might want to find out whether people’s drinking habits varied across the days of the week. The days of the week would be the independent variable and the alcoholic drinks they consumed on each day would be the dependent variable. You then need to construct a statement that identifies an association between the two variables.

Your hypothesis might be, for example, ‘people drink more alcohol at the weekend than they drink in the week’. However, before setting up the test you would need to consider the statement more carefully.

• Who are the ‘people’? This might be determined by what participants you can include in your sample.

Would it be interesting to compare the drinking habits of different age groups, e.g. young people/young adults compared with older adults with family responsibilities and a mortgage to pay? How could you distinguish between the different groups? By age? By gender? By whether they are a parent or not? How will you find this information out?

• Are you going to ask the respondents to tell you whether they drink more on particular days? Do you think they would know this reliably? Could you ask a less direct question that would enable you to calculate how much they drank? Would it still be valuable to ask the question because there is often a mismatch between people’s perceptions of what they consume and what they actually consume. And would this mismatch be interesting to investigate in its own right?

• Could you recall what you drink (alcoholic or nonalcoholic) each day? If you cannot, then are your respondents likely to remember? Could you ask them a more specific question or, rather, give them some answer options that enabled them to give you more accurate information about their drinking habits?

• What do you mean by an alcoholic drink? Do all alcoholic drinks contain the same amount of alcohol? Does it matter? Do you need to know what people drink more specifically? How could you get this information from your survey questions? Do you need to know the size of the drink as well as what type it is (e.g. wine can be served in a range of different-sized glasses)?

• What about students who are under 18? It is illegal for them to consume alcohol other than in their home. Does this mean you should ask respondents where they drink their alcohol? Do you think under-18s might not give you a truthful answer because they are drinking illegally? Do you think people might not give you a truthful answer because they are aware they probably drink too much? How could you ask the questions so that you reduced respondents’ concern about giving you this information? What are the ethical issues that might be relevant here specifically?

• What is ‘the weekend’? Friday, Saturday and Sunday nights or just Saturday and Sunday?

Would a different combination of days be more representative of the days when you suspect people drink most? Do you think all your respondents will think of the weekend as being the same days?

How could you make sure in your questionnaire that there was no confusion about how you were defining a ‘weekend’?

This sort of questioning can be relevant to any project but it is important to be clear about what you are actually measuring when testing a hypothesis, otherwise the validity of the test may be considerably reduced. For example, you may end up with an amended hypothesis such as ’people consume most alcohol at the end of the week‘.


Now, on the selecting a subject forum- carefully word a hypothesis question under your original post. 

For example, my hypothesis might be; "People who experience nuclear melt-downs are more likely to develop PTSD than people who experience natural disaster."

Last modified: Tuesday, 31 Mar 2020, 15:57