A hypothesis is a testable statement which seeks to explain a phenomena by examining the relationship between two or more variables. It is important to remember that a hypothesis seeks to explain a phenomena – it does not merely seek to describe it. Statements such as “people don’t like the government”, “people are generally opposed to the legalisation of cannabis” “some people support the death penalty whilst some are opposed” are not hypotheses. They don’t explain a phenomena they simply describe it – there is a lack support for the government, opposition to the legalisation of cannabis, and different views on the death penalty – these things may all be true but they are not explanations, they are not hypotheses.
It is also important to remember that a hypothesis is not a question – “why has support for the government fallen?”, “who supports the death penalty?” “which people are opposed to the legalisation of cannabis?” You may, indeed should, have a research question but a hypothesis posits a potential answer to that question – “Sun readers are more likely to be influenced by the political position of their newspaper than readers of the Daily Mirror”, “women are more supportive of the death penalty than men”, “adults in full-time employment are more likely to be opposed to the legalisation of cannabis than adults in full-time education”.
A hypothesis is a testable statement – it may be true or false – descriptive statements such as “people’s attitudes towards the death penalty”, and questions such as “who supports the death penalty?” are not a testable statements because they can’t be shown to be true or false. Less obviously, the statement “people are opposed to the Government” is not a hypothesis (or at least not a very good one) because it is not really possible to test this statement unless one can prove that nobody is opposed to the government.
In order to show that a hypothesis is not true, one should be able to define a null hypothesis which explains the situation which exists if the hypothesis is proved not to be true. In the case of the hypothesis “women are more in favour of the death penalty than men” the null hypothesis is not that men are more in favour of the death penalty than women (although this would disprove the hypothesis), it is simply sufficient to prove that there is no difference between men’s and women’s attitudes to the death penalty. Establishing a null hypothesis is an effective way of discovering whether you have a statement which can be proved true or not, i.e. a hypothesis.
In order to make their hypotheses testable social scientists introduce the use of variables. These are different factors which may influence the phenomena which is being explained. Hypotheses in the social sciences usually compare two or more variables. So if we use one of the examples offered above, in examining attitudes towards the death penalty one could hypothesise:
“men are likely to be more supportive of the death penalty than women”
In this case gender is being used to explain support for the death penalty, the variables are gender (which may be male or female) and attitudes towards the death penalty (which may be supportive or opposed).
A statement such as “students are in favour of the legalisation of cannabis” – is not a hypothesis. Firstly because it does not seek to explain a phenomena it just describes it. Secondly it doesn’t seek to compare variables. The only variable in this statement is attitudes towards the legalisation of cannabis, “students” is not a variable. There is nothing with which to compare the attitudes of students. One could create a hypothesis on this subject (it’s extraordinary how often students want to look at this issue) along the following lines:
“students are more in favour of the legalisation of cannabis than people who are not students”
However, if this was your hypothesis you would then need to bear in mind that the educational status of your respondents may not be the only factor which explains the phenomena which you have observed. So the factor (or variable) which affects individuals attitude towards the legalisation of cannabis may not be their educational status, but some other variable like their age, gender, religion, or even political affiliation.
The different variables which you identify in testing your hypothesis have different names. The variables for which you are suggesting there is a relationship are called the independent and dependent variables. In this case the independent variable is educational status (whether the respondent is a student or not) this is a fact, or external reality, which will not be changed by the research. The dependent variables are their attitudes towards the legalisation of cannabis – these vary from person to person but are not usually fixed and unchangeable in the way that independent variables like gender, age, social status are. What you are claiming is that people’s attitudes towards cannabis are dependent on their educational status.
This means that such research is essentially predictive. If your hypothesis is proved to be true then knowing someone’s educational status would allow one to predict their attitude towards the legalisation of cannabis.
Any other variables which may have a relationship with the dependent variables but which are not covered by the hypothesis are known as intervening variables – ie they intervene between the suggested relationship between the independent and dependent variable. Intervening variables tend to be other factors similar to the independent variable posited in your hypothesis. So if your hypothesis suggests attitude towards the legalisation of cannabis is dependent on educational status (the independent variable), it may actually be dependent on age, gender or political affiliation – these then are intervening variables.
When designing questionnaires to collect data with which to test a hypothesis you need to try to include questions which establish all the variables you are interested in independent, intervening and dependent. Generally speaking you would begin by covering the independent and any potential intervening variable – Are you a student? What is your gender? which age group do you fit in? What political party did you vote for at the last election? You should then move onto the questions which establish people’s attitudes towards the issue being discussed (the dependent variables), – Are you in favour of the legalisation of cannabis? Do you think cannabis is less damaging than cigarettes? And so on.
Large scale social attitudes surveys such as British Social Attitudes usually ask a very large number of questions about the background of the individuals being surveyed before moving on to the more substantive questions about attitudes. This enables a large number of independent variables to be tested. Small scale surveys, of the kind carried for most undergraduate research projects are usually more limited, not least because students often find it difficult to persuade a large number of people complete surveys comprised of several hundred questions. This doesn’t mean that such surveys can’t produce valid results but it does mean that careful consideration needs to be given to which variables to include.