What are experimental hypothesis and null hypotheses? (HL Only)
The null hypothesis states that no significant difference is expected to be found between the groups on the measure of the dependent variable, and that any difference found is due to chance. This will become important later on when you apply the inferential statistics to your data. The purpose of a null hypothesis is to test whether or not the results you got were in fact from the manipulation of the variables, or if it was just by chance.
Experimental Hypothesis (also called “research hypothesis”)
The research hypothesis must be a clear, concise prediction of what is expected to be demonstrated in the experiment.
This must be operationalized: it must be evident how the variables will be quantified, and may be either one- or two-tailed (directional or non-directional). It’s imperative your research hypothesis is formulated based on your background research. This is the mistake that most students make. It should be clear in your introduction how you came to formulate your hypotheses.
HL students need to have an experimental and a null hypothesis. They should make note if their experimental hypothesis is one-tailed or two-tailed as this will be important when they come to conduct the statistical tests.
When researchers make a prediction it’s only natural to want to “prove” that prediction right. So a null hypothesis is important because it ensures the researchers are thinking about the possibility that there won’t be any cause-effect relationship determined. It should always be the null hypothesis that is accepted or reject for this reason.
It’s important that the null hypothesis is also operationally defined. That is to say, it’s defined as it exists in the present study. To use the Mozart effect as an example, it wouldn’t suffice to say “music won’t affect intelligence.” This is not operationalized because the term “music” is too broad as is “intelligence.” It’s not known how they exist in the study. A better operationalized null hypothesis would be:
“There will be no significant difference in the amount of accurately completed multiplying fraction problems between the participants listening to Mozart’s operas and the participants with no music. (two-tailed)”
One tailed hypothesis means that “Variable X” will affect “Variable Y” in a certain direction (e.g. it will cause it to increase). A two tailed hypothesis means that X will increase or decrease Y. So a one-tailed test only tests the significance of the data in one direction (e.g. the probability that X will cause Y to increase). Whereas a two-tailed test will test the significance of the data in both directions.
Here are some examples…
Operationalized One-Tailed Null Hypotheses Examples:
- The verb “smashed” in the critical question will not significantly increase the participants’ estimates of the speed of the vehicle.
- Knowing the title of a passage of informational text will not significantly increase the participants’ ability to recall details from that passage.
- There will be no significant increase in the ability to correctly solve fractional equations when listening to Mozart’s opera music.
Operationalised Two-Tailed Null Hypotheses Examples:
- The verb choice will have no significant effect on the participants’ estimates of speed.
- Knowing the title of a passage of informational text will not significantly affect the participants’ ability to recall details from that passage.
- There will be no significant effect in the ability to correctly solve fractional equations when listening to Mozart’s opera music.
Notice how the two-tailed hypothesis do not state a direction (increase or decrease) but simply state an effect. Note also how the variables in each experiment have been written specifically (i.e. operationally).
So if the study has a one-tailed hypothesis, it means a one-tailed test should be conducted. If the hypothesis is two-tailed, a two-tailed test should be conducted.