Experiments

Date: 2015-10-07; view: 575.

Experiments are generally the studies that are the most precise and have the most weight to them due to their conclusive power. They are particularly effective in proving hypotheses about cause and effect relationships between variables. A hypothesis is a prediction of how one variable relates to another. There are two types of hypotheses, null and directional. The null is a prediction that there will not be any change in the dependent variable when the researcher changes the independent variable. The directional hypothesis states that the change in the independent variable will induce a change in the dependent variable. In a true experiment, all variables are held constant except for the independent variable, which is manipulated. Thus, any changes in the experimental groups can be solely attributed to the action of the independent variable. This is called being objective.

For instance, in an experiment to test whether music improves people's memories, we would have a sheet of paper with ten unrelated words on it for people to memorize. The control group would have no music playing in the background while the experimental group would have some music in the background. Because as researchers we have adhered to the scientific method and held all variables as constant as possible, if the experimental group does report better recollection of words, then we could assume that the music had an effect on memory. However, we must be certain to do our best to ensure that any controllable differences between the two groups are eliminated in order to ensure that no confounding variable interferes with the experiment.

There are two main ways to pick, or sample the subjects in an experiment, random and stratified. In a random sampling each person has an equal chance at being picked. This means that if 90% of the population being sampled from are Christian then 90% of the sample will be Christian. If the researcher wanted all religions represented equally he would employ stratified sampling. For instance, the experiment could be performed only on women, or on mixed groups with equal numbers of each sex in them, to eliminate the possibility of biased results from one gender having better average memory than the other.

Steps must be taken to make sure that there is no experimenter bias. Two common forms of bias are demand characteristics and expectancy effects. If a researcher expects certain results from an experiment and influences the subjects response this is called demand characteristics. If the experimenter inadvertently interprets the information to be as expected in his hypothesis it is called expectancy effect. To counteract experimenter bias the subjects can be kept uninformed on the intentions of the experiment, which is called single blind. If the people collecting the information and the subjects giving it are kept uninformed then it is called a double blind experiment.

The experiment should also be reported so that other researchers can repeat it. If an experiment isn't repeatable it will not hold much weight in the scientific community. To help an experiment be repeatable the researcher should have the variables be measureable, this is called being empirical.

Whether researching humans or animals the experiment should be ethical. When humans are the subjects they should be informed of what the study is, consent to being in it, be debriefed afterwards, and their information should also be kept confidential.