There are many different ways to test hypotheses in Psychology.
Working with your Research Advisor (RA), you will decide the best Method for your investigation. Students sometimes begin the thesis process with a “Method in Search of a Question," instead of the other way around. While there are always limits to the scope of thesis research (e.g. time, expense, access to populations and equipment), you should consider all possible approaches suitable to your topic.
1. Experiments include research conducted in a laboratory or in the field.
- Data may be obtained using observations, biochemical assays, questionnaires, psychometric tests, interviews, physiological measurement, etc.
- In an experiment the researcher exposes participants to some treatment, experience, or activity while carefully controlling all other conditions,and examines outcomes of that exposure to determine the effect of the experimental manipulation
- Most often, there will be a comparison to participants who are exposed to something different or who do not experience the experimental manipulation under the same controlled conditions
- It is expected that there will be random assignment of participants to conditions
- It's the only research method that permits you to test a causal hypothesis
The advantage of an experiment is that the researcher has control over all of the variables affecting outcome. The disadvantage of an experiment is that the results may not be generalizable, or may not be fully generalizable, to the “real world.”
- Also, it may be very costly to set up the controlled conditions
2. Quasi-experiments apply experimental designs to pre-existing groups. For example, you cannot “assign” people to the male vs. female “condition” in a study
3. Correlational Designs can be used across settings.
- In these studies, you measure your variables without any “treatment” or intervention
- What makes it correlational is that you examine the ways that variables are related to one another
- Even when the relationship is strong, and you believe that one variable is influencing another, you cannot draw conclusions about causal relationships
- Correlational designs allow you to learn about ways that one thing may predict another (e.g., how socioeconomic status is related to academic outcomes)
An advantage of correlational design is that it does not require the researcher to ask participants to do anything (typically, it uses data about existing characteristics, thoughts, behavior, etc.). A disadvantage is that the researcher cannot control many of the factors influencing participant behavior or response
4. Surveys and questionnaires are often used to collect data both in experimental and correlational designs. They may be used by themselves, or in conjunction with other sources of data (e.g., observations, measurements, medical records, face-to-face interviews, etc.).
- Some characteristics cannot be observed directly, and self-report forms may be the only way to obtain the data (e.g. mood, cognitions)
- Surveys may be used without any direct contact between the experimenter and the participant. Online data collection is very popular right now
- Who is sampled is very important to consider, in order to avoid biased data
Many ALM candidates want to develop their own paper and pencil measures. To obtain valid results using the survey method, there are a number of considerations:
- The survey must be reliable
- If items are combined into scales, there must be evidence that they represent a single dimension—this concept is referred to as “internal consistency”
- There must be evidence that the measurements are valid - that is, that the test items or survey questions actually tap the construct the experimenter is trying to measure
- Survey construction is labor intensive and time consuming, often involving multiple pilot tests and extensive analysis before a new measure can be used to test the hypothesis of interest
The advantages of these measures are cost, ease of administration, and the ability to gather large samples. A disadvantage is that questionnaires and surveys are often indirect measures of actual behavior. If you rely on random volunteers, they may not be representative of the population of interest.
5. Sample Size
Students often want to know “how many subjects do I need for a thesis?” This is the wrong question! The correct question is: "how many do I need to test my hypothesis?"
The answer is: “it depends”
… on your area of inquiry, on the type of outcome data you are collecting, on the effect size you are expecting, etc.
- Review “Power Analysis” from your Stat 150 text, talk to your RA, and see what others are doing in similar investigations
6. Statistical Analysis
As you are designing your investigation, you should already be thinking about how you will analyze your data.
- How will responses be coded or scored?
- Will you need to demonstrate the reliability and/or validity of your measures?
- What type of statistics will you use to evaluate your hypothesis?