The Experiment Flashcards
What is the purpose of an experiment?
The point of conducting an experiment is to gather data about some phenomena of interest. New insights are always derived through inspection of the data collected from a prior experiment.
What is the pretest-posttest design and how does it work?
Pretest – posttest design is a simple design where you first observe one group, followed by treating the group (apply some independent variable) and observe the effects. This is a very limited technique as you cannot rule out other factors which may explain the differences, or the lack thereof, between the two measurements. Using a control group is a technique to reduce this issue. The control group should receive no, or an alternative treatment. This allows you to better understand the effect of the treatment. If the group assignment is randomized, we have a true experimental design, if not, the design is quasi experimental.
What is the Solomon four group design and how does it work?
The Solomon four group design expands on the pretest posttest design by adding two additional, randomly assigned groups. Group three is treated equally to the first group but is not observed in advance. The fourth group is only observed after the treatment. Such a design enables to determine whether the first observation effects the final. If so, we can determine the effect of the treatment by adjusting for the effects of the initial observation. In short, the Solomon four group design enables us to take control over other potential explanations.
What is Time serial designs and how do they work?
Uses multiple observations of a group. These observations enable us to establish a baseline which represents the true value of a phenomenon. When we later perform a manipulation followed by a sequence of observations, we have gathered data which can be used for comparison with the baseline. Control groups are not that necessary, but a possible configuration is to perform observations on two groups simultaneously where we abstain from performing the manipulation on the control group.
What is Factorial design and how does it work?
Factorial design is useful when comparing the effect of two different techniques, methods, etc. By using four groups we can determine which methods or construction of methods is the most effective. The first uses method 1 twice, the second begins with method 1 and finishes with method 2, group three does the opposite of group two, and group four only uses method 2. The result allows us to inspect which method or combination of method to be superior.
Describe descriptive statistics and the conditions the prior experiment must satisfy to enable its use.
When the data is gathered, and if the measurements are of the interval or ratio scale, we can utilize several statistical techniques to work with the collected data. Utilizing the normal distribution or chi-square tests are common approaches for testing hypothesis. There are some assumptions that must be met before descriptive statistics can employed. Firstly, the data must be from a random sample, while the sample size should be of sufficient size (a rule of thumb is a minimum of 20 samples). Finally, the observations needs be independent from one another.
What is the role of inferential statistics?
Eliminating the element of chance is critical when utilizing statistics in research. Even though chance can never be fully eliminated, we can reduce its impact on the observations. One approach is to repeat the experiment a few times over. This is however often too expensive and or time-consuming endeavor. This is where inferential statistics come in handy as they allow you to reduce the element of chance through statistical procedures. Some examples of inferential statistics are confidence intervals and t-tests.
What is standard deviation and variance?
Both measure the variability of scores around the mean. Standard deviation is the square root of variance.
What are confidence intervals?
A confidence interval gives an estimated range of values which is likely to include an unknown population parameter, the estimated range being calculated from a given set of sample data.
Confidence intervals may be more informative than the simple results of hypothesis tests (where we decide ‘reject H0’ or ‘dont reject H0’) since they provide a range of plausible values for the unknown parameter.
