What are the similarities between descriptive and inferential statistics? What are the differences? When should you use descriptive and inferential statistics?
When conducting a data analysis, both descriptive and inferential statistics are employed. Descriptive and inferential statistics are mostly used where research is conducted on a group. These two branches of statistics bear both similarities and differences. These differences and similarities also determine when it is appropriate to employ either descriptive or inferential statistics to a research project.
Descriptive statistics refers to the analysis of data in a manner that gives it meaning. Descriptive statistics is the branch of statistics that allows analysts to highlight patterns that emerge from certain data sets. However, descriptive statistics do not allow analysts to draw conclusions or prove any hypotheses (Montero, & León, 2007). On the other hand, inferential statistics are employed when making inferences or predictions. Inferential statistics make suppositions about a populace depending on the analyzed and observed samples.
Descriptive statistics are only used to describe a specific group. Consequently, the outcomes of a particular sample cannot be comprehensive enough to represent a bigger group. On the contrary, under inferential statistics the results of a certain sample can be generalized to represent a larger group. The main connection between inferential and descriptive statistics is that they both use similar data before an analysis is conducted. Differential statistics use this data to exemplify a bigger group while inferential statistics focus on this data to describe the group being studied.
Descriptive statistics are used when a sample does not need to be extended to represent a larger group. Applications of descriptive-statistics include pie charts and graphs. Population censuses are a good example of this application. Inferential statistics are applied where a relationship is expected between variables. This means that a small sample can represent a larger population. Instances of inferential-statistics comprise of survival-analyses and correlation analyses.
What are the similarities between a case study and an ABAB research design? What are the differences? When should you use case study and ABAB research designs?
ABAB research designs fall under the category of single case research designs. The ABAB research design has often been considered similar to case studies. However, there are foundational differences between these two research designs.
The ABAB research design focuses on a single individual and it is often used to monitor how effective a certain treatment or intervention is. Under the single case design, a baseline of the research is established before the research introduces a change – (AB design). When using the ABAB design, the change that is introduced is then withdrawn to monitor its permanent effect. After being tested, it is then re-introduced. A case study is a non-experimental observation of activities that are either going to occur or have already occurred. Case studies can be retrospective or prospective. The main difference between these two designs is that the ABAB design is more experiment-oriented (Cook, & Campbell, 2002). The case study design is mostly observational. Another difference is that the ABAB design is quasi experimental and it depends on internal interventions. On the other hand, the case study design is a more descriptive design. The main similarity between the two designs is that they both focus on a specific subject.
Case studies are normally used when a researcher wants to access the impact of an event or a situation on a participant. The study could be trying to determine the impact of an event’s cause or its effect. For instance, the causes and effects of an event like a hurricane would constitute a good case study. ABAB designs are best applied to experimental treatments or therapies.
What are true experiments? How are threats to internal validity controlled by true experiments? How are they different from experimental designs?
Whenever an experiment is conducted, variables are manipulated and the effects of this manipulation observed. One type of experiment that can accomplish this task is the true experiment. True experiments are considered as the most accurate types of experiments. True experiments make use of both a control group and an “experimental group” (Durbin, 2011). Data is collected concurrently between these two groups.
True experiments control internal validity threats that are often associated with experiments. For instance, a true experiment randomizes and compares both the control groups and the experimental groups. By mixing these two groups, these experiments ensure there is equivalence between the two groups (Frankfort & Leon, 2006). Another way of controlling internal validity threats is by ensuring that both the control and experimental groups are pretested. This means that the state of participants before the experiment begins is noted. This ensures that the relationship between cause and effect is not interfered with by pre-experimental activities. For example, if an experiment involved investigating changes in students’ grades this semester, events that happened in the last semester can affect the outcome of this experiment.
There is a big difference between experimental designs and true experiments. True experiments are refined forms of experimental designs. Experimental designs do not include a control group in their experiments. Therefore, the study focuses on only one group. This means that in experimental designs there is a high chance of the collected data being affected by internal factors. The other difference is that most experimental designs do not make the use of pretests or posttests. All true experiments include both a pretest and a posttest.
What are quasi-experimental designs? Why are they important? How are they different from experimental designs?
Quasi-experimental designs refer to experiments that do not incorporate a random assignment. Unlike experimental designs, the internal validity issues of these experiments are not completely addressed. However, quasi-experimental designs are easier to implement as opposed to experimental designs. Examples of quasi-experimental designs include the nonequivalent group designs, analysis of covariance designs, regression-discontinuity designs, and proxy pretest designs.
There are many known uses of quasi-experimental designs. In average, the nonequivalent group and the regression discontinuity designs are used more regularly than the other Quasi-experimental designs. The Nonequivalent group design (NEGD) lacks the incorporation of random designs. When conducting experiments using this design, two similar groups are picked. This similarity is important when comparing a treated group with a non-treated one. However, there is no way of knowing that the similarity between the two groups is not subject to internal variables. Just as the name of the design suggests, the groups may be similar but they are non-equivalent. The regression-discontinuity design (RD) is mostly used to determine the effectiveness of a method or treatment. The RD is a pretest-posttest design. The only difference between RD and other pre/post test designs is that in RD there is a set cut off score. Participants have to achieve this score before their inclusion in an RD experiment. This design is very useful to programs that are aimed at helping those people with serious problems.
The main difference between quasi-experimental and experimental designs has to do with the type and amount of information collected from participants before and after an experiment (Kenny, 1975). Quasi-experimental designs are very type-specific when it comes to pretest information. On the other hand, experimental designs are not type-specific.
Cook, D. & Campbell, T. (2002). Experimental and Quasi-Experimental Designs for Generalized Causal Inference. Thousand Oaks, CA: Wadsworth Cengage learning.
Durbin, J. (2011). Incomplete blocks in ranking experiments. British Journal of Statistical Psychology, 4(2), 85-90.
Frankfort, C. & Leon, A. (2006). Social Statistics for a Diverse Society. Thousand Oaks, CA: Pine Forge Press.
Kenny, D. (1975). A quasi-experimental approach to assessing treatment effects in the nonequivalent control group design. Psychological Bulletin, 82(3), 345.
Montero, I. & León, O. (2007). A guide for naming research studies in Psychology. International Journal of Clinical and Health Psychology, 7(3), 847-862.