How do I know what type of statistical analysis to conduct?
Researchers working on studies, theses, and dissertations, often ask this question. Ideally this is a question to ask at the beginning of the research project, rather than the end. It is at this point that research needs should be assessed and hypotheses determined. During the initial stages of planning is the time to involve a statistics consulting firm to help you formulate the basis of your study. There are many things that can influence the type of statistical analysis to use but here are a few important ones:
1. Empirical information (statistical analysis) related to your study
If you are at the point in your study or dissertation that you have begun to consider the type of statistical analysis you need to conduct, then hopefully you have considered the statistical analysis of other researchers in similar studies. It is wise to do this, if only for comparison purposes. However, utilizing the same statistical analysis that has been utilized by other researchers and experts in your field will lend credence to your findings and avoid the difficult, if not embarrassing, task of defending why you chose a different type of statistical analysis.
2. University/College requirements
While I don’t believe it appropriate, many colleges and universities stipulate the type of statistical analysis their dissertation candidates can use. Often, the statistical analysis requirement is the use of multivariate analysis. Make sure you are aware of any of these requirements prior to formulating your research questions and hypothesis.
3. Formulation of the research questions and hypothesis
It is very important to have an idea of the statistical analysis to be used before formulating your research questions, hypothesis, and methodology. The wording of your research questions and hypothesis will determine the type of statistical analysis to be conducted.
· Words like relationship and correlation, tend to correspond with correlational statistical analysis. Sometimes these words also represent regression statistical analysis, but usually correlational statistical analysis.
· Words like predict, affect, and impact tend to correspond with regression statistical analysis, both linear regression and logistic regression.
· Research questions using the difference tend to correspond with t-tests, both dependent and independent and ANOVAs.
This means you should have an idea of the statistical analysis to be used before crafting your research questions and hypothesis. It is much easier to do this before the proposal is approved than after.
4. Ability of the researcher
While every researcher aspires to change the world, the statistical analysis should fit within the comfort and abilities of the researcher. If you are having difficulty formulating your methodology around statistical analysis with which you are comfortable, it may be wise to consult a statistics consulting firm to help you with the data analysis and statistical analysis. If you have had two or three statistics classes three years ago, you may not want to choose a complex statistical analysis utilizing structural equation modeling or even multivariate analysis. Though a statistics consulting firm will help tremendously in explaining and using complex analysis, you are ultimately the one that will be defending it and explaining it to your committee. Just a reminder, a wise person knows their limitations.
For a customized, confidential help formulating your research questions and hypothesis, and/or conducting your statistical analysis, please call Statistics Solutions Inc. (877) 437-8622 for a free 30 minute consultation .
