Reflection paper: My thoughts and experiences with statistics and the stages of the research process

I have often seen individuals and organizations shy away from inferential statistics. They prefer to perform all their analysis using descriptive statistics, summarizing and describing the data only (Tokunaga, 2018) because it is easier to explain and less “complicated.” Besides, if a client doesn’t demand inferential statistics, why go through the additional hassle? Yet, I have seen other individuals and organizations use inferential statistics and other advanced methods to serve their clients and carve a niche for themselves. In my opinion, the use of inferential statistics is efficient. It helps the researcher get as much insight as possible from the data and provides detailed guidance than frequencies, means, other measures of central tendency, and descriptive statistics.

Conducting efficient research is also one reason why the planning stage of a research project is crucial. It is when the researcher has to decide the depth of analysis they will be conducting. When designing the data collection instrument, the researcher designs it with this proposed analysis plan in mind. The researcher will make sure that the questionnaire has all the variables of interest. For example, forgetting to code gender or state would be problematic — if the researcher needs those variables for analysis.

The planning stage is also when the researcher finalizes the research question and decides if the research hypothesis will be directional or non-directional. The way researchers choose to measure their variables also influences how they state their research questions and hypotheses (Tokunaga, 2018). Researchers can also be led by the norm and format used by other researchers in their field. Researchers should also be prepared to provide a logical and scientific justification for the choices they make.

For some research studies, there are so many confounding factors to be controlled. Although experimental approaches such as using a control group and random assignments (Tokunaga, 2018) attempt to answer these kinds of questions, they are at best still a proxy. I think a proxy is good when there are no other valid alternatives. Is there any research, especially social sector research conducted with humans as the subject, that isn’t riddled with confounding factors and all findings at best proxies?

Interpreting results when using inferential statistics is also important. Although statistical analysis programs automatically perform many analyses, learning what the calculation does without using a program helps the researcher interpret the statistic better (Tokunaga, 2018). The level of measurement also influences interpretation. With two ratio variables, a regression output is easier to state and understand than a regression output where the independent variable is a nominal or categorical variable.

In conclusion, I agree with Tokunaga that the research process is not “a free ride but rather is filled with starts and stops, dead ends, and wrong turns.” It is a tedious process; therefore, it is important to study something you consider interesting if you have the opportunity to choose. Your interest in the topic will be a constant source of motivation and help you put in appropriate rigor and work when designing your research and throughout data collection, analysis, and reporting.

References

Tokunaga, H. (2018). Fundamental Statistics for the Social and Behavioral Sciences. Second edition. Reflections from: Chapter 1: Introduction to Statistics.