As we mentioned earlier, it is important to not just state the results of your statistical analyses. You should interpret the meanings, because this will enable you to answer your research questions. At the end of your analysis, you should be able to conclude whether your hypotheses are confirmed or rejected. To ensure you are able to draw conclusions from your analyses, we offer the following suggestions:
- Highlight key findings from the data.
- Making generalized comparisons
- Assess the right strength of the claim. Are hypotheses supported? To what extent? To what extent do generalizations hold?
- Examine the goodness of fit.
Your conclusions could be framed in statements such as:
“Most respondents …..”
“Group A (e.g., Young adults) were more likely to ___than group B (older adults)
“Given the low degree of fit, other variables/factors might explain the relationship discovered”
Access and Organize the Dataset
- I have checked whether an Institutional Ethics Review is needed. If it is needed, I have obtained it.
- I have recorded all the ways that I manipulated the data
- I have inspected the data set and have noted the limitations (e.g., sampling, non-response, measurement, coverage) and have inspected it for reliability and validity.
- I have inspected the data to ensure that it meets the requirements and assumptions of the statistical techniques that I wish to perform
Cleaning, Coding, and Recoding
- I have re-coded variables as appropriate.
- I have cleaned and processed the data set to make sure it is ready for analysis.
- If it is secondary data I am using, my methodology has documented their method for deriving the data.
- My methodology documented the procedures for the quantitative data analysis.
- I have highlighted my research questions and how my findings relate to them
- I have reported on the goodness of fit measures such as r2 and chi-square for the likelihood ratio test in order to show that your model fits the data well.
- I have not interpreted coefficients for models that do not fit the data.
- I have not merely provided statistical results, I have also interpreted the results.
- You must test relationships. Univariate statistics are not enough for quantitative research. Make some inferences supported by tests of significance. Correlations, Chi-square, ANOVAs, Regressions (Linear and Logistics) etc.
- I have stored all my statistical results in a central file which I can use to write up my results.
- My tables and figures conform to the referencing styles that I am using.
- Report both statistically significant and non-statistically significant results. Do not be tempted to ignore the non-statistically significant results. They also tell a story.
- I have avoided generalizations that my statistics cannot make.
- I have discussed all of the relevant demographics