Statistical Design and Analysis Grid

Statistical Design and Analysis Grid
Name
Institutional affiliations
Statistical designs and analysis grid
Statistical Test Definition Strengths Weaknesses Examples of where/how used
Correlations Correlation is a statistical method that is used to describe and measure the direction and strength between two variables (Goos & Meintrup, 2016).
It requires two results from the same sample. Correlation enables researchers to gather more information compared to experiments (Goos & Meintrup, 2016).
The results of correlation are applicable in day to day life since the research occurs outside the lab.
Correlation provides room for further studies by other researchers, and it may be easy to find results and cause of relations. One problem with correlations is that they show relationships but not their cause (Graham, 2008).
Similarly, the finding of a correlation does not reveal the variable that influences the other.
Unknown variables may influence others in a relationship causing confusion (Graham, 2008). Correlation shows whether a relationship is negative or positive (Graham, 2008). It is an appropriate method of showing quantifiable data.
t-tests
(student t-tests) These are analysis methods where two populations use statistical examination (Goos & Meintrup, 2016).
Two samples are used for the test with small sample sizes, and the variance determined even when the normal distributions are not known (Rencher & Schaalje, 2008). In research, t-tests are easy to calculate since computers are used.

These tests allow for ease of gathering information as little data is required (Rencher & Schaalje, 2008).
T-tests also accommodate robustness as two populations are normally distributed with similar variance.
The results of these tests are easy to interpret. A researcher may determine the significance level and therefore can reject or use the null hypothesis based on their decision (Rencher & Schaalje, 2008). As such, a researcher can influence the outcome of their study.
The results of research are only true for normal populations except for real samples. The t-test is used when the sample population is unknown, or the size is less than 30 (MacDonald, 2007).
ANOVA A statistical method that is used to determine how two or more sets of data vary in an experiment (MacDonald, 2007).
When there exists a large variance, then research is said to be significant. Anova is reliable when using more than two samples (MacDonald, 2007).
It can be effectively used even in cases where the observations in each group are different. ANOVA does not accommodate missing data and non-Gaussian dependent values (MacDonald, 2007).
Also, it forces problems that are difficult to study using experiments. One way ANOVA is used in the comparison of more than two groups using a single factor (Goos & Meintrup, 2016).
Two-way ANOVA is used when comparing two variables, for instance, employee conditions and hours when determining their productivity.
Chi-Square A statistical measurement that shows the comparison of expectations and results (Qian, 2009). The data used in summation should be raw, random, and mutually exclusive and derived from independent variables (Qian, 2009). It can be used to determine associations between variables (Graham, 2008).
Also, tt differentiates between expected and observed values (Qian, 2009).
It is easy to compute values.
It can be used on data that has been measured on a nominal scale. It does not use percentages.
The tests get invalid when less than 5 values are used (Graham, 2008).
Data used should be numerical.
Also, it is difficult to get the right formula Chi-Square test is used in testing data dependence. Here, random sampling should be used, variables should be categorical, and at least five sets of data should be used (Graham, 2008).
Works Cited
Goos, P., & Meintrup, D. (2016). Statistics with JMP: Hypothesis Tests, ANOVA and regression.
Graham, A. (2008). Statistics. Blacklick, OH: McGraw-Hill.
MacDonald, T. H. (2007). Basic Concepts in Statistics and Epidemiology. Oxford: Radcliffe Pub.
Qian, H. (2009). On data-driven chi square statistics.
Rencher, A. C., & Schaalje, G. B. (2008). Linear Models in Statistics. Hoboken, N.J: Wiley-Interscience.