A t-test is used to determine, compare or asses whether the mean of two groups / samples statistically differs from each other. This analysis is significant when comparing the means of two groups. It is most suitable for analysis of two group’s post-test experiments. There are two types of t-test that can be used based on the assumption made. The first one is the paired t-test and the second one is the independent t-test. Independent t-test is further grouped in to two i.e. pooled independent t-test, and separate independent t-test (Student’s t-test, n.d.).
The difference between matched (paired) and independent samples is that paired samples t-test can be used when testing the same person twice. For instance, when a person is given a pre-test and a post-test, the scores are compared at the two different times to determine whether there is a significant difference between the pre-test scores and the post-test scores, whereas independent samples of t-test can be used when testing different people, such as men versus women or smokers verses nonsmokers etc. According to Trochim (2012), scores from the two samples are analyzed to determine whether there is a significant difference. In this case, it doesn’t matter whether the sampling is random.
From the table of variables, a good example of how I would conduct a t-test would be a scenario where the three people i.e. one male and two females undergo a liver transplant and a t-test is conducted to analyze the body weight loss of a man and women after treatment. Firstly, I would conduct a normality assumption by evaluating the distribution of the data through histograms or alternatively I would perform a normality test. Secondly, I would do an F test or levene’s test to determine whether the two group’s variance is significantly different. According to the theoretical assumption, I would expect the paired differences to be normally distributed. The report from the t-test will give the probability of error needed in accepting the research hypothesis i.e. the existence of a difference. If the variances are significantly different then I would expect the sig to be less than or equal to (<=.05), but if it is greater than (>.05) then I will assume that the variances are equal. Lastly, I would determine the 95% confidence interval associated with the mean difference between men and women for weight loss (What are Basic Statistics, n.d.).
In the arrangement of data I would have one independent variable (grouping) e.g. male / female and one or more dependent variables e.g. Body Mass Index which are mandatory to calculate the mean.
The grouping variables are;
- Gender
- Ethnicity
- Case
The continuous variables that I would include in the analysis are;
- Pre-BMI
- Post-BMI
- Age
- Days-tx
- Weight loss
In this analysis I would use the independent t-test. This is the most appropriate t-test for the analysis since we have two independent groups i.e. males and females.
Fig. 1. Data set and t-test example.
References
Student’s t-test. (n.d.). Student’s t-test. Web.
Trochim, W. M. (n.d.). The T-Test. Social Research Methods. Web.
What are Basic Statistics. (n.d.). Big Data Analytics, Enterprise Analytics, Data Mining Software, Statistical Analysis, Predictive Analtyics. Web.