In: Statistics and Probability
A randomized trial was performed to evaluate the effectiveness of a new drug on controlling Type I diabetes in teenagers. A random sample of 100 patients were obtained from the pediatric diabetes clinic at Sick Kids in Toronto, Ontario; 50 were randomly assigned to the treatment group (new drug) and 50 were randomly assigned to the control group (existing drug). You may assume that basic factors such as validity of the inclusion criteria, blinding, etc. were performed appropriately. Baseline information such as age and gender were collected and key outcomes of A1C level and number of hypoglycemic events were measured after four weeks. A1C levels indicate what percentage of your hemoglobin is coated with sugar (glycated). Higher A1C levels indicate poorer blood sugar control and a higher risk for diabetes complications. A hypoglycemic event occurs when the plasma glucose levels become too low; this is a common and adverse effect of diabetes therapy which has been shown to negatively impact on quality of life.
Put the data into SPSS and find out what if there is a statistical significant difference in A1C levels between the treatment and control groups? Run the appropriate test at the 5% level of significance and decide on a 1-tail or 2-tail test.
I do not know what type of test to run with this data
Observe that the researchers consider a group of 100 pediatric diabetes patients, randomly assign half of them, that is 50, to the treatment group of new drug, and the remaining half to the control group of existing drug.
Note that the two groups are unrelated to each other, as completely different sets of people are assigned to the two groups. So, the two groups are independent of each other. Moreover, the population standard deviations of the groups are unknown, and as a result, the sample standard deviations must be used.
Considering the above two points on independence of the groups and unknown population standard deviations, the independent samples t-test must be used.
Although the researchers want to test the effectiveness of the new drug, the final question asked here is, “find out what if there is a statistical significant difference in A1C levels between the treatment and control groups”. Thus, the two-tailed test must be used (as it is not specifically asked whether the new treatment is more effective than the existing drug, or that the new treatment is less effective than the existing drug).
Since the data set is not provided, we are unable to solve the problem for you. However, we can provide the steps, which we are sure would be helpful when you try to solve the question yourself.
Open the SPSS Data Editor window.
Go to Variable View and enter the names of your variables in unique rows, along with their scale, number of decimal places, labels, if needed, etc.
Go to Data View. The variable names will already have occurred as column headers. Enter the data in the relevant columns.
Go to Analyze > Compare Means > Independent-Samples T Test.
Send the relevant columns of variables that you want to test, to the Test Variable(s) box.
[Ideally, you should not enter the data from the two different groups in two different columns. You need to enter all the observations in one column, and in another column, enter the grouping variable, which takes values, say for example, 1 and 2; then, in rows adjacent to all the Group-1 observations, you should write 1 in the grouping variable column, and in rows adjacent to all the Group-2 observations, you should write 2 in the grouping variable column].
Send the variable used for grouping, into the Grouping Variable box. Go to Define Groups, and enter whatever value you assigned to each group.
Go to Options. Enter the Confidence Interval Percentage as 95, as your significance level is 5%, which is (100% – Confidence Interval Percentage).
Click OK in all dialog boxes.
You should get your output in the output window. Since this is a both-tailed test, use the value given in the column “Sig. (2-tailed)” to conclude your results. If this value is less than the significance level of 0.05, reject the null hypothesis, and conclude that there is significant difference between the groups. Otherwise, fail to reject the null hypothesis, and conclude that there is no evidence of significant difference between the groups.