You should see two datasets, which were obtained from the same set of individuals. Let's again imagine that these data come from an experiment in which a researcher aimed to determine whether reading a chapter first and then going to a lecture is a better learning method than going to a lecture first and then reading the chapter. To address this question, the researcher recruited 25 people, and subjected each individual to the reading then lecture condition (condition 1) and to the lecture then reading condition (condition 2). The order of these conditions was randomized to make sure that any differences that could be found between these methods is not due to the fact that people had more practice for the second condition they performed. Again, the researcher assessed learning by having each participant complete a test on the material afterwards. In this case, two different sets of materials were used to avoid that people were tested on the same material twice. Column A of the Excel sheet displays the scores for condition 1, and column B displays the scores for condition 2. Step 1: Calculate the t-value for the data set Based on the two datasets, we will calculate the t-value in cell G13. For this, we use the paired sample t-test procedure. Step 1a: MD and D First, calculate the difference score for each individual by subtracting the score in condition 1 from the score in condition 2. Put the difference scores in cells C2 to C26. Calculate the mean difference score and put the result in cell H2. Usually, no difference is predicted between two conditions based on the null hypothesis H0. Thus, D = 0. This is displayed in cell G4. Step 1b: sMD In the next step, we need to calculate the standard error of the mean. Step 1c: T-value Calculate the value for t based on the quantities you calculated in the previous steps. Step 2: Determine the one-tailed and two-tailed probabilities As we saw in the previous lesson, Excel has a build-in function to calculate probabilities with a t-test. This function is called =TTEST(). This function takes four input arguments; the data for the first condition (array 1), the data for the second condition (array 2), whether the test is one-tailed or two-tailed (nondirectional), and the type of test. One-tailed probability test We will calculate the one-tailed probability for the paired sample t-test in cell H16. For array 1, you want to select the data for condition 1 (cells A2:A26. For array 2, you want to select the data for condition 2 (cells B2:B26). First, we will do a one-tailed (directional) test of the null hypothesis H0. To do so, put in the value 1 for Tails. For Type, put in the value 1. This corresponds to a paired sample t-test. Thus, the formula in cell H16 should say = TTEST(A2:A26,B2:B26,1,1). A probability should appear in cell H16 after you hit Enter. Two-tailed probability test Now, perform a two-tailed probability test on the data by adjusting the formula for the t-test and putting the answer in cell H17. Save your work Step 3: Making inferences Based on the probabilities you found for the one-tailed and twotailed t-test, answer questions Q1 and Q2 in cells I16 and I17. Put your Yes/No answer in cells J16 and J17. Save your work Step 4: Comparing the independent sample and paired sample t-test Finally, go back to the previous learning lab and review the results. Compare them to the results you got in this learning lab. You should find that even though the data were the same in the two cases, the result of the t-test is different. Answer question Q3 displayed in cells I19 and I20 by putting your Yes/No answer in cell J20 Condition 1 90 75 63 85 77 45 75 73 85 64 51 65 85 68 75 56 95 76 83 76 75 65 68 80 74 Condition 2 85 65 53 64 65 64 59 63 75 62 68 35 90 76 85 68 62 71 70 70 57 60 76 65 53 X2 - X1 (D) MD - D (MD - D)2 Mean X2-X1 (MD) H0 difference SS df s2 s s MD T value T-test probability One-tailed: Two-tailed: 0 Q1: Based on a one-tailed test with = .05, the researcher would reject H0 Q2: Based on a two-tailed test with = .05, the researcher would reject H0 Y/N Y/N Q3: From comparing the results from this week to last week, a paired sample t-test has more power than an independent sample t-test Y/N