In: Statistics and Probability
Hi, I have some doubs about answers to following questions and I apprecite if someone could help me with the answers: We want to investigate the following research problem: We have been developing a new learning and teaching application and want to know if this application is more efficient on a tablet or on a mobile phone. Efficiency is measured in the speed in which answers in the form of text are entered. i. Formulate the hypothesis H1 and the null-hypothesis H0. ii. We are testing two groups of users (15 people each), one group uses the tablet, one group uses the mobile phone. We want to compare the means of the text entry for the two groups: which test shall be used? iii. The same testing is done, but we only use one group of users for both conditions (of course the conditions are counter-balanced). Which test shall be used? iv. We asked users at the end of the test if they had prior experience with a tablet/mobile phone. How can we report the influence of pre-knowledge on the text entry speed? Kind Regards,
i.
Null Hypothesis(H0):
There is no significant difference between the means of text entry for the two groups.
Alternative Hypothesis(H1):
There exists a statistically significant difference between the means of text entry for the two groups.
ii.
The appropriate test is:
"t- test for independent samples(groups)"
15 users who use mobile and 15 users who use tablet are different and are said to be independent groups. Thus, "unmatched pairs t- test" also called as "t- test for independent groups/samples" must be used for the testing of hypothesis.
iii.
The appropriate test is "t- test for dependent samples/groups".
The same group of 15 users use tablet as well as mobile and thus, they are said to be dependent groups/samples. So, "t- test for matched pairs" also called as "t- test for dependent samples/groups" must be used for the testing of hypothesis.
iv.
Generally, when the samples are selected, the appropriate care must be taken regarding their pre-knowledge of the test to which they are assigned to and accordingly, the sample members are selected. For example, all the members who are almost equally capable of using both mobile and tablet must be chosen and can be divided into two independent groups (in case of independent samples t-test) and if the sample of mobile users are better at using mobiles whereas the different sample/group (unmatched pairs t-test) or same sample (matched pairs t-test) of tablet users are poor at using tablets, then the results obtained from the respective t-tests cannot be trusted as they can be misleading.
And if this care is not taken, i.e.,pre-knowledge is not considered before the test, then the same thing must be reported at the end after reporting the conclusion of the test. It must be clearly stated that these many(say, 8)members who used mobile in test have pre-knowledge of using mobile.