I am performing a content based conversational analysis. I wish to know how can T-test be helpful in removing the word 'probably' in my conclusions.
Dear Audil Khaki
Thank you for replying. I have gathered a data which consists of audio recordings of the conversations of target population. I aim to put-forth the differences of the tone, and the time spent talking on a specific topic by two different groups. Do you know how T-test helps and if it will be a good option here? Inability to share details would surely make this discussion less productive. But I still wish to grasp as much as possible.
Dear Abhishek,
T-test is usually used in a context in which we are interested in rejecting or accepting a null hypothesis that the two populations are similar rather than performing a statistical analysis to accept an alternative hypothesis exactly framed opposite to the above presumption. Usually we are interested in statistically approving or disapproving a population statistic (mean) to calculate our T-statistic and either accept of reject the hypothesis accordingly. So far as I can understand from your comment above, you must have got some measurement scales to measure the responses from the 2 different groups that you intend to study or already studied. If such is the case, you are already having a quantitative figures available for all such variables or parameters that you are trying to analyse. I am not an expert in the field but so far as my understanding of T-Test and its applications is concerned, and so far as it fits in what I have already discussed above I think there is nothing to worry about using Student t-test.
All the Best.
Regards,
Khaki
Thanks a lot Audil Khaki.. Your comment has an educational value to offer... Yes, I wonder why do I need the T-test when I already have the basic parameters (from conversations) for comparison... At times, these tests seem to be a way making simple comparisons look complicated. Thank you..!!
Abhishek,
No doubt we can make simple comparisons without making use of any statistical tests. but for making a particular study or for that matter to be accepted by the Researcher Fraternity it needs to carry a statistical validity. It is in order to give that statistical validity to the already simplified comparisons that we perform tests on our Sample/population statistic. And particularly when you are trying to make a comparison between 2 groups you have to render validity to collected data by way of performing t-test in current case. However, measurement of scores needs a rather more cautious handling here rather than making use of t-test. I still suggest going on with t-test if scores have appropriately been recorded.
Best of Luck.
If Abhishek has quantitative data valuable, then only he would be able to compare two groups with student t test
That is true Mr. Fredrick. However, I wonder how structure of normality in a distribution is going to effect the application of T-Test.i.e., as you mentioned taller distribution. All that matters in a t-test is that samples should be normally distributed irrespective of how taller it is..!!
Regarding Significance Level.. It is like with what certaintly are we able to say that a particular arguement is true or false..!! So P values are not either going to hinder application of T-Test, no doubt P values will be there by default after performing the Test.. and that will represent a level of certainty we can accept a particular hypothesis, bbe it null or alternate..!!
@ Abhishek Sharma
To compare two groups in terms of a variable of your interest, you may collect the quantitative raw data of each group members then calculate the average (i.e.; the mean) for each group. The statistical analysis helps you decide whether the noticeable difference between the two average values is REAL and not due to CHANCE. That's why you need it inevitably.
Please, try out the online calculator below and get your statistical test calculated at a glance: http://www.graphpad.com/quickcalcs/ttest1.cfm
The expected t test results would be either:
- [This difference is considered to be statistically significant] = Difference between the two groups is REAL, or
- [This difference is considered to be NOT statistically significant] = Difference between the two groups is due to CHANCE
A very nice guide is included in this website as well.
Good Luck
Dear Drs. Audil Khahi, Fredrick Navarro, Asif Patel and Wafa Habbal.. Thank you very much for your comments. I am happy that I posted the question, the discussion holds an educational value.
I think I will explore further and will be using spss and the web-tool suggested by Wafa Habbal. Will post you if any doubts, and the outcomes. Regards
MS Excel is very helpful as well. Please, let me know if you need any know-how support.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS