A question about Null Hypothesis for 1 Sample T-test

hazwan2283

Involved In Discussions
Hi to all the respected experts in Elsmar, my name is Hazwan.
I have a question regarding 1 sample T-test.
When we want to do a 1 sample T-test, we need to identify first what is the null and alternative hypotheses.

Let say my first null hypothesis is Mean for the dataset is higher than or equal to 10.
My alternative hypothesis is Mean for the dataset is lower than 10.

Second way of looking into this is..i can also say null hypothesis is Mean for dataset is lower than or equal to 10.
Alternative hypothesis will then be Mean for dataset higher than 10.

My question is, regardless of which null/alternative i choose from above...higher than or lower than...i will get the same result correct?.
The reason why i'm asking this is because i'm using Excel to do this test and i cannot see anywhere in Excel where we can specify whether greater than or lower than.
While i can see from Youtube Minitab videos, in Minitab they can specify whether its greater than or lower than.

I am now wondering why is it important to indicate in Minitab whether greater than or lower than at all?.
 

Miner

Forum Moderator
Leader
Admin
How are you doing a 1-sample t-test in Excel? As you can see below, it is not even an option in Excel.
A question about Null Hypothesis for 1 Sample T-test
 

Miner

Forum Moderator
Leader
Admin
To answer your question about Minitab, I created a standard normal distribution (mean = 0, stdev = 1) and ran a 1-sample t-test for the following:
  1. Ho: mean = 0.5, Ha: mean < 0.5
  2. Ho: mean = 0.5, Ha: mean > 0.5
See the results below. One is significant and the other is not.
  1. A question about Null Hypothesis for 1 Sample T-test
  2. A question about Null Hypothesis for 1 Sample T-test
 

hazwan2283

Involved In Discussions
To answer your question about Minitab, I created a standard normal distribution (mean = 0, stdev = 1) and ran a 1-sample t-test for the following:
  1. Ho: mean = 0.5, Ha: mean < 0.5
  2. Ho: mean = 0.5, Ha: mean > 0.5
See the results below. One is significant and the other is not.
  1. View attachment 29441
  2. View attachment 29442
Hi Miner, i'm using attached excel file to do the One-to-Standard T-test. i have a feeling that excel is not completely correct...isn't it?.
 

Attachments

  • One to Standard Comparison.xlsx
    10.8 KB · Views: 138

hazwan2283

Involved In Discussions
Hi @Miner , i realized now that the t-statistic always a single formula. But the p-value formula differs depending on the alternative hypothesis whether single tail or double tail. I'm now in the midst of doing an excel template for my company for one-sample T-test, two sample T-test, one sample F-test, one proportion test. Since the p-value has multiple formulas..i will be dead by the time i research the formulas..

We are in a super tied budget so not possible to buy any softwares or anything...
 

Miner

Forum Moderator
Leader
Admin
Hi Miner, i'm using attached excel file to do the One-to-Standard T-test. i have a feeling that excel is not completely correct...isn't it?.
I verified that you are calculating everything correctly for a 2-tailed 1-sample t-test.
 

hazwan2283

Involved In Discussions
I verified that you are calculating everything correctly for a 2-tailed 1-sample t-test.
Yes, to my suprise i managed to add formula to the 1-tailed as well (left and right) to cover all 3 possible alternate hypotheses. Thank you Miner
 
Top Bottom