I have a final in a couple of weeks, and a lot of homework due soon, but I am super confused about how to do a few things, and my professor is not really great at walking me through the steps when I've met with him.

My main questions are:

I have a problem with a box labeling some numbers like so:
n1 = 81
n2 = 80
y1 (with a line above it) = 41
y2 (with a line above it) = 54
s1 = 21
s2 = 19

I can figure out the whole extended problem to get a t-value, which is -4.12 (rounded), and there are df = 157.80

The next step says the formula I want to use is 2*P(t157.80 (less than symbol) -4.12)

My issue is that I have no idea what that is supposed to mean, and my online homework helper simply says "Check your use of technology". Can someone walk me through what I need to do? I am absolutely lost.

And before it is mentioned or suggested, I am trying to do this problem with "homework helper" assistance to figure out the steps first. This means that once I have completed it, the problem will be refreshed and all values will be randomly generated again into a new problem. So assistance will not give me an easy answer. I truly just need to know what to do next. Thank you.

Looks like you are putting in some effort so I think it's legit to offer help with the problem. We don't do that without evidence the person tried to figure it out first.

OK, so how did you get a fractional degrees of freedom? The SD's look pretty similar to me, so I probably would have done a pooled variance t-test, which would have df = (80-1) + (81-1) = 159. Did you have a good reason to use a more complicated formula?

As for the next step, once you get a t value, you need to determine how likely it was to get a t value that large or larger by chance ("larger" in absolute value, which includes smaller in the negative direction, assuming the null). So the formula asks how likely you would be to get a t value < -4.12 (observed t), from a t-table, with your df. Excel will also give you that. But that would be a one-tailed p-value, so you need to double it for two tailed.

Last edited by EdGr; 04-22-2016 at 04:09 PM.
Reason: Noted small misstatement

Oh, thanks for the show-your-work heads up. Did not know that.

Well, this is supposed to be an intro to Stats class, and we're supposed to know the process for doing this already, but I don't, which is why I am looking for a walkthrough.

For this problem I had to to find SE (y1-y2) = ? and plug it into the formula. My formatting-fu is not strong, and I don't know how to get/find symbols, so I won't bother writing it. The answer I got for the SE was 3.1555 (rounded), and plugged that back into the big equation to find t, getting a -4.12. The fractional degrees of freedom came explicitly from the homework problem itself (that's a number it told me to plug in).

Perhaps the problem is that I don't know how to use a t chart? Referring to your last paragraph, I understand that a t table is required, but I don't understand how one uses one. Or how one determines whether a problem requires a one-tailed thing or a two-tailed something. When it gets to here, I don't know what to do at all, and I'm stonewalled.

Most t charts have selected p values at the top, or some indication of a tail probability associated with a particular t value and degrees of freedom. Your exact values will not appear in the chart, so you have to estimate or use an online program or Excel.

For example, a t table in front of me shows a one tailed p of 0.05 for a t value of 1.812 with 10 df. If I had a strong prdiction about the direction of the difference and thus was using a one-tailed test, and if I got t = 1.812 with 10 df, I would report p = 0.05. The probability of a t value as big as 1.812 or larger by chance is 5%. If my t was bigger but not as big as the next t value in the chart, I would report p < 0.05. Etc. I think you need to look into this a bit more if my explanation isn't clear.