# Paired t test unequal sample sizes?

#### wabao

##### New Member
I want to find if there's a significance between the two levels average mean values from my dependent variable "X," which are "High X" and "Low X."

High X (85 items) + Low X (65 items) = X (150 items)

Both have unequal sample size, but they're from the same group "X." I opted for paired sample t-test, but the output results would only appear N=65 for both groups. I don't think I should run an independent t-test since "High X" and "Low X" technically aren't two unrelated groups since they come from "X." Can anyone help me understand what test I should run? I simply want to know if there's significance between my dependent samples.

#### Karabiner

##### TS Contributor
Could you please describe your actual problem and the actual facts, instead of "X" "low X" etc? The description is unintelligible, I'm afraid, at least for me.

With kind regards

Karabiner

#### wabao

##### New Member
Could you please describe your actual problem and the actual facts, instead of "X" "low X" etc? The description is unintelligible, I'm afraid, at least for me.

With kind regards

Karabiner
So my dependent variable is the persuasiveness of an ad, and so I have a total of 150 ads. From these 150 ads, 85 were categorized as high message persuasiveness (i.e., High Persuasion) and 65 as ads with low message persuasiveness (i.e., Low Persuasion). I already have the average means for the 150 ads (Each ad have their own average means). Now, I need to run a test to find the sig. value of the two groups average (High P and Low P) and also find their standard deviation. I hope this explains my question.

#### Karabiner

##### TS Contributor
You can compare men and women, although both belong to mankind. Accordingly, you can compare ads which were classified as highly persuasive with ads that are less persuasive, altthough they both are ads. I do not know, however, how the classification was carried out and/or how the data on the dependent variable where collected, so there might be a source of depenedency.

BTW, why do you want to perform a test of significance on whether ads classified as highly persuasive are more persuasive than ads classified as less persuasive, on average? Is there reason to assume that the classificiation was totally false, or made 100% by chance? Your test of signficance deals with the question whether the mean difference in persuasiveness is exactely 0.000000000...0000... in the populations from which the ads were sampled. Is the answer to this question informative? "Statistical Significance", as you are certainly ara awre of, has nothing to do with effect size relevance of an effect, or iportance of an effect. Only whether te Nully hypothesis might be rejected or retained.

With kind regards

Karabiner

Last edited: