Paired Samples t-test - SPSS
HTML-код
- Опубликовано: 21 авг 2024
- I demonstrate how to perform and interpret a paired samples t-test in SPSS. I also point out that many people fail to test the homogeneity of variance assumption in the paired samples t-test, but that this can be done relatively easily with a Pitman-Morgan test.
paired t-test
You are the man. I went from zero to top of my class. I am in Graduate School for Public Administration and my first day of research design, I sat to far away from the screen to learn what she was doing. Also, she was a very fast teacher throughout the year. It was a 3 hour class and I wanted to quit school after the first. Thought I was to old and just beyond my capabilities. I have been using all of your videos throughout the class. I went from knowing nothing to being the spss guru for the class. Thanks man. I hope your making money off these videos because you deserve it.
im starting to feel like a spss god
I owe you my life. My exams are coming up and you have saved me, keeo making more
I'm in research methods, this saved me! Thank you so much.
I wouldn't bother with z-scores in this case. For the purposes of estimating the p value, you could use the Bootstrap utility that is part of the paired samples t-test in SPSS. Or you could use a non-parametric approach such as Friedman's ANOVA.
Thanks Sir
Thank you so much. Explained so well. Greetings from Stanthorpe QLD Australia 🇦🇺
Thanks for a great video! I don't believe you actually need to worry about homogeneity of variance with a paired-samples t-test, though. The denominator of the paired-samples t statistic is the standard error of the distribution of differences between the scores, not the distributions themselves. Since we are only concerned with the distribution of differences, there is only one variance term to worry about, and therefore no need to worry about homogeneity.
6:34 the p value is on the top in the properties box.
you really help me a lot. Thank you so much for making this tutorial :)
I am trying to compare resistance to change in an organisation against three other variables: trust, communication and participation...how can I use the t-test if each variable is calculated using 4/5 sets of questions (answered using a Likert scale)? Do I add them all up?
Thank you so much, great video!
Incredibly helpful! Thank you so so much
Hi Sir, first thank for the valuable video, Second as I am new to SPSS, I have a question, that why the sample size is more than 30 in t test.
Thanks
Thanks for this video help me for master thesis
Hi!
Please help me with this!
I need to see if the relationship partners have similar personality traits and then correlate that result with dyadic satisfaction.
How do you recommend me to do that? I already operated the paired sample t test.
Thank you! from Greece
saved my degree lol. your the best
how can I run a paired T test for 2 different groups as they are correlated.
I'll be more precise, I recruted different people from different (let's say 2) centers, than test them for an X outcome. I think that the fact that they're recruited from 2 different centers, will make the people's characteristics from the same center more correlated. So my first question, If I want to compare the mean of age between the X negative and X positive persons do I count for the correlation and use a paired T test and if so, how will I do it on SPSS? or am I wrong and I don't know it? because I kind of feel it.
thanks in advance, great videos.
Hi there,
Just wanted to know how I might go about dealing with a situation in which I have an extra time1 measure for one of the groups (or you can consider it a missing time2). In my case, I'm trying to compare a group of 33 with a group of 32--I'm aware I can omit the extra 33rd data point but is there a way I can still take it into account?
I was suggested that mixed models was the way to go but I don't quite know what the levels/groups would be here. Or maybe there's another way.
Thanks!
Pp
Hello doctor. If I want to see the change in sales before and after the covid19. Then should I use paired sample t test or any other test. Can you please suggest.
but when you introduce a control group, is that still a paired t-test?
lmao bro u just saved my life god bless u
i cant thank you enough!!
Sir it's quite helpful but how did u fill the answers from questionnaires into the data view.
if I have 15 people in a survey and I have 17 questions on both type of my questionnaires pre & post,both have MCQ type of questions,marked by participants differently..some people marked a.) some marked b.)...so how do we enter these answers of people onto the data view...did u calculate the mean for all the questions first & then enters their means into first two rows of data view? if u did so,then where did u enter the sample size?? PLEASE CLARIFY AS SOON AS POSSIBLE...I'm in a mist of completing my dissertation but don't know how to fill answers market by people on the data view sheet of spss. We can calculate the mean easily by spss because the entries of the questions & sample size are quite easy...but I don't understand how to fill the answers of questionnaire on the data view to analyse the paired t test.Pls reply😳🙏🙏
Can SPSS perform paired sample t-test if the pretest is continuous data (100 pts) but the posttest is ordinal data(ABCDE or 54321) ?
Can you please tell me what happens if I had 55 students from a class that I taught, I took pre-test scores, but only 35 participated, then I did 2 months of teaching and now 40 participated in post-test, and let's say that only 30 are the same (pre and post), can I assume that they came from the same population so I can use all data or should I only use data where their initials filled out on the forms are the same (exclude those who weren't there in pre-test) and have a much smaller sample?
And if I can assume the first; they came from the same population (they were in the same class), should I use paired or unpaired sample T-test considering some who are in post, didn't fill out pre (and vice versa)?
Sir you have sample of 50. Why at many places it is written t test should be used with below 30 sample.
Or t test can be used with n no of samples, doesn't matter. Is that so?
You have said nothing about the rest of the table... What does the confidence interval stand for?
Can we calculate paired sample t-test on the average of two variables but the sample is same.?
Hello. What does negative value in the mean shows? Is it fine to get negative value when we are doing pre test minus post test?
may i ask a question sir? when is the exact time we need to use t-test?
What does it mean when you have a negative t-value (compared to having a positive t-value)? I know it have to be mentioned when reporting the results. But what negative t-value means for the data? What information it gives to the reader of the report?
Thanks & my apologizes if that was already explained in the video... English is not my mothertongue.
No need to apologise. It's a perfectly good question. A negative t-value means that the second mean was larger than the first mean. In the video's example, the resilience2 mean was larger than the reslience1 mean, and the t-value came out negative; had the resilience1 mean been larger than the resilience2 mean, the t-value would have come out positive.
Great thanks for the quick answer! So the negative t-value has nothing to do with significance, at least you can't say that result is/is not significant only because of negative t-value. This was a lot of help :)
so why so many people use t value to explain significant level. I have seen some people quote result is not statistically significant when t value is negative.
so what does that mean? if our p-value is
A very small p-value is perfectly possible. It doesn't mean that you did anything wrong.
@@how2stats ok thanks, really helpful
you rock sir, thank you!
Record for the most statistically significant result ever 😂
Very handy, thank you!
Chech this vid
watch?v=nFqq9FWzxF0
It's about how to perform paired-samples t-test in Doronix Math Toolbox
What if I have one sample, that has answered a dependent measure about regarding two possible conditions?
Say, the sample rates the truthfulness of 10 medical reports. 5 are true, 5 are false (however, only I know this).
participants were randomly assigned and presented all articles. If I calculate the mean scores for each group of 5 articles, and compare the differences, should I use this t-test?
I'm thinking yes, but not 100% sure
Participants were randomly assigned to what? Two different groups?
I worded this wrong - the sample was randomly selected from the population, who were exposed to both possible conditions (however, they did not know there were two possible conditions).
need advice :let say I have pre test and post test data in %.. do I need to convert the data into Z-Score before doing paired t test?
If I want to compare two conditions but I also have two groups, should I use split file. Because if I do not I get a significant p value, if I do I only get a significant p value for one of the two groups.
Yeah, you loose statistical power, when you split your sample. My hunch is that you might be interested in conducted a mixed-design ANOVA, as it is well-suited to test hypotheses with your data: ruclips.net/video/APvlPjYSSaI/видео.html
Thanks for your quick answer, I was also looking into doing a repeated measures ANOVA. The only problem is that my data is not normally distributed, I did a wilcoxon for the condition effect and a mann-whitney for the group effect. Could I still be able to measure some kind of interaction or is a repeated measures ANOVA my only option?
goodDay Sir, If the p-value is 0.0000000000 does it always meant that the paired variables are statistically significant? and to how to cite/state that the null hypothesis is rejected? thanks ahead
Yes, the null hypothesis has been rejected when p = .00000000. Report as p < .001. Check out how to write-up the results here: ruclips.net/video/LPOEdtw7Sjo/видео.html
may I also asked what statistics to be used if this the null hypothesis, "There is no significant effect of _____________ on the Math Review". And what do you think is/are the variable fits to my null hypothesis? These are the sources of our data;
Students Profile:
-Gender
-Study Habits
-Mathematics Grades
Teachers Profile:
-Trainings attended
-Highest Educational Attainment
-Length of Service
I’m not finding the information for testing homogeneity of variance for paired samples on your blog. Can you direct me to it?
It's the Pitman-Morgan test: www.how2stats.net/2011/06/testing-difference-between-correlated.html
Can you do a paired samples t-test when your data are ordinal? My data are reading levels at time 1 and time 2, but the data levels are A, 1 2,3,4,6,8,10, 12, 14, 20, 24, 28, 30, 34, 38, 40, 50, 60, 70, 80. They appear to be continuous but they are not. Can I still use a paired t-test? These are the levels from k - 8th grade.
+Lynpe You must compute a new variable. You want to get a mean, then run the test.
+Lynpe A month later is no help, feel free to email me.
Hey, instead of showing a 2 tailed test how do I change it into a 1 tailed test?
To my knowledge, it's not possible to get SPSS to report a one-tailed t-test. However, you simply have to divide the reported p-value by 2 and that will reflect the one-tailed probability. Alternatively, you could get SPSS to calculate the 90% mean difference confidence intervals, instead of the default 95%, and I believe that would essentially reflect a one-tailed t-test.
how to do t test for just 2 data of adjusted r2?
How about normality test before t test ?
is it necessary that the data must be normal distribution???
Yes. Imagin the distribution of wealth in any country. It is not normally distributed. Now, how would you describe this distribution? You would need to say something about the average (mean), the variation in income (variance) AND the skew (How far to the left/right is the "blob" of most people?). If you describe this distribution only with the mean and the variance to someone, he is thinking of a very different distribution than you are. You are thinking of a distribution that looks like a slide. He is thinking about a hill/pyramid. The chance to become rich would be very different in both scenarios.
Since the T-test is only working with the variance and the mean you are misleading others (and yourself) into thinking about a very skewed distribution as a normal distribution. You would missunderstand the chances of any given person to get rich, how many people are likely to have X amount of income and so on.
Is there any connection between statistical significance of the result and t value.
Yes, a very close connection: as the t-value increases in value (positive or negative), the smaller the p-value becomes.
You can even eyeball the t-value as a rough estimate wheter you will obtain a significant result. if t is greater than 2 You will reject the null at a significance level of 0.05 in most cases.
Good day sir, I'm having a bit of a trouble. I'm not sure if I used the right statistical test. I want to know the difference between levels of offense in swearing by comparing the level of offense made when parents swear toward an individual vs. the level of offense made when friends swear toward an individual. Can I use a Paired T-test?
Gwyn Minoza o
How to take it within group??
Is it okay to use it in experimental research with post test only group design?
Sounds more like an independent samples t-test scenario that you've got there.
@@how2stats But we have two groups, which are experimental and control group. What do you think?
Whay my significance is two sided🥲
one sided p = 0.002 and two sided p = 0.003
Epic thanks
Thank you for the advice"
How do we do this analysis for two groups? (expe and control group)
If you only have two groups, independent samples t-test; if you have two groups plus a within-subjects factor (e.g., pre-test; post-test), mixed-design ANOVA is an option.
thank you for the video ;)
thank you by Portugal...:)
B.Statistics brought me here.
Thank you :)
Is the t the t obt?
That's correct: obtained t-value.
saved my gpa
If you're a psych major please tell me it gets better than this
thanks for explaining absolutely nothing