Previously I was using your videos as reference tool but now I consider them as the main source of learning and inspiration. Very lucid, understandable, and to the point. Thank you so much!
thank you so much this is extremely helpful! My only remaining question was how to get a one-tailed result but I have just found the answer… so if anyone else needs to know it's quite simple. To convert a two-tailed result into a one-tailed one you divide the result by two (ie: the number under "Sig. (2-tailed)") (according to Field's SPSS introductory book so hopefully that is correct…)
Please dont take this video down its so good. I ran the same test setting the groups variable as nominal and got the same results as you so it dose not matter.
Thank you very clear and better than my tutor :D. A little side note: you may need to cut the Cohen's d of the video as the new version of SPSS calculates it when running the t-test
+Laura Menezes Yes, so long as skew is less than 2.0; check out my video "Does the t-test and ANOVA really assume normality?" ua-cam.com/video/yNdlGRz-Z04/v-deo.html
if there is no dependent variable in the list of variables, because in survey only independent variables are included with questuions, then what is the solution because in regression, ANOVA and t test it is compulsory to enter dependent variable but it is not in the list so what is the solution now?
Consult the results associated with the 'Equal variances not assumed' row in the t-test results table. It's called Welch's t-test and it doesn't assume equal variances.
What if you want to compare the proportion of smokers in your sample to the national proportion of smokers? How would you be able to compare the proportions since the smokers are only labeled as 0 or 1?
the test doesn't generate anything for me because it says that at least one of the variables are empty. I don't know why. Also the value lables are not changing.
You probably labeled your variables from. It DOES matter if you use 0,1 or 3,5 for example. If you have different grouping numbers then SPSS cant read that specific group hence it saying one is empty. But this is probably way too late
Thank you, this was very helpful and you were great. What if the Sig value (under Levene's) is less than 0.05, which you mention is not good, how do we proceed? Do we read the data from the second row (equal variances not assumed)?
+mghusson Yes, that's a good option. SPSS doesn't say it, but the second row is Welch's t-test (it can handle unequal variances and unequal sample sizes).
Thank u for educating... can u plz also tell how to form a norm group in SPSS 17 of a personality test. I have data of 200 people and test has 35 variables. I have got the zscore and tsore and I do not know the next step... basically need to make a norm group and name it. Its really urgent for my project. Plz help me on this.
Hi, I really enjoy your videos!!! I have a big question... I have a group with depression and after a couple of months we did an exercise test to the whole depression group... but at the time of the exercise test, only half of the group have depression. Should I compare the groups? Or do you recommend another approach? Thanks a lot!!! I am desperate ...
There are two confusion here! (1) sig. two tail, it suppose to be p value right, and p value and signifincant level (alpha) they are different right, why used it interchangely? (2) If the equal variance assumed and equal variance not assumed equal to 0.004, which less than 0.05, then what's the conclusion?
+Yeap Eugine Yes, sig. two tail is the p-value associated with the t-test. Alpha is different. Alpha is the maximum error rate you are willing to accept in the analysis. Usually, people specify alpha at .05. They then look to see if the p-value associated with the t-test is less than alpha (i.e., .05). If t-test p value is .004, it's less than .05, thus, the conclusion is to reject the null hypothesis of equal means.
Thank you for your useful video. I have followed your steps, however I cannot see the Sig (2-tailed), instead the table presented One-side p (0.242 and 0.250) and Two-side p (0.484 and 0.499). My question is how can I draw findings with these information? thank you so much.
I'm a little confused as to why the significance for the F value has to be above .05. What is the actual null hypothesis for Levene's homogeneity test? Is it "the samples are from different populations?"
and why do we need to use the t-test if we already have the correlation coefficient? I'm sorry about the beginner-ish questions..I'm completely new to statistics
I have created weights/expansion factors for my dataset (to match the survey universe). My weighted vs. non-weighted results are very different. I assume that it is best to use the unweighted/unexpanded data?
Doesn't the p value only need to be greater or less than +-.025 to have the difference of the population means be declared significant? You were saying .05, but it's two tailed so wouldn't it be half that?
No. A one-tailed test would include all of the "alpha" on one side of the t-distribution, which would correspond to a smaller critical t-value. You would still report a significant result (greater observed t-value) as p < .05. A two tailed t-test distributes the .05 alpha equally to both sides of the t-distribution and leads to a larger critical t-value. Again, though, an observed t-value greater than the critical t-value would still be reported as p < .05, in this case. This webpage might help you understand: statisticsbyjim.com/hypothesis-testing/one-tailed-two-tailed-hypothesis-tests/
Hello. Thank you for this video. I have a question. I want to compare if a numeric variable A has an impact on a variable B which has 3 numeric variables, B1, B2, and B3. What I want to do is to demonstrate if A has an impact on B1, B2 or B3. Should I run a t-test or just correlate if A has any impact on B1 and then if A has an impact on B2 and the same with B3 ( analyze->correlate-> bivariate) . Which of these possibilities is the correct one? I just started using SPSS and learning statistics and I really need this to know for my thesis. Thank you very much.
why is there a equal variances assumed and not assumed and how did SPSS got that answer. and why is there levene`s test? i also did not understand upper and lower.??
Sorry if my question is elementary: what is the meaning of the t-value and the degree of freedom? We keep looking at significance but not these values, what are they for?
Hi thanks so much for your video! Could you explain when we should look at data in the row for "equal variances assumed" and when we should use the row "equal variances not assumed"? I used a Shapiro-Wilk test to see check my data is normally distributed, before using the T-test but it seems that the values I get under the column "Sig. 2-tailed" are all not significant anyway...
It's impossible to have a negative significance (p-value). A negative t-value simply implies that the second mean is numerically larger than the first mean. However, it's essentially arbitrary which mean is mean #1 and which is mean #2, so, you just have to look at which mean each group belongs to and interpret the results accordingly.
In this context, you would use a non-parametric test if the _dependent_ variable was measured on a nominal scale. In the independent samples t-test, the independent variable is always nominal in nature and the dependent variable is interval/ratio in nature. In this case, the dependent variable was brain volume, which is continuous (interval/ratio).
The smoker variable is just the grouping criteria so that the software knows which case belongs to which group. Thats not the variable which is being analyzed in the t test. The variable being analyzed is the scale one...
+MIBCN2008 With a calculator: Cohen's d = (Mean 1 - Mean 2) / ((SD1 + SD2) / 2) If you're keen, you can get eta-squared through the GLM menu option in SPSS
Many thanks for this easy ablicable lec. just one question ,,i tried it in my data the result of significance give me 0.000 is this possible or some thing wrong ?
No, nothing is wrong. If you push out the decimal place limit in the output, you'll see a number eventually. People report 0.000 as p < .001. You rejected the null hypothesis.
I would like to ask, if i want to do the one-tailed F test on this, what the Nul Hypothesis and the alternate hypothesis will be? Many thanks in advance ^^
If someone can help me, I have to measure an Independent Variable with two levels (control group and experimental group) and also a Dependent Variable with two levels (reaction time and mistakes made during test). What test should I use, and how should I put the values?
i have same problem.. an expert in this field suggested me this.... "The type of data you have may determine the statistical approach. Replicated plot data, for example, lends itself to statistics that assume a "normal" distribution. Statistics like mean, standard deviation, analysis of variance, etc. It may be appropriate to use non-parametric procedures if data is skewed or "messy". Median and quartiles or quintiles may be more appropriate and better reflect the data frequency distribution is skewed. Some of these tests use the data rank rather than the mean to identify differences. Examples are Mann-Whitney test, Kendall test-for-trend, Sen's slope estimator, and Kruskal-Wallis non-parametric analysis of variance. These tests can identify significant differences, but are not necessarily able to quantify the difference. Scientists in psychology or wildlife biology collect data that is not replicated and certainly "messy", but are able to make statistically valid conclusions. It appears that the data you have collected may not necessarily be replicated, but you could use a non-parametric statistical approach". hope this helps.
Dear How2stats, Thank you so much. You are a real life saver. I'm currently working with an spss file created by another, who defined male and female and 1 and 0. Which is fine. I would like to know though, how can I have it so that the group is described not as 0 and 1, but the word male and female in the output file. I hope the question makes sense? I just don't know how to back track and change the original file from 0 and 1 to male and female. The information was verbally relayed, but would help me if I could have the word in the result! Thanks.
eduffles Just got it! When I double click on the table, in the output file, directly on the group 0 or 1 (whichever I want to change), I can type in whatever the true description should be.
In one sentence
I LOVE YOU
You just saved a soul from failure
Thnk u sir
That's more than one sentence, but I'll take it. You're welcome!
6 years later and he's still saving souls from failure LOL
So a free video on UA-cam is way better than what we paid for in Uni ? What on earth is wrong with this world
Previously I was using your videos as reference tool but now I consider them as the main source of learning and inspiration. Very lucid, understandable, and to the point. Thank you so much!
Thank you all these years later. I am working on my DNP and this is a required course. You helped me so much!!!
By a large margin the best approach to teaching SPSS! Truly grateful for your help!
Of all the videos I've seen, this was the most helpful in dealing with two groups with unpaired data
I wish you were my Statistical Applications teacher! You just made my life ten times easier
Love your videos. I use them all the time. You're gifted at making something easy to understand for anyone!
This was the best example of conducting an independent samples test!!!! Thank you!!
Using this for my dissertation after my supervisor recommended it. Thank-you.
Thank you for saving my stats in undergrad research... Thank you again for my postgrad in progress 😊
Very thorough, explained very good, more importantly I understood you clearly. I am looking forward to viewing more of your lessons, THANK YOU!
Thank you for this clear explanation. This video saved me from calling my professor.
thank you so much this is extremely helpful! My only remaining question was how to get a one-tailed result but I have just found the answer… so if anyone else needs to know it's quite simple. To convert a two-tailed result into a one-tailed one you divide the result by two (ie: the number under "Sig. (2-tailed)") (according to Field's SPSS introductory book so hopefully that is correct…)
Cohen's d and a few others are now available on SPSS. It'd be great if you could make a video about reading the new additions to this! Love the video!
It's been so helpful thank you!! So hard to find good videos that help to interpret the output...Thx again, you'll save my grades!
Please dont take this video down its so good. I ran the same test setting the groups variable as nominal and got the same results as you so it dose not matter.
THANK YOU A LOT.
STILL USEFUL THESE DAYS.
I have a Stats exam this next Tuesday so for the next week I'm gonna abuse your youtube channel!
Thank you sir!
Thank you very clear and better than my tutor :D.
A little side note: you may need to cut the Cohen's d of the video as the new version of SPSS calculates it when running the t-test
thanks for the video... it really did help me in understanding the main concept of t test in spss..
Great professional presentations, Very GRATEFUL FOR YOUR HELP.
I was badly stuck n didnt find a gud video until i found your's so so perfect. Thankssss alot
Hope this helps me in ma finals tomorrow
Did it work?
@@onmyown9552 i guess it didn't
In SPSS 29, Cohen's d is not included in the output of any independent sample t-test
this was so helpful even in 2020
I am confused... is it Accept the NULL? Because there is no significant difference between the smokers and non smokers?
No, the difference between the smoker/non-smoker means was found to be statistically significant, so, the null hypothesis was rejected.
Thank you for saving my dissertation!
I thank you from the bottom of my heart :)
This is a good tutorial .and helpful tutorial using SPSS independent t tests
Thank you so much, your explanation made this really easy to understand
Thank you so much sir for this data , May I get another tutorial about f- test please
if the scale has skweness >1 can u stil use t-test? or is mann whitney?
+Laura Menezes Yes, so long as skew is less than 2.0; check out my video "Does the t-test and ANOVA really assume normality?" ua-cam.com/video/yNdlGRz-Z04/v-deo.html
Have skewness= 1.388 kurtosis 2.142 and shapiro-wilks p= 0.001
it is very helpful in understanding the t-test operation
if there is no dependent variable in the list of variables, because in survey only independent variables are included with questuions, then what is the solution because in regression, ANOVA and t test it is compulsory to enter dependent variable but it is not in the list so what is the solution now?
great vid cus its exactly what we did in my stats class and its a great review thanks
Can we get a total SEM just like we use ANOVA beacuse when I use t-test I only get SEM of two groups only and not total SEM. Please help
Thanks for great video
What if the leven's test shows different variance in two groups?
What test we can use then?
What to do if Levene's test for equality of variances are significant? What alternate could be done?
Consult the results associated with the 'Equal variances not assumed' row in the t-test results table. It's called Welch's t-test and it doesn't assume equal variances.
What if you want to compare the proportion of smokers in your sample to the national proportion of smokers? How would you be able to compare the proportions since the smokers are only labeled as 0 or 1?
another soul saved, thank you very much!
Thank you for this video it is very useful and has helped me with my assignment.
Great explanation. Thanks a ton!
the test doesn't generate anything for me because it says that at least one of the variables are empty. I don't know why.
Also the value lables are not changing.
You probably labeled your variables from. It DOES matter if you use 0,1 or 3,5 for example. If you have different grouping numbers then SPSS cant read that specific group hence it saying one is empty. But this is probably way too late
Thank you, this was very helpful and you were great. What if the Sig value (under Levene's) is less than 0.05, which you mention is not good, how do we proceed? Do we read the data from the second row (equal variances not assumed)?
+mghusson Yes, that's a good option. SPSS doesn't say it, but the second row is Welch's t-test (it can handle unequal variances and unequal sample sizes).
YES I THINK
please keep a last description on chart
Thank u for educating... can u plz also tell how to form a norm group in SPSS 17 of a personality test. I have data of 200 people and test has 35 variables. I have got the zscore and tsore and I do not know the next step... basically need to make a norm group and name it. Its really urgent for my project.
Plz help me on this.
No, it is not. I can't think of any case where it would be necessary to convert data into z-scores for the purposes of conducting a t-test.
you just saved me a lot of time for my Bachelor Thesis. Thank you very much!!!!
Hi, I really enjoy your videos!!! I have a big question... I have a group with depression and after a couple of months we did an exercise test to the whole depression group... but at the time of the exercise test, only half of the group have depression. Should I compare the groups? Or do you recommend another approach? Thanks a lot!!! I am desperate ...
There are two confusion here!
(1) sig. two tail, it suppose to be p value right, and p value and signifincant level (alpha) they are different right, why used it interchangely?
(2) If the equal variance assumed and equal variance not assumed equal to 0.004, which less than 0.05, then what's the conclusion?
+Yeap Eugine Yes, sig. two tail is the p-value associated with the t-test. Alpha is different. Alpha is the maximum error rate you are willing to accept in the analysis. Usually, people specify alpha at .05. They then look to see if the p-value associated with the t-test is less than alpha (i.e., .05).
If t-test p value is .004, it's less than .05, thus, the conclusion is to reject the null hypothesis of equal means.
can you explain to me what is manipulation check using t test
How about explaining the contents of the other columns in both Data & Variable view
How can I simulate 10000 samples of size n=5 in a normal population (10;10) in spss??
Thanks a lot for this! Test on Friday
Also, what does it mean by Sig 0.679? it means for equality? why not just compare the null hypothesis?
Sig/p value which is 0.679 is > 0.05 so equality is assumed
Thank you for your useful video. I have followed your steps, however I cannot see the Sig (2-tailed), instead the table presented One-side p (0.242 and 0.250) and Two-side p (0.484 and 0.499). My question is how can I draw findings with these information? thank you so much.
How do I do significance testing for mean scores for more than 2 groups. Let's say I have Group A, Group B, and Group C...?
One-way between groups ANOVA
Very useful indeed! Is it not necessary to convert brain size numbers into Z scores before running the t test?
I'm a little confused as to why the significance for the F value has to be above .05. What is the actual null hypothesis for Levene's homogeneity test? Is it "the samples are from different populations?"
and why do we need to use the t-test if we already have the correlation coefficient? I'm sorry about the beginner-ish questions..I'm completely new to statistics
THANK YOU :) NOW I KNOW HOW TO DO THE HYPOTHESIS THANK YOUUU :)
hi
please gave me the all description of independent T-test.produces by spss software
I have created weights/expansion factors for my dataset (to match the survey universe). My weighted vs. non-weighted results are very different. I assume that it is best to use the unweighted/unexpanded data?
Doesn't the p value only need to be greater or less than +-.025 to have the difference of the population means be declared significant? You were saying .05, but it's two tailed so wouldn't it be half that?
No. A one-tailed test would include all of the "alpha" on one side of the t-distribution, which would correspond to a smaller critical t-value. You would still report a significant result (greater observed t-value) as p < .05. A two tailed t-test distributes the .05 alpha equally to both sides of the t-distribution and leads to a larger critical t-value. Again, though, an observed t-value greater than the critical t-value would still be reported as p < .05, in this case. This webpage might help you understand: statisticsbyjim.com/hypothesis-testing/one-tailed-two-tailed-hypothesis-tests/
Doesn't the groops variable, (0 non smokers 1 smokers) have to be set at nominal? or dose it not matter?
If the Levine's test is significant should we then still use the independent t-test or should we then use t(32.321)=3.072 ?
Use the adjusted t-test, 3.072, if levene's is significant. It's Welch's t-test and is robust to heterogeneity of variance.
Thank you so much! GOAT 🐐
Hello. Thank you for this video. I have a question. I want to compare if a numeric variable A has an impact on a variable B which has 3 numeric variables, B1, B2, and B3. What I want to do is to demonstrate if A has an impact on B1, B2 or B3. Should I run a t-test or just correlate if A has any impact on B1 and then if A has an impact on B2 and the same with B3 ( analyze->correlate-> bivariate) . Which of these possibilities is the correct one? I just started using SPSS and learning statistics and I really need this to know for my thesis. Thank you very much.
Cristina Ciocsirescu n
Can I ask a link for the IBM SPSS statistical software you use. I'm having a hard time in finding it
what I doing when the value of t less than 1 and the sig. for t _test
is more than 0.05 ?
Thank you!! This was very useful.
why is there a equal variances assumed and not assumed and how did SPSS got that answer. and why is there levene`s test? i also did not understand upper and lower.??
Sorry if my question is elementary: what is the meaning of the t-value and the degree of freedom? We keep looking at significance but not these values, what are they for?
Hi, please can you explain how the column Std Error difference is calculated? Thanks
Hi thanks so much for your video! Could you explain when we should look at data in the row for "equal variances assumed" and when we should use the row "equal variances not assumed"? I used a Shapiro-Wilk test to see check my data is normally distributed, before using the T-test but it seems that the values I get under the column "Sig. 2-tailed" are all not significant anyway...
hii, i need to compare from the published article with my research article from their mean which test should i apply
How is the degree of freedom (df) calculated in SPSS for the independent t-test?
You helped me so much! Is there a donation button? I wouldn't mind supporting this.
But how do I have to think when my sig, t, mean difference are all negatives?
It's impossible to have a negative significance (p-value). A negative t-value simply implies that the second mean is numerically larger than the first mean. However, it's essentially arbitrary which mean is mean #1 and which is mean #2, so, you just have to look at which mean each group belongs to and interpret the results accordingly.
The video is nice but i am yet to grasp it
like equal variance assumed means, not assumed means, levenes test equality variances means
Thank you for the explanation. But smokers or nonsmokers data are nominal. Should not we use a non-parametric test with nominal variables?
In this context, you would use a non-parametric test if the _dependent_ variable was measured on a nominal scale. In the independent samples t-test, the independent variable is always nominal in nature and the dependent variable is interval/ratio in nature. In this case, the dependent variable was brain volume, which is continuous (interval/ratio).
Btw, take a look this vid, it's about Doronix Math Toolbox - tool for statistics
watch?v=SWA86d_ZMpQ
The smoker variable is just the grouping criteria so that the software knows which case belongs to which group. Thats not the variable which is being analyzed in the t test. The variable being analyzed is the scale one...
hi, good time, The standard deviation can be greater than the mean?If so, what is the reason?How can this problem be solved? much thanks
Hi, is there a way to test Granger Causality using SPSS? Thanks.
What would you advise to do if the significance of the Levene´s test is less than .05?
Welch's t-test: check this out: ua-cam.com/video/fl6wgE4Edqw/v-deo.html
How do you calculate effect size if SPSS doesn't do it?
+MIBCN2008 With a calculator: Cohen's d = (Mean 1 - Mean 2) / ((SD1 + SD2) / 2)
If you're keen, you can get eta-squared through the GLM menu option in SPSS
+how2stats Thanks.
Many thanks for this easy ablicable lec.
just one question ,,i tried it in my data the result of significance give me 0.000
is this possible or some thing wrong ?
No, nothing is wrong. If you push out the decimal place limit in the output, you'll see a number eventually. People report 0.000 as p < .001. You rejected the null hypothesis.
Is this Nominal or ordinal measure?
I would like to ask, if i want to do the one-tailed F test on this, what the Nul Hypothesis and the alternate hypothesis will be? Many thanks in advance ^^
is it possible if the significance value in the Levene's test is 0.00?
thank you for this❤❤😭😭
If someone can help me, I have to measure an Independent Variable with two levels (control group and experimental group) and also a Dependent Variable with two levels (reaction time and mistakes made during test). What test should I use, and how should I put the values?
i have same problem.. an expert in this field suggested me this.... "The type of data you have may determine the statistical approach. Replicated plot data, for example, lends itself to statistics that assume a "normal" distribution. Statistics like mean, standard deviation, analysis of variance, etc. It may be appropriate to use non-parametric procedures if data is skewed or "messy". Median and quartiles or quintiles may be more appropriate and better reflect the data frequency distribution is skewed. Some of these tests use the data rank rather than the mean to identify differences. Examples are Mann-Whitney test, Kendall test-for-trend, Sen's slope estimator, and Kruskal-Wallis non-parametric analysis of variance. These tests can identify significant differences, but are not necessarily able to quantify the difference. Scientists in psychology or wildlife biology collect data that is not replicated and certainly "messy", but are able to make statistically valid conclusions. It appears that the data you have collected may not necessarily be replicated, but you could use a non-parametric statistical approach".
hope this helps.
What do you do if you want to run an Independent samples t-test and your data is not normally distributed? (eg. positively skewed).
James Cowan Watch this: ua-cam.com/video/yNdlGRz-Z04/v-deo.html If your data are very non-normally distributed, you have the option of bootstrapping.
It calculates effect size now :)
Why Std.Deviation is greater than one?
There is no limit to what value a standard deviation can take. Perhaps you are thinking of correlation?
do we need to check for normality in T-test?
+John tialcungling Yes, but you should check out this video before you make a decision about normality: ua-cam.com/video/yNdlGRz-Z04/v-deo.html
Thank you very much for your prompt reply how2stats as i am in the urgent... i will try.
Thank you very much fos the huge help!!!
Hi,
I did this and followed all the steps but its saying it cant be done because at least one group is empty, what does this mean ? Thanks
twoofakind786 Hi I am having the same problem now. Did you ever figure out how to fix it?
Glad to see someone's figured it out since then... I actually needed to know the answer to this.
Dear How2stats,
Thank you so much. You are a real life saver.
I'm currently working with an spss file created by another, who defined male and female and 1 and 0. Which is fine. I would like to know though, how can I have it so that the group is described not as 0 and 1, but the word male and female in the output file.
I hope the question makes sense? I just don't know how to back track and change the original file from 0 and 1 to male and female. The information was verbally relayed, but would help me if I could have the word in the result!
Thanks.
eduffles Just got it! When I double click on the table, in the output file, directly on the group 0 or 1 (whichever I want to change), I can type in whatever the true description should be.
+eduffles You can also use value labels so that you don't have to edit the tables.
i am new to spss, but it triggers me to see categori numbers with decimals