sta

u08a1SPECIALWORKSHEET.new.t-test.200227.xlsx

Home >Mathematics homework help >Statistics homework help >sta

SPECIAL WORKSHEET u08a1

ENTER NAME HERE--->				0
When you submit your work, will you include a histogram on GPA?
Please select a choice	1			0
When you submit your work, will you include a descriptive statistics table for GPA (including skew and kurtosis)?
Please select a choice	1			0
When you submit your work, will you include the SPSS output for the Shapiro-Wilk test?
Please select a choice				0
When you submit your work, will you include the SPSS output for the Levene's test?
Please select a choice				0
When you submit your work, will you include the SPSS output for the results of the t-test?
Please select a choice				0
Variable	What kind of Variable Is This?		What is the Scale of the Variable?
GENDER =	Please select a choice		Please select a choice	0	0
GPA =	Please select a choice		Please select a choice	0	0
What is the overall sample size?	Please select a choice			0
Assumptions	Select 'Yes' for each assumption that needs to be satisfied when conducting a t-test		Select 'Yes' for each assumption that was satisfied in the data set you're working with IMPORTANT: For each assumption, only answer this if you selected "Yes" in the cell to the left. If you selected "No," the answer choice will go black in this column
Independence of observations	Please select a choice		Please select a choice	0	0
Outcome (or dependent) variable is quantitative and normally distributed	Please select a choice		Please select a choice	0	0
Homogeneity of variance	Please select a choice		Please select a choice	0	0
Variables are linearly related	Please select a choice		Please select a choice	0	0
To evaluate these assumptions, please fill in the blanks with the results of the following tests:
Shapiro-Wilk Test, Statistic =		<-------- Use THREE decimal places		0
Shapiro-Wilk Test, p-value =		<-------- Use THREE decimal places		0
Levene's Test, Statistic (F-value) =		<-------- Use THREE decimal places		0
Levene's Test, p-value =		<-------- Use THREE decimal places		0
Articulate a research question relevant to the statistical test	Please select a choice	0		0
Articulate the null hypothesis	Please select a choice	0		0
Articulate the alternative hypothesis	Please select a choice	0		0
Specify the alpha level	Please select a choice	0		0
t-test: Please fill in the blanks:
degrees of freedom =		0		0
t-value =		<-------- Use THREE decimal places		0
p-value =		<-------- Use THREE decimal places		0
effect size (calculate this: eta2) =		<-------- Use THREE decimal places		0
For the next question, please select from the drop-down box.
Based on the results of the t-test, should the null hypothesis be rejected?	Please select a choice	0		0
	Descriptive statistics for FEMALES
Mean =		<-------- Use TWO decimal places		0
Standard Deviation =		<-------- Use THREE decimal places		0
	Descriptive Statistics for MALES
Mean =		<-------- Use TWO decimal places		0
Standard Deviation =		<-------- Use THREE decimal places		0
Mean difference between the means for males and females =		<-------- Use TWO decimal places		0
Please enter the LOWER and UPPER bounds of the 95% confidence interval for the difference between means for males and females
Lower =		<-------- Use THREE decimal places		0
Upper =		<-------- Use ONE decimal place		0
What is the main strength of a t-test?
Please select a choice				0
What is the main limitation of a t-test?
Please select a choice				0

Hold on! You're not done. You still have 41 question(s) to answer.
(and your name is one of them)

SPECIAL REQUIRED WORKSHEET FOR U08A1 - t-Tests This worksheet is absolutely positively required. Why? It compels you to engage with each question and will guide your efforts on the long-form Word doc part of the assignment. What's in it for you? Hints and points. Lots of them. Failing assignment submissions become passes. C's turn to B's. B's turn to A's. So, please hand this in. Remember: This worksheet is required but does not replace your fully-written u08a1 DAA Word doc. You must submit BOTH.

First things first... You are required to show your SPSS output when you submit your assignment. Please indicate whether you intend to do so.

Section 1 - DATA CONTEXT: In this section, you'll be making choices from drop-down menus.

Section 2 - TESTING ASSUMPTIONS: In this section, you'll be using drop-down menus to choose 'yes' or 'no'

Section 3 - Research Question, Hypotheses, Alpha Level: In this section, you'll be using drop-down menus

Section 4 - Interpretation: In this section, you'll be using drop-down menus and filling in the blanks

Section 5 - Conclusion: In this section, you'll be using drop-down menus

TWO MAJOR HINTS -- (1) Use the EQUAL VARIANCES ASSUMED figures; (2) To calculate effect size, square the t-value, then divide that by the sum of the square of the t-value and degrees of freedom

MAJOR HINTS FOR THE ASSUMPTIONS PORTION -- (1) Even if the DV is NOT normally distributed, we can still perform a t-test. (2) Linearity was really important for correlation. Is this assignment about correlation? (3) If homogeneity of variance is not violated, we use the output where equal variance is assumed.

MAJOR HINT FOR THE DATA CONTEXT PORTION -- Gender is neither interval nor ratio. GPA is not ordinal. Do you see why?

MAJOR HINT FOR THIS PORTION -- Hypothesis testing is a researcher's bread and butter. If you're still having trouble with it, please see the handout on 4d1 from a few weeks ago.

MAJOR HINT -- To determine whether to reject the null, we look at the p-value. If p<.05, we REJECT the null.

OUTPUT

SCORING

RUBRIC

12%

ENTER NAME HERE--->

Great job on DAA - lots of explanation and elaboration

Important Note: Your DAA was beautifully comprehensive. As I read through it, I got the sense that you'd really absorbed these concepts! This is an extraordinary achievement for such relatively short engagement with the concepts and materials. I am TRULY impressed! Well done!

Please select a choice

A mediocre job on DAA - some explanation but could benefit from more

Your DAA flowed from start to finish with learning. Could it have benefited from a bit more explanation in places? Yes (e.g., in the section on assumptions and the interpretation). But overall, I could see the effort. Thank you for making that effort.

Rubric #

RAW SCORE

WEIGHTED

Yes

Not a good job on DAA -- needs a lot more explanation

One note about the DAA: I think you’d be doing yourself (and your assignments) a huge favor by explaining and elaborating more in your DAA. The fact is that even if something goes awry in the numbers, even if you made a mistake in the way you used SPSS, it’s still possible for you to grab a few more points if you can demonstrate you knew what you were talking about. Right now I’m not seeing that in your DAA but in future assignments, that sort of explanation could help you out. By the way, the kind of explanation I’m talking about would have to be substantive, relevant, and meaningful; as you know, I read every word of every paragraph so I’ll be looking for content that really shows me you were working to absorb these concepts and make them your own.

20%

Please select a choice

some level of explanation but a lot of statements were inaccurate

You made a strong effort to explain what you were doing; thank you so much for doing that. It gives me a sense first of all, of hard you worked (and you did work hard!) but also of your level of achieved learning. I think the level of achieved learning is climbing but it hasn’t quite gotten where it needs to go. I’m guessing you felt that as you were working through the assignment. All this means is that there’s more work to be done. That’s part of the process. No worries.

10%

Predictor variable

20%

Please select a choice

Outcome variable

10%

Nominal

Moderator variable

20%

Ordinal

Mediator variable

10%

Ratio

100%

10%

Interval

Please select a choice

Continuous

Is there a difference in GPA between male and female students?

Is there a difference in GPA based on school location?

Detailed Comments on Your Assignment:

Is there a correlation between male and female students?

Hi, How good should you feel about what you did here? SUPER-good. Yes, you’ve got some things you need to do here and there to shore up the gaps, but this is just really good work of which you should be tremendously proud. Here are a few comments I worked up: Your DAA flowed from start to finish with learning. Could it have benefited from a bit more explanation in places? Yes (e.g., in the section on assumptions and the interpretation). But overall, I could see the effort. Thank you for making that effort. In section 1, you were asked to describe specific elements of the data set by describing the variables themselves. This is pretty important because if we don't understand the variables and the sort of relationship we'd like to test, then none of this can help us answer any research questions. I bring all this up because I noticed you had difficulty determining which variable was the predictor and / or which was the outcome. For this data set, the independent variable (or predictor) should have been GENDER and the dependent variable (or outcome) should have been GPA. I would check over your work and figure out where things went wrong. Along these lines, I should mention you were asked to evaluate which scale of measurement we would use with the GENDER and GPA variables, respectively. I noticed an error or two here. The correct answers (just so you know, so you can compare them to what you did) for GENDER would be nominal (since gender can either be male or female) and for GPA, would be ratio or interval. Why should we care about any of this? Most of the parametric tests you’ll be learning about in this course (such as the t-test) require the dependent variable to be at least ratio. If you try to run those analyses on incompatible variables, your analysis will go kablooey (that’s a technical statistical term, you understand) fairly quickly. The only other thing I noticed in this section was that your measurement of sample size for the data set appears to be incorrect. It should be 105, reflecting the total number of participants in the study. Okay, so let's tackle section 3 of the rubric, the one asking you to evaluate the assumptions that need to be satisfied before doing a t-test. This is the sort of stuff that you either know (based on your reading of the course material) or you don't. Of course, the real challenge is that once you know it, you've got to be able to understand it. It looks as if you may not have represented your knowledge correctly (or just didn't completely nail down that particular piece of knowledge) because I see at least one error in the part where you had to identify assumptions that would need to be satisfied. It could also be that you didn't completely understand these assumptions, or how to apply them because when I look at your DAA and worksheet, I see at least one case where you misidentified whether an assumption was satisfied in the data set with which you were working. Just so you know what all of the correct answers were, let me list them out. The assumptions that have to be met when conducting a t-test are: Independence of observations; Outcome (or dependent) variable is quantitative and normally distributed; and Homogeneity of variance. Of these, only the first and third were met in this data set. So, all in all, I think you may want to dig back into the materials a bit. As for the Shapiro-Wilk test, you could very well be on the right track in your understanding of it, though, to be candid, I couldn’t be sure from what I saw in your assignment submissions (I saw at least one error of note skulking about in your work). Let’s take a step back. What does this test do? This test helps us determine whether the data set satisfied the assumption of normal distribution. In this case, since p<.05, the assumption of normal distribution was violated. Remember that for this sort of a test where we’re trying to establish whether assumptions have been met, we want to see statistical NON-significance. Conveniently, SPSS provides the output of the Shapiro-Wilk test for your review. I would revisit the material to see if you can get a better handle on the underlying concepts of this test. Another important point: The Levene test is something you’ll want to get fairly comfortable with, since you will see it over and over in discussions of the assumptions of parametric tests. Again, I saw an error or two in your treatment of it in your assignment submissions so this is something you’ll want to review. Just so you know, Levene’s test helps us determine if the homogeneity of variance assumption has been met. SPSS can perform the Levene test and produces the output along with the results. Note that just as we saw with the Shapiro-Wilk test, we want Levene’s test to be non-significant (as judged by the p-level). In the fourth segment of the rubric, you had to come up with a research question, hypotheses and alpha level. Every good study begins with a good research question, which is why it’s so important to be able to get this right. Regrettably, your assignment showed that something went wrong here. So, let’s look at this: The research question is what the researcher would like to investigate. It is literally a question, and has to be asked with some precision of language. What does this mean? This means that the research question isn’t something we ask casually: the words have specific meanings. In this case, we were interested in whether male and female students had the same GPA, on average. The answer, therefore, has to be in the research question. Here’s what the question should look like: Does gender have an effect on GPA? A few more notes: We didn’t have any interest in either the location of the school or students’ age. Similarly, this was a research question suitable for a t-test, not a correlation, so any question about correlation was automatically incorrect. The null hypothesis you specified, also doesn’t appear to be quite right. Again, null hypotheses have to be right; they’re just not the sort of thing you can get sort of right. Here’s why: The null hypothesis is the opposite of what we think will happen in the study. It is a statement that we reject if p is less than the specified alpha level (typically, .05). When it comes to this assignment, we wondered whether there’s a statistically significant effect, so the null hypotheses had to be the opposite of that statement (no difference in GPA between men and women). If articulation of the null hypothesis is still unclear, I would suggest revisiting the course materials. Understanding what the null hypotheses are and how to formulate them will serve you well not only in this course, but throughout your career. Now, turning to the alternative hypothesis, something appears to be not entirely right here. The alternative hypothesis is what we’d like to demonstrate in the study. Typically, it’s the affirmative statement of the research question. In our case here, it’s the statement that says that there is a difference between the GPA of male and female students. As with the null hypothesis, it’s super-important to be precise in the way we articulate it. Please reread the textbook (and/or other sources) to make sure you know what an alternative hypothesis is and how to postulate one. One more thing to know for this section: The alpha level is typically set to .05. Yours wasn't. Let's continue on to t-tests, your understanding of which is evaluated in the 5th segment of the scoring rubric for this unit. Remember that a t-test is all about determining whether the means of 2 groups are different in a statistically significant way (meaning that they're different but not simply because you sampled in a certain way but because if you kept sampling over and over, you'd see the same kind of effect). To get to a p-value which helps you understand whether you are indeed looking at a statistically significant difference, you've got to calculate some other statistics, as well. So, first off, I noticed an error in your calculation of the t-value, which you were supposed to take directly from SPSS output. Any kind of mistake here is due to maybe one of 2 factors: (1) An incorrect test was performed. Please double-check the SPSS step-by-step guide. (2) A rounding error (3) A misunderstanding of what you were looking at When I look both at your worksheet and the Word doc, I can see a problem with your calculation of degrees of freedom. This, too, was provided for you in the SPSS output for the independent samples t-test. If there is an error here, it’s likely because you copied the wrong number or maybe there’s some sort of rounding error. Remember that since Levene’s test was found not to be significant (we fail to reject the null hypothesis which said that equal variances are assumed), we can use the “equal variance assumed” row (I provided a hint in a red box, right on the worksheet to help you with this). As for p-value, well, we know that p-value is what we've been chasing, right? Unfortunately, it doesn't appear to be correct in your assignment submission. Please check to see that you copied it correctly from the SPSS output. Also, keep in mind that since the Levene test was found to be non-significant, we can use the “equal variance assumed” row I provided a hint in the worksheet itself about this (you couldn't miss it -- it's in a big red box). Effect Size: The effect size you listed is not correct. As discussed within this unit, SPSS doesn't calculate the effect size for you. You have to compute it manually using a formula that you'll find in your course materials. I also placed a hint about this in the worksheet. How big of a hint? Well, I did provide the formula. As for interpreting the magnitude of the effect size, Warner (2013) includes a table (5.2) that should help. The announcement I'd posted recently should have helped you with this, as well; I would review it. Importantly, contrary to what you wrote in your assignment, the null hypothesis should be rejected in this case because it is less than 0.05, our present alpha level). Why is this the case? Please take a look at the textbook as well as other course and non-course materials to get a better understanding of hypothesis testing. As I glance down your paper, I see some problems in what should be simple transcription of descriptive statistics from SPSS output. It looks as if the following aren't right: mean for females; standard deviation for females; mean for males; standard deviation for males. So, either you made a mistake when you copied the numbers (or filled out the worksheet; note that if the numbers appear as correct in at least one of those formats, you received partial credit), OR you may not have followed carefully enough the step-by-step SPSS guide. As for the mean difference, the correct answer should have been 0.28 (rounded), which should have been a relatively manageable sort of calculation. Onto confidence intervals: We learned a unit or two ago that confidence intervals help us understand the range of numbers within which a given parameter of interest probably lies. In this part of the assignment, you had to specify the lower and upper bounds of that range, and the SPSS output should have come to your rescue and provided you the 95% confidence interval for the difference between the 2 means. Since we are assuming equal variance here (because Levene’s test was non-significant), the lower and upper bound numbers should have been offered in the first row of the results. Something went wrong here so I would revisit the material, and maybe re-run this in SPSS to see whether you get different results. The last section where you had to evaluate the strength and limitation of a t-test is another of those you-either-know-it,-or-you-don't kinds of situations. Both of your answers were incorrect, so I'm not sure what happened here. You might just want to check out the course material and get this moving in the right direction again. One final thing: When I look over your DAA, I see someone who worked extraordinarily hard to think through and develop this assignment. I do not under any circumstances want you to think that just because I marked some things wrong, took off some points here and there, and wrote a lot of feedback, that you didn’t do well. I think you did REALLY well. This paper is what it looks like to work hard, and even though you may not have gotten everything as correct as you wanted to, I really truly believe with all my heart, that you scored a real victory here. Regardless of what you think about your abilities, you’re in the fight. Thanks for all the hard work! -Dr. Reynolds

Is there a relationship between GPA and student age?

When you submit your work, will you include a histogram on GPA?

Please select a choice

There is no difference in GPA between male and female students.

There is no difference in GPA based on school location.

When you submit your work, will you include a descriptive statistics table for GPA (including skew and kurtosis)?

There is a difference in GPA between male and female students.

Please select a choice

AMAZING

Hi, This is really fantastic work! You swept through these sections and made these concepts your own. Really well done! Just a few comments:

There is no correlation between male and female students.

Great job

Please select a choice

When you submit your work, will you include the SPSS output for the Shapiro-Wilk test?

Great effort

Hi, I say it all the time and I have to say it again, because your work here is living proof of it: This course has to be all about the journey of learning, more than whether you got everything right. As I look through your DAA and the worksheet, I see evidence of that learning. You should feel proud of that; I certainly do. Here are my specific comments:

105

Please select a choice

Good effort

Hi, I say it all the time and I have to say it again, because your work here is living proof of it: This course has to be all about the journey of learning, more than whether you got everything right. I could see some of that journey in this, some evidence of effort, and I am really grateful for it. The task is to continue to try to take that up a notch, and make these concepts your own. Here are my specific comments:

There is a difference in GPA between male and female students.

420

There is a difference in GPA based on school location.

When you submit your work, will you include the SPSS output for the Levene's test?

Thanks for all the hard work! -Dr. Reynolds

One final thing: When I look over your DAA, I see someone who worked extraordinarily hard to think through and develop this assignment. I do not under any circumstances want you to think that just because I marked some things wrong, took off some points here and there, and wrote a lot of feedback, that you didn’t do well. I think you did REALLY well. This paper is what it looks like to work hard, and even though you may not have gotten everything as correct as you wanted to, I really truly believe with all my heart, that you scored a real victory here. Regardless of what you think about your abilities, you’re in the fight.

Unknown

There is no difference in GPA between male and female students.

Please select a choice

There is a correlation between male and female students.

When you submit your work, will you include the SPSS output for the results of the t-test?

Please select a choice

Yes

It cannot evaluate the means between more than 2 groups.

It doesn't require assumptions to be satisfied before we can use it.

Not Applicable

It can evaluate whether there's a statistically significant difference between 2 groups.

It can predict the required sample size for a follow-up study.

Select from drop-down menu boxes

Please select a choice

Does the previous GPA correlate with the number of correct final exam scores?

It cannot evaluate the means between more than 2 groups.

Variable

What kind of Variable Is This?

What is the Scale of the Variable?

Does gender correlate with previous GPA?

It can evaluate the means between more than 2 groups.

GENDER =

Please select a choice

I bring all this up because I noticed you had difficulty determining which variable was the predictor and / or which was the outcome. For this data set, the independent variable (or predictor) should have been GENDER and the dependent variable (or outcome) should have been GPA. I would check over your work and figure out where things went wrong.

Along these lines, I should mention you were asked to evaluate which scale of measurement we would use with the GENDER and GPA variables, respectively. I noticed an error or two here. The correct answers (just so you know, so you can compare them to what you did) for GENDER would be nominal (since gender can either be male or female) and for GPA, would be ratio or interval. Why should we care about any of this? Most of the parametric tests you’ll be learning about in this course (such as the t-test) require the dependent variable to be at least ratio. If you try to run those analyses on incompatible variables, your analysis will go kablooey (that’s a technical statistical term, you understand) fairly quickly.

The only other thing I noticed in this section was that your measurement of sample size for the data set appears to be incorrect. It should be 105, reflecting the total number of participants in the study.

In section 1, you were asked to describe specific elements of the data set by describing the variables themselves. This is pretty important because if we don't understand the variables and the sort of relationship we'd like to test, then none of this can help us answer any research questions. I bring all this up because I noticed you had difficulty determining which variable was the predictor and / or which was the outcome. For this data set, the independent variable (or predictor) should have been GENDER and the dependent variable (or outcome) should have been GPA. I would check over your work and figure out where things went wrong. Along these lines, I should mention you were asked to evaluate which scale of measurement we would use with the GENDER and GPA variables, respectively. I noticed an error or two here. The correct answers (just so you know, so you can compare them to what you did) for GENDER would be nominal (since gender can either be male or female) and for GPA, would be ratio or interval. Why should we care about any of this? Most of the parametric tests you’ll be learning about in this course (such as the t-test) require the dependent variable to be at least ratio. If you try to run those analyses on incompatible variables, your analysis will go kablooey (that’s a technical statistical term, you understand) fairly quickly. The only other thing I noticed in this section was that your measurement of sample size for the data set appears to be incorrect. It should be 105, reflecting the total number of participants in the study.

Do the final exam scores correlate with the geographical areas?

It can evaluate whether there's a statistically significant difference between 2 groups.

GPA =

Please select a choice

The only thing

The only other thing

I noticed in this section was that your measurement of sample size for the data set appears to be incorrect. It should be 105, reflecting the total number of participants in the study.

Does socioeconomic status correlate with the final exam scores?

It can predict the required sample size for a follow-up study.

Please select a choice

nullhyp

What is the overall sample size?

Please select a choice

Previous GPA does not correlate with the number of correct final exam scores.

Gender does not correlate with the previous GPA.

The previous GPA correlates with the number of correct final exam scores.

Gender correlates with the previous GPA.

Please select a choice

althyp

The previous GPA correlates with the number of correct final exam scores.

Assumptions

Select 'Yes' for each assumption that needs to be satisfied when conducting a t-test

Select 'Yes' for each assumption that was satisfied in the data set you're working with

opener

good first col, >1 mistake second col

good first col, 1 mistake second

bad first col, good second

bad first col, bad second col

Gender correlates with the previous GPA.

Independence of observations

Please select a choice

Yes

It looks as if you knew it at the definitional level because you identified the assumptions correctly. However, you may have had a bit of trouble applying these definitions because when I look at the DAA and / or your worksheet, I see a few cases where you did not correctly state whether the assumptions applied. Just so you know what all of the correct answers were, let me list them out. The assumptions that have to be met when conducting a t-test are: Independence of observations; Outcome (or dependent) variable is quantitative and normally distributed; and Homogeneity of variance. Of these, only the first and third were met in this data set.

It looks as if you knew it at the definitional level because you identified the assumptions correctly. However, you may have had a bit of trouble applying these definitions because when I look at the DAA and / or your worksheet, I saw one case where you did not correctly state whether the assumption applied. Just so you know what all of the correct answers were, let me list them out. The assumptions that have to be met when conducting a t-test are: Independence of observations; Outcome (or dependent) variable is quantitative and normally distributed; and Homogeneity of variance. Of these, only the first and third were met in this data set.

It looks as if you may not have represented your knowledge correctly (or just didn't completely nail down that particular piece of knowledge) because I see at least one error in the part where you had to identify assumptions that would need to be satisfied. It could also be that you didn't completely understand these assumptions, or how to apply them because when I look at your DAA and worksheet, I see at least one case where you misidentified whether an assumption was satisfied in the data set with which you were working. Just so you know what all of the correct answers were, let me list them out. The assumptions that have to be met when conducting a t-test are: Independence of observations; Outcome (or dependent) variable is quantitative and normally distributed; and Homogeneity of variance. Of these, only the first and third were met in this data set. So, all in all, I think you may want to dig back into the materials a bit.

Okay, so let's tackle section 3 of the rubric, the one asking you to evaluate the assumptions that need to be satisfied before doing a t-test. This is the sort of stuff that you either know (based on your reading of the course material) or you don't. Of course, the real challenge is that once you know it, you've got to be able to understand it. It looks as if you may not have represented your knowledge correctly (or just didn't completely nail down that particular piece of knowledge) because I see at least one error in the part where you had to identify assumptions that would need to be satisfied. It could also be that you didn't completely understand these assumptions, or how to apply them because when I look at your DAA and worksheet, I see at least one case where you misidentified whether an assumption was satisfied in the data set with which you were working. Just so you know what all of the correct answers were, let me list them out. The assumptions that have to be met when conducting a t-test are: Independence of observations; Outcome (or dependent) variable is quantitative and normally distributed; and Homogeneity of variance. Of these, only the first and third were met in this data set. So, all in all, I think you may want to dig back into the materials a bit.

The previous GPA does not correlate with the number of correct final exam scores.

Outcome (or dependent) variable is quantitative and normally distributed

Please select a choice

Yes

Homogeneity of variance

Please select a choice

Yes

Please select a choice

Variables are linearly related

Please select a choice

0.50

0.00

Please select a choice

0.05

Gender/GPA

0.50

To evaluate these assumptions, please fill in the blanks with the results of the following tests:

Gender/Final

0.90

Gender/Total

0.00

Shapiro-Wilk Test, Statistic =

Statistic

p-value

Another important point: The Levene test is something you’ll want to get fairly comfortable with, since you will see it over and over in discussions of the assumptions of parametric tests. Again, I saw an error or two in your treatment of it in your assignment submissions so this is something you’ll want to review. Just so you know, Levene’s test helps us determine if the homogeneity of variance assumption has been met. SPSS can perform the Levene test and produces the output along with the results. Note that just as we saw with the Shapiro-Wilk test, we want Levene’s test to be non-significant (as judged by the p-level).

As for the Shapiro-Wilk test, you could very well be on the right track in your understanding of it, though, to be candid, I couldn’t be sure from what I saw in your assignment submissions (I saw at least one error of note skulking about in your work). Let’s take a step back. What does this test do? This test helps us determine whether the data set satisfied the assumption of normal distribution. In this case, since p<.05, the assumption of normal distribution was violated. Remember that for this sort of a test where we’re trying to establish whether assumptions have been met, we want to see statistical NON-significance. Conveniently, SPSS provides the output of the Shapiro-Wilk test for your review. I would revisit the material to see if you can get a better handle on the underlying concepts of this test. Another important point: The Levene test is something you’ll want to get fairly comfortable with, since you will see it over and over in discussions of the assumptions of parametric tests. Again, I saw an error or two in your treatment of it in your assignment submissions so this is something you’ll want to review. Just so you know, Levene’s test helps us determine if the homogeneity of variance assumption has been met. SPSS can perform the Levene test and produces the output along with the results. Note that just as we saw with the Shapiro-Wilk test, we want Levene’s test to be non-significant (as judged by the p-level).

GPA/Final

Shapiro-Wilk Test, p-value =

0.961

0.004

GPA/Total

Final/Total

Levene's Test, Statistic (F-value) =

Statistic (F-value)

p-value

Levene's Test, p-value =

0.095

0.758

Please select a choice

Gender/GPA

Gender/Final

GPA/Final

GPA/Total

opener

research q

Final/Total

Articulate a research question relevant to the statistical test

Please select a choice

Is there a difference in GPA between male and female students?

In the fourth segment of the rubric, you had to come up with a research question, hypotheses and alpha level.

Every good study begins with a good research question, which is why it’s so important to be able to get this right. Regrettably, your assignment showed that something went wrong here. So, let’s look at this: The research question is what the researcher would like to investigate. It is literally a question, and has to be asked with some precision of language. What does this mean? This means that the research question isn’t something we ask casually: the words have specific meanings. In this case, we were interested in whether male and female students had the same GPA, on average. The answer, therefore, has to be in the research question. Here’s what the question should look like: Does gender have an effect on GPA? A few more notes: We didn’t have any interest in either the location of the school or students’ age. Similarly, this was a research question suitable for a t-test, not a correlation, so any question about correlation was automatically incorrect.

The null hypothesis you specified, also doesn’t appear to be quite right. Again, null hypotheses have to be right; they’re just not the sort of thing you can get sort of right. Here’s why: The null hypothesis is the opposite of what we think will happen in the study. It is a statement that we reject if p is less than the specified alpha level (typically, .05). When it comes to this assignment, we wondered whether there’s a statistically significant effect, so the null hypotheses had to be the opposite of that statement (no difference in GPA between men and women). If articulation of the null hypothesis is still unclear, I would suggest revisiting the course materials. Understanding what the null hypotheses are and how to formulate them will serve you well not only in this course, but throughout your career.

Now, turning to the alternative hypothesis, something appears to be not entirely right here. The alternative hypothesis is what we’d like to demonstrate in the study. Typically, it’s the affirmative statement of the research question. In our case here, it’s the statement that says that there is a difference between the GPA of male and female students. As with the null hypothesis, it’s super-important to be precise in the way we articulate it. Please reread the textbook (and/or other sources) to make sure you know what an alternative hypothesis is and how to postulate one.

One more thing to know for this section: The alpha level is typically set to .05. Yours wasn't.

In the fourth segment of the rubric, you had to come up with a research question, hypotheses and alpha level.

One more thing to know for this section: The alpha level is typically set to .05. Yours wasn't.

In the fourth segment of the rubric, you had to come up with a research question, hypotheses and alpha level. Every good study begins with a good research question, which is why it’s so important to be able to get this right. Regrettably, your assignment showed that something went wrong here. So, let’s look at this: The research question is what the researcher would like to investigate. It is literally a question, and has to be asked with some precision of language. What does this mean? This means that the research question isn’t something we ask casually: the words have specific meanings. In this case, we were interested in whether male and female students had the same GPA, on average. The answer, therefore, has to be in the research question. Here’s what the question should look like: Does gender have an effect on GPA? A few more notes: We didn’t have any interest in either the location of the school or students’ age. Similarly, this was a research question suitable for a t-test, not a correlation, so any question about correlation was automatically incorrect. The null hypothesis you specified, also doesn’t appear to be quite right. Again, null hypotheses have to be right; they’re just not the sort of thing you can get sort of right. Here’s why: The null hypothesis is the opposite of what we think will happen in the study. It is a statement that we reject if p is less than the specified alpha level (typically, .05). When it comes to this assignment, we wondered whether there’s a statistically significant effect, so the null hypotheses had to be the opposite of that statement (no difference in GPA between men and women). If articulation of the null hypothesis is still unclear, I would suggest revisiting the course materials. Understanding what the null hypotheses are and how to formulate them will serve you well not only in this course, but throughout your career. Now, turning to the alternative hypothesis, something appears to be not entirely right here. The alternative hypothesis is what we’d like to demonstrate in the study. Typically, it’s the affirmative statement of the research question. In our case here, it’s the statement that says that there is a difference between the GPA of male and female students. As with the null hypothesis, it’s super-important to be precise in the way we articulate it. Please reread the textbook (and/or other sources) to make sure you know what an alternative hypothesis is and how to postulate one. One more thing to know for this section: The alpha level is typically set to .05. Yours wasn't.

Articulate the null hypothesis

Please select a choice

There is no difference in GPA between male and female students.

Articulate the alternative hypothesis

Please select a choice

There is a difference in GPA between male and female students.

Specify the alpha level

Please select a choice

0.05

t-test: Please fill in the blanks:

opener

t-value problem

df problem

p value problem

effect size prob

null hyp prob

degrees of freedom =

103

So, first off, I noticed an error in your calculation of the t-value, which you were supposed to take directly from SPSS output. Any kind of mistake here is due to maybe one of 2 factors: (1) An incorrect test was performed. Please double-check the SPSS step-by-step guide. (2) A rounding error (3) A misunderstanding of what you were looking at

When I look both at your worksheet and the Word doc, I can see a problem with your calculation of degrees of freedom. This, too, was provided for you in the SPSS output for the independent samples t-test. If there is an error here, it’s likely because you copied the wrong number or maybe there’s some sort of rounding error. Remember that since Levene’s test was found not to be significant (we fail to reject the null hypothesis which said that equal variances are assumed), we can use the “equal variance assumed” row (I provided a hint in a red box, right on the worksheet to help you with this).

As for p-value, well, we know that p-value is what we've been chasing, right? Unfortunately, it doesn't appear to be correct in your assignment submission. Please check to see that you copied it correctly from the SPSS output. Also, keep in mind that since the Levene test was found to be non-significant, we can use the “equal variance assumed” row I provided a hint in the worksheet itself about this (you couldn't miss it -- it's in a big red box).

Effect Size: The effect size you listed is not correct. As discussed within this unit, SPSS doesn't calculate the effect size for you. You have to compute it manually using a formula that you'll find in your course materials. I also placed a hint about this in the worksheet. How big of a hint? Well, I did provide the formula. As for interpreting the magnitude of the effect size, Warner (2013) includes a table (5.2) that should help. The announcement I'd posted recently should have helped you with this, as well; I would review it.

Importantly, contrary to what you wrote in your assignment, the null hypothesis should be rejected in this case because it is less than 0.05, our present alpha level). Why is this the case? Please take a look at the textbook as well as other course and non-course materials to get a better understanding of hypothesis testing.

As I glance down your paper, I see some problems in what should be simple transcription of descriptive statistics from SPSS output. It looks as if the following aren't right: mean for females; standard deviation for females; mean for males; standard deviation for males. So, either you made a mistake when you copied the numbers (or filled out the worksheet; note that if the numbers appear as correct in at least one of those formats, you received partial credit), OR you may not have followed carefully enough the step-by-step SPSS guide.

As for the mean difference, the correct answer should have been 0.28 (rounded), which should have been a relatively manageable sort of calculation.

Onto confidence intervals: We learned a unit or two ago that confidence intervals help us understand the range of numbers within which a given parameter of interest probably lies. In this part of the assignment, you had to specify the lower and upper bounds of that range, and the SPSS output should have come to your rescue and provided you the 95% confidence interval for the difference between the 2 means. Since we are assuming equal variance here (because Levene’s test was non-significant), the lower and upper bound numbers should have been offered in the first row of the results. Something went wrong here so I would revisit the material, and maybe re-run this in SPSS to see whether you get different results.

Let's continue on to t-tests, your understanding of which is evaluated in the 5th segment of the scoring rubric for this unit. Remember that a t-test is all about determining whether the means of 2 groups are different in a statistically significant way (meaning that they're different but not simply because you sampled in a certain way but because if you kept sampling over and over, you'd see the same kind of effect). To get to a p-value which helps you understand whether you are indeed looking at a statistically significant difference, you've got to calculate some other statistics, as well. So, first off, I noticed an error in your calculation of the t-value, which you were supposed to take directly from SPSS output. Any kind of mistake here is due to maybe one of 2 factors: (1) An incorrect test was performed. Please double-check the SPSS step-by-step guide. (2) A rounding error (3) A misunderstanding of what you were looking at When I look both at your worksheet and the Word doc, I can see a problem with your calculation of degrees of freedom. This, too, was provided for you in the SPSS output for the independent samples t-test. If there is an error here, it’s likely because you copied the wrong number or maybe there’s some sort of rounding error. Remember that since Levene’s test was found not to be significant (we fail to reject the null hypothesis which said that equal variances are assumed), we can use the “equal variance assumed” row (I provided a hint in a red box, right on the worksheet to help you with this). As for p-value, well, we know that p-value is what we've been chasing, right? Unfortunately, it doesn't appear to be correct in your assignment submission. Please check to see that you copied it correctly from the SPSS output. Also, keep in mind that since the Levene test was found to be non-significant, we can use the “equal variance assumed” row I provided a hint in the worksheet itself about this (you couldn't miss it -- it's in a big red box). Effect Size: The effect size you listed is not correct. As discussed within this unit, SPSS doesn't calculate the effect size for you. You have to compute it manually using a formula that you'll find in your course materials. I also placed a hint about this in the worksheet. How big of a hint? Well, I did provide the formula. As for interpreting the magnitude of the effect size, Warner (2013) includes a table (5.2) that should help. The announcement I'd posted recently should have helped you with this, as well; I would review it. Importantly, contrary to what you wrote in your assignment, the null hypothesis should be rejected in this case because it is less than 0.05, our present alpha level). Why is this the case? Please take a look at the textbook as well as other course and non-course materials to get a better understanding of hypothesis testing. As I glance down your paper, I see some problems in what should be simple transcription of descriptive statistics from SPSS output. It looks as if the following aren't right: mean for females; standard deviation for females; mean for males; standard deviation for males. So, either you made a mistake when you copied the numbers (or filled out the worksheet; note that if the numbers appear as correct in at least one of those formats, you received partial credit), OR you may not have followed carefully enough the step-by-step SPSS guide. As for the mean difference, the correct answer should have been 0.28 (rounded), which should have been a relatively manageable sort of calculation. Onto confidence intervals: We learned a unit or two ago that confidence intervals help us understand the range of numbers within which a given parameter of interest probably lies. In this part of the assignment, you had to specify the lower and upper bounds of that range, and the SPSS output should have come to your rescue and provided you the 95% confidence interval for the difference between the 2 means. Since we are assuming equal variance here (because Levene’s test was non-significant), the lower and upper bound numbers should have been offered in the first row of the results. Something went wrong here so I would revisit the material, and maybe re-run this in SPSS to see whether you get different results.

t-value =

1.999

p-value =

0.048

effect size (eta squared) =

0.03734728

descriptive stats problem

mean diff

CI problem

For the next question, please select from the drop-down box.

As I glance down your paper, I see some problems in what should be simple transcription of descriptive statistics from SPSS output. It looks as if the following aren't right:

As for the mean difference, the correct answer should have been 0.28 (rounded), which should have been a relatively manageable sort of calculation.

mean for females; standard deviation for females; mean for males; standard deviation for males.

Based on the results of the t-test, should the hypothesis be rejected?

Please select a choice

Yes

So, either you made a mistake when you copied the numbers (or filled out the worksheet; note that if the numbers appear as correct in at least one of those formats, you received partial credit), OR you may not have followed carefully enough the step-by-step SPSS guide.

Descriptive statistics for FEMALES

Descriptive statistics for MALES

Mean =

2.9719

2.691

mean for females;

mean for males;

Standard Deviation =

0.67822

0.73942

standard deviation for females;

standard deviation for males;

Mean difference between the means for males and females =

0.2809

Please enter the minimum and maximum of the 95% confidence interval for the difference between means for males and females

lower =

0.00215

0.55965

upper =

What is the main strength of a t-test?

Please select a choice

It can evaluate whether there's a statistically significant difference between 2 groups.

The last section where you had to evaluate the strength and limitation of a t-test is another of those you-either-know-it,-or-you-don't kinds of situations.

One of your answers here is incorrect, so I would check out the material and see where things didn't go the way they should have.

Both of your answers were incorrect, so I'm not sure what happened here. You might just want to check out the course material and get this moving in the right direction again.

The last section where you had to evaluate the strength and limitation of a t-test is another of those you-either-know-it,-or-you-don't kinds of situations. Both of your answers were incorrect, so I'm not sure what happened here. You might just want to check out the course material and get this moving in the right direction again.

What is the main limitation of a t-test?

Please select a choice

It cannot evaluate the means between more than 2 groups.

Please select a choice

strongcorr

Strong positive

Strong negative

Weak negative

Weak positive

Please select a choice

evallp

It can evaluate whether there is a relationship between 2 variables.

It can evaluate whether one of the variables causes another variable.

It can evaluate whether the sample size needs to be larger.

It can evaluate whether there is a relationship between homoscedacity and homogeneity.

Please select a choice

astudio

It cannot evaluate whether there is a relationship between 2 variables.

It cannot evaluate whether one of the variables causes another variable.

It can evaluate whether the sample size needs to be larger.

It can evaluate whether there is a relationship between homoscedacity and homogeneity.

Put an "X" in Col H to describe quality of DAA writing