One way ANOVA

| December 16, 2015

One way ANOVA

Is the statistical test clearly and accurately defined?
Not fully correct. I think your team tried your best, but you screwed up on one very important point – on slide 1 you stated, “A one-way ANOVA has one nominal variable and three or more independent ratio variables” – this is incorrect. An ANOVA has THREE OR MORE groups in your nominal variable (as your independent variable) and ONE interval/ratio variable (as your dependent variable).
Does the team explain what levels of measurement (or what types of variables) are appropriate for the test?
Your team explained the levels of measurement, you just did not specify them correctly, as explained above. You guys simply swopped the two.
Are the variables presented in fully developed hypothesis statements?
Yes, you guys did this correctly *verbally*, but not in the written form on Slide 4 and slide 12. In your revision, I suggest elaborating and stating all the Hypotheses in full sentences.
Does the team clearly explain the independent (predictor) and dependent (outcome) variables in each of the 2 (or more) distinct examples/hypotheses? You can have multiple hypotheses to a single SCC example, but not the other way around.
Yes, you guys did this correctly.
Does the team discuss how the variables can be measured? In other words, are the key variables (or important terms) fully defined both conceptually and operationally?
Yes, you guys did this correctly. In your revision, you can more specifically state, “Conceptually,…. Operationally,…” to make sure your audience don’t miss it.
If a composite scale is suggested to measure an interval variable (e.g., an attitude, a perception, likelihood of a behavior, etc.), are the indicators/items presented and the concepts/constructed discussed? Does the team effectively discuss items used for each scale/measure during the presentation?
N/A. For your SCC examples, you did not use a variable that would call for a composite scale.
Does the team state both the null hypotheses and the research hypotheses?
Yes, you guys did this correctly, although not in a fully written form. In your revision, please update.
Does the team demonstrate an unusual understanding of the statistical test, beyond an average student?
You guys did OK. I wouldn’t say you demonstrated an unusual level of understanding. You could have demonstrated an unusual understanding by adding a Levene’s test to the analysis. See my comment below.

Mastery of SPSS to Conduct the Statistical Test (15%/30%)

Does the team show how the hypotheses can be tested using the statistical test in SPSS? The team can use fake data to show how to interpret SPSS output.
Yes, you guys did this correctly overall, but I would like your team to update 2 things in your revision:
On Slide 5, you stated, “Open up SPSS program and click on variable view and input the labels for your variables” – you should add one more slide to discuss data entry. Show how to do that on your “Data View”.
On slide 9, yes, most intro text would say just to use Tukey and call it a day. But as you guys stated, “For this example we will want to assume equal variances. The Tukey test will test for pairwise comparisons as well as control the type 1 error and generate confidence intervals” – if you want to be more rigorous, you will run “Homogeneity of Variances” test (aka Levene’s test), and the result will help you to consider if you want to consider “Equal Variances Not Assumed” and choose “Games-Howell” (instead of Tukey) test. – Morgan et al does this well. Please consult and follow it.
I showed this procedure in class. Don’t know if you guys remember that. What I described above is to make your ANOVA test more parallel with t-test, which automatically gives you a Levene’s test result.
Does the team show how to read and interpret the associations or differences being tested?
Yes, you guys did this correctly.
On Slide 11, you stated, “We now see that the sales don’t differ between the head and the chest shelf height. But sales did differ between the head shelf height and the waist shelf height” – you could have extended the interpretation and said the analysis shows that the difference between head level shelf and waist level shelf is -4, indicating that product placement at the waist level actually generated more sales than product placement at the head level. The strategic decision would be to capitalize on the waist level shelf space, and to always keep the waist level shelf fully stocked, even if that means to keep the head level shelf space empty.
Does the team show how to read and interpret effect size appropriate to the statistical test?
No, you guys did not do that. But for an ANOVA, it is ok.
Does the team show how to read and interpret the p-values? Are results presented clearly relative to each hypothesis? Does the team discuss the appropriate scope of generalizability of the results?
Yes, you guys did this correctly.
While your script is correct, Alex did not read it correctly on Slide 11, “Significance is measured by a p value that is <.05” – he read “greater than 0.05” which is wrong.
On Slide 11, you stated, “SPSS notes with an asterisk when data is significant” – not quite, SPSS notes with an asterisk when THE DIFFERENCE BETWEEN TWO GROUPS is significant.
Your team did not discuss the appropriate scope of generalizability.
If a composite scale is used, does the team explain Cronbach’s alpha to establish the reliability of the scale using SPSS? Does the team establish the validity and reliability of the evidence gathered and tested in SPSS?
N/A. For your SCC examples, you did not use a composite variable.
Does the team show how to write up the results using APA style? See the “Results” paragraph in Morgan et al (and see Wrench et al for more examples)
No, you guys missed this one.
Does the team demonstrate an unusual mastery of SPSS, beyond an average student?
You guys did OK. I wouldn’t say you demonstrated an unusual level of SPSS mastery.

SCC/COM Examples & Overall Presentation (15%/20%)

Does the team provide at least 2 distinct SCC/COM examples/hypotheses (e.g., one in ‘employee relations,’ and another in ‘image and identity.’)
Yes, you guys did this correctly.
Are the examples of SCC significance? Are they worthy of SCC research efforts? Are the “so what” of the hypotheses (or expected findings) clearly articulated with practical suggestions for implementation of findings? Are the examples impressive? Would the department chair be wow-ed if she sees this presentation?
Yes, you guys did this correctly.
However, for “Are the “so what” of the hypotheses (or expected findings) clearly articulated with practical suggestions for implementation of findings? Are the examples impressive?”, you guys did not explain why the audience should care about your SCC examples.
Is the statistical test discussed in a way that shows how the test can be applied across various SCC/COM topics (e.g., employee relations, identity and image à corporate advertising, etc.)?
Yes, you guys did this correctly.
Does the presentation/ discussion suggest how the assigned statistical test can be used to investigate future corporate problems and/or organizational issues? These applications can be examples of problems/issues common to most companies.
No, your team did not go into this.
Is the presentation in the form of a voice-over PowerPoint file?
Yes, you guys did this correctly.
Is the verbatim script for each slide provided in the textbox?
Yes, you guys did this correctly.
Are the slides visually enhanced with images?
Yes, you guys did fine on this. But I always think visuals can be enhanced.
Does the presentation appear as “one voice”, and not simply a collection of fragmented pieces? It is really bad when a team divides up the work and strings fragmented pieces together, instead of working on the project “as a team” to make sure the whole presentation flows and is coherent.
Yes, you guys did this correctly.
Does the team effectively answer questions from the audience?
Yes, you guys did well.
Does the team show that they have gone above the basic expectations, beyond even what an excellent team would do?
You guys did OK. I wouldn’t say you have gone above and beyond, because on Slide 12, you provided a good example, “Suppose a store owner wanted to examine the possible relationship between music genre played in the store and the money spent. Over the course of 2 months the music genre alternates every other day between three genres; hip-hop, pop, and classical music. The independent variable is the genre of music being played at the store and the dependent variable is the amount of money shoppers are spending. We hypothesised that shoppers will spend more money when hip-hop is being played in the store and our null hypothesis is that the genre of music being played does not affect the amount they spend. To test this hypothesis we use a one-way ANOVA and be able to determine if music genre influences shoppers’ spending habits” – in you revision, perhaps you guys can generate another (fake) SPSS output to guide your audience through an interpretation exercise. Some teams did two full examples. But you guys only did one. So I wouldn’t even say doing two full examples is “going above and beyond”.

Multiple Choice Questions (7%/20%)

Write 10 well-crafted multiple choice questions with 5 answer choices for each question.
The questions must include an explanation of the answer – why the correct answer is correct AND why the incorrect choices cannot be correct.
Indicate the correct answer with (***) AND the page number reference in the chapter.
The multiple choice questions are emailed to the professor after the presentation, but within 24 hours from the end of class.
[No, you guys missed the deadline, so you lose 10% (or 10 points) of the project grade]

TOTAL SCORE: 63%. But you turned in your Multiple Choice Questions late, so you lose 10%. FINAL SCORE = 47%. Alex was fired right before the Multiple Choice Questions were turned in. So Alex will not receive any points on the last section. His score will be based on the first 3 sections.
Sample Questions
(from another class on ‘big data,’ related to but not research methods)

Sample Question #1: What is one of the most difficult things for society to accomplish in order to succeed in the era of big data?
a. New processing technologies
b. Not knowing why, only knowing what (***, pp. 18-19)
c. By changing the amount of big data, we change the essence
d. Big data is about predictions
e. Big data is about descriptions

Explanation: Choice (b) is correct because, according to Mayer-Schönberger and Cukier, “big data refers to things one can do at a large scale that cannot be done at a smaller one” (p.18). In this way, our society must “shed some of its obsession for causality in exchange for simple correlations” (p.18). This is the hardest concept for our society to understand and adopt because of our long history with making decisions and viewing reality that is always based on why. You can find the specific quote and answer on pp. 18-19, “Most strikingly, society will need to shed some of its obsession for causality in exchange for simple correlations: not knowing why but only what.”

Choice (a) is incorrect because big data is more than just technology. Although “Initially the idea was that the volume of information had grown so large that the quantity being examined no longer fit into the memory that computers use for processing,” Mayer-Schönberger and Cukier say there is no single, specific definition of big data. Our society is constantly evolving and innovating new technologies, and this has never been a new concept to us. While new processing technologies are, indeed, important to the success of big data, they are not the most crucial aspect that we, as a society, must learn to overcome. AS stated in the text, “the real revolution is not in the machines that calculate data but in data itself and how we use it” (p. 19).

Choice (c) is incorrect because although this is a true statement, it is not an obstacle that our society must overcome in order to ensure that big data is successful. The example used in the book to explain this concept is that “A movie is fundamentally different from a frozen photograph,” but that a picture drawn by hand and a photograph of the same image are quite similar (p. 22). While it is important to be aware of the new characteristics that are created so that we can adapt, this is a familiar act to our society. We are constantly changing and evolving and adding new and increasing information to computers. Unlike choice (b), we do not have a long history or tradition of being unable to grow with the acquisition of new information of the maths and sciences. Choice (b) is a better answer because changing mindsets and attitudes is always more difficult.

Choice (d) is incorrect because this is not a problem for our society. This is exactly why we want to use big data. The machines that are built to operate and help calculate large amounts of information “change fundamental aspects of life by giving it a quantitative dimension it never had before” (p. 24). Already these systems look out for patterns so they can constantly improve predictions and be more accurate. An example is Amazon and how it can give recommendations based off of what you and other users have clicked on and showed a liking to. This is not an appropriate answer to the question because this is precisely why big data exists, to help apply large amounts of data and make predictions that we could not do ourselves.

Choice (e) is incorrect because this is not a problem for our society. Descriptions are the opposite of predictions in (d), and the use of big data is not for description.

Sample Question #2: Which of the following is an example of ways that random sampling succeeds in gathering the accurate information for populations?
Random sampling scales easily to include subcategories
Random sampling reaches a niche audience effectively and precisely.
Random sampling reduces big data problems to more manageable data problems (***, pp. 23-25)
Random sampling removes all chances of biases.
Random sampling is for nonparametric tests.

Explanation: According to Chapter 2, pp. 23-25, random sampling is effective in low cost and efficiency terms, yet the data can be skewed due to many different variables.

Choice (a) is not the correct answer because random sampling does not scale easily to include subcategories. Breaking the results down into smaller and smaller subgroups increases the possibility of erroneous predictions. (p. 24).

Choice (b) is not the correct answer because random sampling does not reach a niche audience precisely. The tiny biases in the overall sample will make the errors more pronounces at the level of subgroups.

Choice (c) is the correct answer because random sampling “reduces big-data problems to more manageable data problems.” It is used to make improvements much easier and less costly.

Choice (d) is not the correct answer because random sampling increases the number of biases especially when dealing with a niche audience.

Choice (e) is incorrect because random sampling is for parametric test.

GUIDELINES FOR FIRING TEAMMATES
If your team reaches a consensus to fire a team member, please follow the steps outlined below:
Give the member a warning (perhaps in an email first for documentation, and then follow up with a phone conversation and/or face-to-face to make sure the message is received and understood). State specifically your team’s reasons. Make it objective and not personal. (e.g., “We respect you as a person, and we know you have the capability to do the work. However, the work consistently fails to meet the team’s deadlines/expectations.” So you point out what is lacking in the work that could be improved if s/he chooses to).
Respectfully ask to find out the reasons behind this team member’s consistent failure to meet deadlines/expectations. Perhaps there were misunderstandings and miscommunication. If so, this is your team’s chance to clarify.
Your team needs to give at least one warning before firing a member. Along with the first warning, request a commitment to specific actions and a timeline if the member wishes to stay on the team. Be specific with dates/times/qualities/quantities/standards for actions. State that if the person does not follow through by a certain date/time, and remain consistent until the project is completely done, then the person will be fired within your team’s specified timeframe/tolerance/grace period.
Furthermore, you can also state what the member has to do to earn and/or regain the trust and confidence of the rest of the team members. Be specific! Ask for commitment.
If s/he begs for mercy, decide what the rest of the team members are willing to compromise/negotiate without resenting it later. Your team makes the call.
If your team has given this member multiple chances (or warnings) already and has indicated the possibility of firing him/her (i.e., Step 1 has already been carried out clearly), you can decide to fire the member now. If you do, do it gently, save face in the process, be a bit tactful. Document it clearly.
Explain to the fired member that his/her scores from team assignments before being fired will remain on his/her Blackboard gradebook. However, everything from this point onwards is on his/her own effort. That means, if the team project is not completed yet, the fired member will carry out his/her own final project individually. S/he cannot use any of the materials already shared with the team, including his/her previous contribution, in his/her own individual project. When fired, the fired member’s previous contribution remains the materials of the original team, because the materials were gathered and/or produced within the context of the team.
Learn it now or learn it later. Here are the real-life lessons: (a) If you get fired: free riders don’t always get away with not contributing to collective efforts yet benefiting from others’ hard work. If you got away with it once, that was your lucky break. (b) If you feel like firing a member: Learn to communicate realistic and reasonable expectations objectively and effectively in life. Otherwise people will always disappoint you, or you will resent working with people all the time. This can happen at school, at work, or even… at home! The source of the problem may be you, not others.
Add or improvise as you see fit. You and your team members know the situation the best. This is just a list of general suggestions.
Those who get fired can get adopted by another team or form their own team with other classmates fired by other teams.

Get a 30 % discount on an order above $ 50
Use the following coupon code:
COCONUT
Order your essay today and save 30% with the discount code: COCONUTOrder Now
Positive SSL