The management-research question hierarchy is a method of ...
Unit 1.2
The management-research question hierarchy is a method of approaching research that breaks down into steps a specific dilemma. These 6 steps provide a road map for reaching a logical, and hopefully meaningful, conclusion to a specific dilemma. The steps cumulate with a recommendation(s) for management on how to solve the dilemma.
Below I have applied the management-research question hierarchy to the NCRCC case.
Management Dilemma
Attract more members in order to support the renovated facility.
Management Question
How do we increase membership?
Research Questions
How do we improve service?
Do we modify and improve the golf course layout?
Why did the last membership "drive" have only moderate success?
Who do we want to focus on attracting… young, older, local, single, …etc?
Should we add a pool, health club, and/or tennis courts?
Can we afford to upgrade if we can't support our current renovations?
Investigative Questions
How many members will you lose if you increase dues?
If we do a survey of potential-members, how would the results differ?
Besides a survey, how else can we gage member satisfaction and desires?
Measurement Questions
The survey in the book is fairly detailed in terms of questions. There are a couple questions I have that would need to be derived from the survey answers.
How do the answers we get compare to other country clubs of similar size?
If current member demographics are not the same as the demographics of the new members you are looking for, how valuable is the survey?
Management Decision
Obviously this is unknown. Given the existing data, I would tend to think that adding a pool, tennis courts, and a health facility would be the way to go. Attracting a younger member means more future years of a profit generating member. Also, I found it odd that less than a third of all members wanted the addition, but a healthy 41% belonged to a pool or health club in addition to NCRCC. These additional facilities clearly need to be added if the goal is to increase the membership via younger families.
NOTE:
AFTER I wrote my response, I called NCRCC (I live about 45 minutes from the club). Last spring they added a pool and health facility, but no tennis courts. They have totally redesigned the golf courses by moving tee boxes and greens. The have re-sodded both driving ranges. They have stricter clothing standards for both members and employees. And next month they are re-paving the parking lot.
By the way, golf playing memberships are currently at 683 members… their goal was 680.
Bari Courts
Unit 1-2 R
Kelly,
I agree that new facilities must be added to attract the younger age groups need to build and sustain growth. Your suggestion to use the pay per use fee for new facilities is also a good idea. It would be a legitimate method of generating the revenue needed to support the new facilities. I do have some reservations about how to get it to work. First, the current members were not really in support of adding new facilities (only 30%). Thus, most would be very reluctant to want to pay more for facilities they feel they do not want. Second, new members would not feel it appropriate that their dues are higher to finance the additional facilities that all members are entitled to use (even if they don't want to use them).
My brother works at a country club and I ask him how they would handle this situation. He said that they are very picky about everyone paying the exact same amount in annual dues, but new membership fees are a little more flexible because they are a one-time occurrence. My brother indicated that his country club goes in cycles where they lower membership prices on occasion in order to attract more members. And on other occasions they raise the membership price in order to finance upgrades.
It looks like my brothers country club found a way to institute you two-tier dues idea without offending anyone.
Bari Courts
Unit 1-3 R
Carol,
I did not use the Analyze, Descriptive Stats menu on SSPS. I simply sorted by "sea" and then did everything using the bar graph feature. The SSPS program is basicly a spreadsheet so figuring the averages in each area was no problem.
I read the definition of descriptive statistcs in the text book and was trying to use the data to determine an interesting result that I though might prompt more questions.
Hopefully this helps.
Caren,
First… case information is on page 664… it helps a lot.
Second, I also looked over the SSPS program and have figure out a couple things. I'm pretty good with graphs so I played with it the most. I hope to spend more time on the Anaylze menu this week as I think it is suppose to do some cool things. I am most familiar with Excel so I spent time making it do things I know Excel can do like sorting and filtering.
I personally would like to look at the data in terms of regional averages. There seemed to be some interesting findings that could be derived from comparing the west to the south, midwest, and east.
While I actually enjoy learning new computer program, it certainly is a time consuming process. I sure could use the SSPS for Dummies book right now!
Bari Courts
Unit 2 Questions from Healthy Lifestyles
Q1
A couple interesting numbers jump out at me.
Alabama is 10 percentage points higher than any other state in the no leisure time off the job category.
Only 5% of the people in Hawaii don't use a seat belt (or only occasionally), but a whopping 60% admit to not using seat belts in North Dakota (and 57% in South Dakota). Does it not snow in the Dakotas? Is driving on snow and ice without a seat belt a good idea? I'd like to see the traffic mortality rate in the Dakotas.
Q2
I weighted the statistics by assigning the "best" percentage a 1 and the "worst" percentage a 45. In the event of a tie, I used the simple average of the tied states. The sum of the weighted averages yielded my "most healthy lifestyle" winner… West Virginia! It is interesting to note that West Virginia was near the bottom (39/45) in seat belt usage, but was 14th and above in every other category. Also of note, I travel to West Virginia on occasion for business and it is one of the few states with a speed limit of 70 mph.
The big loser in my weighted averages was South Dakota, 25th or higher in every category.
Florida ranked second just behind WV in my rankings with a small gap. California was about a 50% increase in the sum of the weighted averages (73 to 113.5). And Minnesota was a very "unhealthy" 140% above West Virginia (73 to 180.5).
Q3.
My data is limited in many ways. First, I omitted the states that did not respond. Is it a true reflection of the US if there are missing states? Why is DC considered a state?
I used a fairly simple weighted average example. I not a statistics guru, but would think a better method of weighing the averages might be a logarithmic graph. Due to time constraints, I did not dwell into the logarithmic graph option.
Note: I am still playing with SSPS and have discovered many things it can do. Unfortunately, I can never seem to make it do what I want it to do. In the above calculations, I reverted back to Excel to provide the calculations I needed.
Bari Courts
Unit 2-1R Response to quantitative literature critique
Carol,
I dug up the article you critiqued and agree it is quite long. I did not read it all, but did seem to think it was redundant in many places. I was interested in the article because for 10 years I was an entrepreneur. I am in the corporate world right now, but see myself getting back to the entrepreneurial life style after I get the necessary industry experience. It is interesting that their findings are the things I dislike about corporate life so much.
Other items about the study that I found interesting. The survey used only people who were currently entrepreneurs, but left out any like me who had been their own business owners. With only 153 people responding to the survey, I'd have liked to see a bigger data set.
Regarding your method of critiquing, there was not much I thought was wrong. I would like to have seen a section on the background or history, and I probably would understand more if the analysis section had more statistics. However, your template seems logical, flows nicely, and provides all the information needed in a concise format. Good job.
Bari Courts
Unit 2-2R Response to Healthy Lifestyles
Carol,
After looking over your results I find that you weighted the data in a different manner than I did. In fact, I have found a few different interpretations of the word "weighted" in our course room. Yours is interesting to me because you weigh the statistics against the other statistics. The sum of all your statistics equals the whole data set. Interpreting your data is more beneficial when taken in context with the other statistics. Meanwhile, I weighted each of the 7 statistical columns individually. My data seems to best be analyzed in isolation to the other statistics. While you are in essence weighing data horizontally, I am weighing the data vertically. Of course, neither is wrong, but I am surprised at what completely different information can be interpreted depending on the way the data is organized.
As a side note, I have been thinking about the omission of the 4 states in the data and how everyone (including me) says it is a "limitation" in the data. Is it really? With DC added, we still have data on 92% of all possible respondents. I can't think of many studies whose sampling includes 92% of all possible respondents. In my critique of the quantitative article I read for Unit 2-1, the surveyors only tested 16 kids. I'm sure they would have liked to have had 92% of all kids who qualified for their test. I'm beginning to think 47 of 51 states responding is a good representation of the actual data.
Bari Courts
Unit 3-1 Qualitative Journal Article Critique
Bibliography
Langer, Stephen, (1998). Compensation and Benefits in Consulting Firms. Journal of Management
Consulting, 10, (2), 27-30.
Summary of the research article
Purpose
The purpose of this study was to provide information regarding the compensation packages of employees at consulting firms. The researchers were looking for solid information regarding the benefits in the consulting field.
Background
The researchers of Abbott, Langer, and Associates conduct research on compensation and benefits of consulting engineering firms. This particular report is an extension of previous work.
Hypothesis
It is postulated that the higher the officer in a consulting from, the larger percentage of this compensation is wanted in the form of benefits, and not salary.
Research design
The research was conducted from a survey received from a hundred 96 consulting firms in the United States. The questions involve to both filling the blacks, opinionated rankings, and straight opinions.
Ethics
The only mention of ethics was the researcher promise that detailed information would not be shared outside their company. Only the results of the study would be released to the public.
Research results
The results showed that indeed a larger portion of an executive's compensation is in the form of benefits at consulting firms. The article also released some of the statistics regarding benefits so that they may be compared to the readers current benefits.
Limitations
There were many limitations to this survey and research. To begin with, the author acknowledges that the sample is large enough to be indicative but too small to be representative of the entire field. The author also admits that the sampling was not good. The people who were mailed the survey came from a rented commercial mailing list that resulted in many returned letters and phone calls. Not every organization received the survey was even a consulting firm. Thus it is difficult to determine what the true size of the mailing or whether they received a reasonable response rate.
Critique
Are the findings reasonable? I find the information that they received to be reasonable given the nature of their mailing list. However, there appears to be no accommodations made to account for region of the country, type of service provided, annual fees charged, or number of employees. These are all demographic variables that appear to be unaccounted for.
Are the findings useful? I suppose there is a use for the findings because the article states that there is very little solid information existing in the field. But given the limitations involved, I'd be hesitant to consider any of the findings important.
Were the findings explain? There was a large listing of their findings, but very low explanation of what the findings mean. They seemed to draw little conclusions from the data they collected, and many questions were left unanswered as I read through the article.
Was the significance of their study? While I am confident that their findings are accurate, I question the purpose of this study. The study seemed to be a last minute thought, or derived from some information they had received for another surveys. Even the article explaining their study I do not feel was well written. It jumps from conclusion to conclusion without any valid explanations, and I can only assume their research was just a sporadic.
Note:
My background is such that I am best suited to use a purely quantitative approach in my dissertation. However, given the last few weeks of readings, I am convinced I will need to incorporate some qualitative research as well. I am making an extra effort to better understand the qualitative aspect of research and I'm working on how to use it in the most effective manner in my dissertation.
Bari Courts
Unit 3-1 Qualitative Article Critique – Response
Melissa,
Your critique is well organized and appears to summarize the study in a most effective manner. One of the things I do when I look at someone's critique is to see if it leaves more questions unanswered than answered. After reading your critique, I feel I understand what they did.
After spending some time trying to find something I didn’t know, I came up with a question...
If their sampling was mostly first year ACP teachers, would this not reflect a bias toward teaching? I would think new teachers in their first year are still in their "honeymoon" period where their new career is just great. Would the study not yield more accurate results by taking a better mix of experienced and new ACP teachers?
It appears you have either forgotten, of left off, a section reflecting your opinion on the article. I would be interested in what you thought of the article. However, I am not trying to be negative, I really feel you did an excellent job of summarizing the study.
Bari Courts
Unit 3-2 QSR vs SPSS
What are the similarities and differences in the SPSS and QSR software packages?
SPSS is mainly a statistical analysis package that is used for quantitative analysis and interpretation. QSR is mainly a qualitative analysis package used to help analyze open-ended survey questions. (Example, how do you feel about this issue and why?)
SPSS is used in commercial, governmental, and academic research, whereas QSR seems to be focused in academic arena with only moderate use in the commercial and governmental areas.
For SPSS, the GUI (Graphic User Interface), or the way it looks and feels, is more like a spreadsheet. But QSR looks and feels like a word processor. Not surprisingly, the SPSS output tends to be a bit bland in terms of visual appearance, but the QSR output can easily take advantage of the font characteristics and spacing benefits of a word processor.
SPSS results come out very logical and analytical. You must provide the interpretation of the data. QSR allows for more subjective results due to its method of grouping and organizing the data. Here you might be given new ideas due to the method or presenting the results.
SPSS stresses how it can handle large data sets. QSR does not say it can't handle large data sets, but the nature of the program makes it seem as though very large data sets would create a problem.
Finally, while reading the QSR information, I kept thinking of how these programs would handle a written test. It seems that SPSS would be used to analyze the True/False section and QSR would handle the essay questions. I realize this is not a "profound" conclusion, but it is an easy way for me to keep the two programs separate in my mind.
Bari Courts
Unit 3-2 QSR vs SPSS - Response
Paul,
My undergraduate degree was in math and I find myself interested in different types of math problems. It is embarrassing to admit, but I actually took a little time to go through SPSS looking for how it did certain math functions. For example, SIN, COS, and TAN equations. I know Excel fairly well and I noticed that SPSS has some built in functions that you need to build a formula for in Excel. For example, computing means and regression calculations.
Also, I see you found references to "ANOVA models, Bivariate (Correlate) analysis, Case Summaries (Reports), Crosstabs (Descriptive Statistics), Descriptive Ratio Statistics, Frequencies, Linear Regression, Reliability, and t-tests". I have heard of a couple of these things, but some are completely foreign to me. In any event, you have provided me some great new words I can use on my kids when I am trying to totally befuddle them!
Bari Courts
Unit 3 – Research Problem
Paul and Kelly,
I am planning on doing my dissertation on the effect of the internet on the insurance industry. I spent 10 years selling insurance before moving into the IT field. I also have a web site I use to sell some insurance (start-a-).
My personal preference is to do a large quantitative study on sales that have been logged so far through web sites. However, the important part of my dissertation will consist of what the online benefits have been despite the poor sales. This will require a qualitative study.
My plan is to use the results of my qualitative survey to explain why sales have been slow, but the cost savings benefits have been good. I then plan to use the study to present a plan for what should be done in the future.
My survey will be mainly focused on the agent view of the internet. Thus, my survey will be with licensed insurance agents across the country. It will be substantially qualitative in nature.
Bari Courts
Unit 4-3 Research topic and problems
As previously mentioned, my research topic is on the effect of the internet on the insurance industry. Specifically, looking at the question from the perspective of the agent/company relationship.
My research hypothesis is that there have been substantial improvements in communication and paperwork flow due to the internet. However, there are still many opportunities that are not being taken advantage of.
My data collection will be in the form of a survey. I intend to survey active, licensed insurance agents in the United States. My survey will be online at a web site I create. The agents will be "invited" to answer the survey by going to the web site.
The agents will be marketed through the use of a service called the Agent Card Pack produced by AMC Publishing. The Card Pack is a collection of advertisements the size of a postcard that are purchased by insurance managers to advertise their products to agents. Typically, the managers want the agent to sell their products. I have used their service in the past when trying to recruit agents. The "National" Card Pack is sent 4 times per year to 50,000 active insurance agents. Having a card in the Card Pack cost about $2500+, but I am confident I can find an insurance company who will find the results important enough to fund the cost of my marketing.
(There are other options in terms of driving traffic to my survey web site. Most notably, I can run advertisements in insurance trade magazines. But this creates issues in determining the size of my sample.)
Given the narrow specifications of using the Card Pack to gather data, I do not anticipate any validity or reliability issues.
The study will only be practical if I can write the survey well and use the results to draw a purposeful conclusion.
Research Problems
There are a few problems I can think of now, and I'm sure many more will appear as I proceed with my study.
1. Will I be able to pinpoint a topic that is researchable and not just my opinion?
2. Can I write a survey well enough to provide me with the information I need to make useful conclusions?
3. Is there a need for personal interviews?
One more item to be addressed is that the survey will be biased in favor of agents who are already on the internet. Due to the nature of my data collection, if an agent is not online, they can not participate in my survey. This is my intention because I need opinions from agents who use the internet, not agents who have never been online.
Bari Courts
Unit 4-1 NCR Case Revisited
Attachments include:
NCR_Descriptive.spo
NCR_LocHome.spo
NCRCrossLocHomeGender.spo
NCRCrossSocialGender.spo
How appropriate were the various sampling design decisions?
I felt overall the design of the sample was fairly good. The survey was meant to extract the feeling and thoughts of the members and to that extent I feel it succeeded. The sampling of members included every member so there was no room for improvement. However, they could have surveyed more non-members since it is the non-members that they are ultimately trying to attract to the club.
Explain how the questionnaire was coded and prepared for data entry into SPSS.
Most of the questions were coded to provide an answer from 1-5. There were occasional questions with only 2 choices (gender, member), and current membership level had only 4 options. This type of coding makes it easy to analyze in SPSS.
Cite examples of missing data or odd occurrences.
I don't like the fact that missing data is represented by 99 in SPSS. I could not really find a reasonable explanation for this. Also, why were there 79 people who did not know their gender, yet everyone answered the membership question? (Apparently the gender question was too vague.)
Cite examples of various variables at different measurement levels.
Most of the questions were coded to provide an answer from 1-5 based on either satisfaction level or importance level. Of course, some questions were easier like member/nonmember and male/female.
Select one variable at each measurement levels and use "frequencies" with statistics and charts and "descriptive" statistical analysis using SPSS's "ANALYZE" pull-down menu.
See attached.
An interesting note on the Location-from-Home statistic I looked at was the fact the graph of the statistic was almost a perfect Bell Curve. You'll recall the Bell Curve is a particular form for the distribution of a variable which, when plotted, produces a ""bell"" shaped curve- symmetrical, rising smoothly from a small number of cases at both extremes to a large number of cases in the middle.
Bari Courts
Unit 4-1 Response to Kelly regarding NCR Case
Kelly,
You put together some nice statistics and graphs on SPSS. Good Job. You mentioned you found the Crosstabs feature useful. Likewise, I also found the Crosstabs feature useful and produced some interesting results from it.
For example, I was interested in the distribution of the answers based on gender. Using the Crosstabs feature I found that more men than women (35% to 31%) found it either important or very important that the club be located near their home. I would have thought the opposite in that men are usually more willing to travel some distance to play on a good golf course and golf is the main attraction for the men of the club.
However, women were much more interested in the social aspects of the club than me (46% to 32%). I would have suspected this result based on the fact that a fewer percentage of women play golf at the course than men.
Bari Courts
Unit 4-2 Ramada Case
The research question is a pretty basic question asked by many companies; how can I improve customer service? As the article mentioned, Ramada was not getting lower customer satisfaction rating yet, but they chose to be proactive in their approach to customer satisfaction.
Their hypothesis was that they needed to improve on what they feel are the 3 most important components of customer service… hiring, training, and motivation.
The data was collected primarily 2 sources. The first was a field force of 24 researchers who visited each Ramada and conducted interviews with management and employees. The employees were given the opportunity to express their opinion in a festive-like atmosphere that was both relaxful to the employee and informative to the research team. The second source of data was the third party interviews and observations with three leading companies of customer satisfaction (Disney, SWA, and TGIF's).
I like the fact Ramada went outside the company to help find answers. It shows they are secure in their business practices, but are not afraid to learn from other companies.
There were many strengths to the approach Ramada took. First, and most importantly, it produced positive, quantifiable results (increase in survey rating). Second, the party-like roll out of the information gathering sessions were a morale boost to the individual locations. And third, the depth of research done can certainly yield additional information on other problems Ramada may have and possible solutions for these problems. For example, the fact that a small group visited each location could yield some additional ideas on good staff management ideas at some locations and bad ideas at other locations. These ideas could then be circulated among the locations for use.
As far as weaknesses involved I can see a couple. The first is cost. A survey of this magnitude had to have been very costly. Most companies could not afford such a costly survey whose ultimate purpose was to improve a subjective measure (customer service). I realize that the customer service rating is a quantifiable measure, but even increasing this rating has subjective results. How can you measure how much more money you will make if you customer service rating goes up by a point? I'm not saying the survey was not necessary, just costly.
Secondly, I felt time was a weakness. Six months of visits probably meant the whole survey took over a year to complete and even longer to implement the changes. Perhaps the fact that they were proactive in addressing the problem in the first place yielded them the lengthy time they needed to do such a good survey.
Finally, it is hard to be critical of this survey based on the outstanding results it yielded. Given the results, how much better could it have been handled?
Bari Courts
Unit 4-3 comment
Dr. Hollis,
I believe I can control my population via the mailing. The mailing is through AMC Publishing and they send a packet of postcards to insurance agents in different states. For example, Ohio may go to 25,000 while West Virgina's list is only 8,000. I can send a combo pack of 100,000 names to a a list of insurance agents from all 50 states. I will have to do a bit of work with AMC to get the percent undeliverable, but I believe I can get what I need from them.
Thanks,
Bari
Unit 5-1 NCR – Fitness Center Issue
Overall, I felt the survey was adequate for retrieving the information they wished to get. I felt the McMahon group made an effort to arrange questions in a logical order. They also made attempts to provide clear directions. However, there are a few items I would address to improve the survey.
The survey is rather long, and I do not feel that long is necessarily bad if it is done correctly. Unfortunately, the McMahon survey is too long because it has redundant and non-useful question. For example, question #14 is looking for satisfaction levels for the food service facilities. There are 4 questions for Service, 4 for Food, and 5 for Other. That is 13 questions just to get the members opinion on the food service. Limiting this section to maybe 8 questions would have probably yielded the same conclusions and would have cut down on the length of the survey. There are other sections that could be improved in the same way.
I believe it takes a couple moments for members to get focused on the club when they start to fill out the survey. Thus, asking their overall opinion of the club in the first question is not proper. I feel the overall feeling of the club should be asked at the end of the survey. In addition, moving the demographic questions in the "About You" section to the front will allow members to get focused on the survey while answering generic, quantitative questions.
While I'm on the subject of members, why did they only survey members? If they are trying to attract people who are potential members, would it not make sense to survey some non-members also?
There were also some ambiguous questions. For example, question #21 ask member opinion on eliminating some outing during the week and if the member would be willing to have a dues increase to improve the course and make up for lost revenue. If outing were eliminated, revenues would be lost, but what does that have to do with improving the course?
Regarding the fitness center issue, I would add the fitness center. There are 2 main reasons why I'd add the fitness center. First, 25% of all members belong to a fitness center someplace else. By adding a fitness center you can potentially attract the 25% who are working out elsewhere. These members would be spending more time at the club and thus be spending more money at the club on food and beverages.. Second, almost half (49%) of your target age group to recruit (under 46) feels a fitness center is either important or very important.
I have attached an SPSS output supporting my recommendation.
Bari Courts
NOTE:
I called Jay at the NCR Golf Club in Dayton (I live in Cincinnati). He said they added a fitness center late last year that includes a pool. There is no spa, and tennis courts have not been added. Jay says there are no plans for either. He did say most improvements in the last 6 months have been to the golf courses and driving range. I asked about membership levels and he says they have remained level at the 680 member range. His opinion was the fitness center was nice, but did not seem to be a major influence on attracting new members.
Unit 5-2 Raid Display Experiment
Question:
Given the new Raid displays, do the displays meet the desired requirements for customer and retailer satisfaction? If not, what can be done to improve the displays?
In trying to think "out of the box", I tried to approach the problem from a purely quantitative standpoint. My experiment starts with the Insect-a-Guide. There are 2 options with the first being an interactive touch screen that is informative to both the consumer and the seller. The consumer gets all the bug information they want. The seller gets statistics on what the consumer looked at. For example, how many hits were made on the "mosquito's information" screen. These statistics can be used to determine interest in different products.
The second option is a little less high-tech, but requires simple brochures and not the touch screen. The brochures would still provide a statistic on the use of the information but far fewer details than the touch screen.
To test visual appearance, vary the colors and layouts at different stores. Comparing sales between stores may not yield much useful information because of all the variables involved (time of year, store location, weather during the testing period, etc…), but comparing the sales during to testing period to prior years sales at the same store would be beneficial. These numbers could then be converted to percentages to allow comparison between stores.
While the above steps in me experiment will yield quantifiable results, I have yet to address consumer and retailer satisfaction. I suppose it is time to move to a qualitative solution. Again, trying to think out-of–the-box, I propose short surveys on the can of Raid. Offer a free can of Raid in exchange for a completed survey. The survey can be either the peel-off type on the can, or can be folded up and stuck on the lid. You can limit the amount of surveys dispersed in order to keep from giving away too many free cans of Raid. The surveys will address the satisfaction levels of consumers. I see no reason similar surveys can not be circulated to store employees and managers requesting the same type of information.
OK, so there would probably be drawbacks to my experiment. For example, is a survey from a customer really worth the price of a can of Raid? It could cost a fortune to get enough surveys to be useful. Also, the brochure idea would not really tell us how the information is used. If sales did not increase, would that mean the displays are to blame, or are there too many additional variables that can not be controlled (weather in the store location, are bugs heavy this year, etc…)
I'm not even sure my experiment would work, but I felt it better to list another option than just rehash some of the other good, and probably more practical, ideas of my classmates.
Bari Courts
Unit 5-2-Response
Raid Display Experiment - Response
Chan,
I liked your idea of such a comprehensive study that includes assessing the bug-knowledge level of both consumers and employees. I do wonder about the need to spend time in determining "how effective the employees will be at conveying any unclear information to the average consumer." We don't have a big bug problem here in Cincinnat, but do not imagine many people in stores ask questions to the store employees about which product to use on which bug. I would think Raid would be best served by working under the assumption that their display and cans will be the sole source of information for consumers who have questions about which product to use.
Another question I have has to do with how demographics are handled in you experiment (and mine for that matter). Different areas of the country have very different needs for Raid products. For example, a southern store, near a body of water, with a humid climate would have a much different need for Raid products that an northern town where it is below 50 degrees for 6 months of the year. Is the answer to do large quantities of research to try and get representative data for most scenarios? Is this amount of data feasible to collect? I don't have the answer, just the question.
Bari Courts
Unit 6 - #1 – Rubbergate
Rubbergate is a name given to the scandal in the Congress involving an abnormally large number of bounced check written by members of Congress. A statistician has been asked to collect and interpret some data about the scandal for a lobbying firm in an attempt to gage how the scandal will affect the upcoming elections. Below is a general description of the data. (Note, attached is the full statistical output.)
SERVICE (total years served in Congress)
Mean = 10.64
Median = 8.5
Std. Dev. = 8.04
Range = 42
Min = 0
Max = 42
AGE (Congress members age)
Mean = 53.26
Median = 53
Std. Dev. = 10.10
Range = 48
Min = 32
Max = 80
VOTES (percent returned for the most recent election the Congress member was in)
Mean = 69.407
Median = 65.5
Std. Dev. = 14.855
Range = 57
Min = 43
Max = 100
DISTRICT (Population of Congress member's home district in thousands)
Mean = 572
Median = 555
Std. Dev. = 89.11
Range = 567
Min = 413
Max = 980
STATE (Population of Congress member's state)
Mean = 10336
Median = 7730
Std. Dev. = 8194.98
Range = 29306
Min = 454
Max = 29760
In addressing how the data may differ between those Congress members who bounced checks and those who did not, I made the following observations.
|Averages |Service |Age |Votes |District |State |
|Rep/no Checks |7.69 |55.77 |66.49 |595.46 |10,117.12 |
|Rep/Checks |9.48 |50.39 |66.02 |548.29 |8,915.06 |
|Dem/no Checks |9.66 |50.72 |66.11 |584.76 |11,643.24 |
|Dem/Checks |15.27 |64.57 |87.07 |670.88 |12,393.61 |
Full excel file attached.
Average service time for Democrats who bounced checks was significantly longer than Republicans who bounced checks.
Average age for Democrats who bounced checks was significantly higher (14 years) than Republicans who bounced checks.
Democrats who bounced checks had , on average, bigger districts and a larger portion of the votes than Republicans who bounced checks.
Bari Courts
Unit 6 - #2 – Rubbergate – Party Affiliation Test
Relevant Statistics
The sample is made up of 92 Democrats (63%) and 54 Republicans (37%).
Democrats (Value =1 )
AGE
mean = 52.772 years, standard deviation = 10.061,
SERVICE
mean = 11.663, standard deviation=8.444
VOTE
mean = 70.5141, standard deviation=14.643
DISTRICT
mean = 556,652, standard deviation=73,817
STATE
mean 9= ,874,358, standard deviation=8,174,530
All are positively skewed
Republicans (Value =0)
AGE
mean = 54.093years standard deviation = 10.215,
SERVICE
mean = 8.889 standard deviation=7.027
VOTE
mean = 67.5204 standard deviation=15.1605
DISTRICT
mean = 600,740 standard deviation=105,442
STATE
Mean= 11,124,060 standard deviation=8,246,037
All are positively skewed
Null Hypothesis
Democrats and Republicans are equally likely to bounce checks. Party affiliation is an independent variable.
Hypothesis
Democrats and Republicans are not equally likely to bounce checks. Party affiliation is not an independent variable.
I have chosen to use the Chi-Square method to test my hypothesis. The Chi-Square Test will test for significant differences between observed distribution of data and the expected distribution stated in the null hypothesis . Level of significance chosen is .01.
The observed distribution of data is the number of bad checks written in the sample.
The expected distribution of data is the expected a return of an even number of bad checks written between the 2 parties.
Correlation Results:
Pearson Correlation = .131
Sig (2-tailed) = .22
Results of Chi-Squared test:
Chi-Square by: Party = 9.89 Bad checks = 7.918
Df by Party = 1 Bad checks = 1
Asymp. Sig. By Party = .002 Bad checks = .005
From Appendix Table F-3 in the Cooper and Schindler textbook:
Critical Value = 6.64 (Alpha - .01)
Conclusion:
Chi-Square value of 7.918 exceed Critical Value of 6.64.
Since the ChiSquare value of 7.918 is greater than the Critical Value of 6.64, we would fail to reject the hypothesis.
Unit 6 – 2R – comment
April,
I liked the readability of your review. I often struggle with keeping my responses simple and short. Either I write too much or I can't find an easy way to clearly organize my answers. Thanks for the idea.
Bari
Unit 6 - #3 – Rubbergate – Summary
Rubbergate is a name given to the scandal in the Congress involving an abnormally large number of bounced check written by members of Congress. A statistician has been asked to collect and interpret some data about the scandal for a lobbying firm in an attempt to gage how the scandal will affect the upcoming elections. (Cooper & Schindler, 2001,pp. 703-704) Below is a general description of the total number of bad checks written.
A total of 7,322 bad checks were written by only 90 Congress members in 1992. Of those, 6,000 were written by Democrats and 1,322 were written by Republicans. That works out to an average of 81.35 bad check written per Congress member. (Does anyone else find this alarming?) The minimum number of bad checks written was 1, but the maximum was 878. (I hope that person was not re-elected.) The average service time of the bad check writers is 11.77 years in office. The average age for this group is 52.93.
The result of the Chi-Square Test shows that we can not reject the hypothesis that Democrats and Republicans are not equally likely to bounce checks. The Critical Value at alpha = .01 is 6.64, and the results of the Chi-Square test at 7.918 exceed this value.
Reference
Cooper, D. R., & Schindler, P. S. (2000). Business research methods (7th ed.). Boston, MA: McGraw-Hill Irwin.
Unit 6 summary
The lessons from this weeks assignment are multiple and far reaching. There are numerous new statistical terms we used this week including correlation, normality and t-distributions just to name a few. The tests centered around hypothesis testing which can be quite intimidating if not handled in small pieces. It is very important to understand your data before you try and decide what type of test to use. In fact, by understanding your data well, you can begin to migrate to the most appropriate tests for your data set by eliminating test that do not fit the measurements you are looking for.
The most important point from this week is that there are numerous quantitative test that can be used to analyze your data. Many of the test will yield useless results or be statistically insignificant. But there are some quantitative test that will yield powerful insight into you research topic and provide you with statistical evidence for your research hypothesis (or null hypothesis).
Bari Courts
Unit 7-1 Performance Evaluations
Management Dilemma
There have been employee complaints regarding the consistency and fairness of the XYZ Company performance appraisals. Given the use of the appraisals in helping determine merit raises, accurate appraisals are important.
Management Question
Are there any non-work related biases in the evaluation process? If so, can they be eliminated?
Research Question
Are average scores higher or lower when compared to department, grade, or type?
Ho = Average scores do not differ by Department
Ha = Average scores do differ by Department
Ho = Average scores do not differ by Grade
Ha = Average scores do differ by Grade
Ho = Average scores do not differ by Type
Ha = Average scores do differ by Type
Investigative Questions
These are questions that will need to be answered for the manager to answer the research questions.
Why are managers late with evaluations?
Do later evaluations usually mean a lower score?
Do interdepartmental biases exists?
Are scores artificially changed to move employee from one level to another?
Are the 3-month performance appraisals necessary?
Should the 3-month performance appraisals be used in determining merit raises?
Is the company lowering scores of older employees trying to force early retirement?
Measurement Questions
The management questions are what should be asked in order to obtain the information the management needs. However, this case already has statistics that include…
SCORE – Evaluation score in points.
LAG – Difference between the due date of the evaluation by the manager and the day on which the results were discussed with the employee, in days.
EMP@EOY – Years with the company at the end of the year.
POS@EOY – Years in the current position at the end of the year.
TYPE – Type of evaluation (0=3 month evaluation, 1= annual evaluation).
GRADE – Employee's grade scale: 1-7 are clerical or technical; 8-9 are skilled employees; 10-12 are management.
DEPT – Employee's department number.
EDUC – Employee's education level: 12 is high school grad; 13 is one year training beyond high school (e.g., secretarial training); 14 is two years beyond high school (i.e., Associate's degree, nursing diploma); 16 is bachelor's degree; 18 is master's degree; 20 is Ph.D.
EMP@EVAL – Years with the company at the time of evaluation.
POS@EVAL – Years in current position at the time of the evaluation
Statistical Test Results
A full set of descriptive statistics are in the attached output, but I'll make a couple observations. The range on the scores was 144 points (184 – 40). In looking at a scatter plot for scores verses grade and department I see no visual evidence of a straight line indicating a linear relationship. In looking at the histogram for scores, I see it does not conform to a traditional bell curve indicating normal distribution. The data is skewed to the right of center indicating a larger than would be expected group of evaluations had better than average scores.
Skewness for : scores = -0.404 dept = -.894 grade = .315 type = -.028
Kurtosis for : scores = .959 dept = .063 grade = -.318 type = -2.028
ANOVA test results (Alpha = .05, Critical Value (CV) from table F-9 in textbook)
Average scores by Department
Df between groups = 12 F = 1.576 CV = 1.75
Since CV is greater than F, we accept Ho that average scores do not differ by Department
Average scores by Grade
Df between groups = 11 F = 1.178 CV = 1.8
Since CV is greater than F, we accept Ho that average scores do not differ by Grade
Average scores by Department
Df between groups = 1 F = 37.825 CV = 3.9
Since F is greater than CV, we reject Ho that average scores do not differ by Type
My initial thought was to do separate test on the employees who were still at the company verses the employees who had left the company. However, the data set for employees who had left the company is very small (19 records). Thus, I added a column to the data for LEFTFIRM and used 1 = employees who were still at the company and 0 = employees who had left the company.
Ho = Average scores for employees who were still at the company do not differ from average scores for employees who had left the company
Ha = Average scores for employees who were still at the company do differ from average scores for employees who had left the company
ANOVA test results (Alpha = .05, Critical Value (CV) from table F-9 in textbook)
Df between groups = 1 F = 9.189 CV = 3.9
Since F is greater than CV, we reject Ho that average scores for employees who were still at the company do not differ from average scores for employees who had left the company
Bari Courts
Unit 7-1R - comment
Paul,
As usual, you have posted a very good, detailed accounting of the Performance Evaluation case. You seem to be grasping the statistical methods a bit faster than most of us. I'm a bit embarrassed to admit I was a math major as an undergraduate, and I am slow to pick up on the concepts. Fortunately, I found evidence from previous courses that I understood all this stuff some 10 years ago so I guess I'll be able to learn it again. (It is amazing what you forget when you go 10 years between Chi-Squared distribution calculations!)
Bari Courts
Unit 7-2 Statistical Experts System
As previously mentioned, my research topic is on the effect of the internet on the insurance industry. Specifically, looking at the question from the perspective of the agent/company relationship. My research topic is how has the insurance agent benefited form the substantial improvements in communication and paperwork flow due to the internet, and are there still opportunities that are not being taken advantage of?
My data collection will be in the form of a survey. I intend to survey active, licensed insurance agents in the United States. My survey will be online at a web site I create. The agents will be "invited" to answer the survey by going to the web site.
The agents will be marketed through the use of a service called the Agent Card Pack produced by AMC Publishing. The Card Pack is a collection of advertisements the size of a postcard that are purchased by insurance managers to advertise their products to agents. Typically, the managers want the agent to sell their products. I have used their service in the past when trying to recruit agents. The "National" Card Pack is sent 4 times per year to 50,000 active insurance agents.
There are other options in terms of driving traffic to my survey web site. Most notably, I can run advertisements in insurance trade magazines. But this creates issues in determining the size of my sample. Given the narrow specifications of using the Card Pack to gather data, I do not anticipate any validity or reliability issues.
I anticipate most of my survey to contain questions that will yield ordinal data. As a result of the data, I expect to use mainly non-parametric test. After doing some work on the Statistics Coach, the research test I am now considering include the following:
• Gamma
• Kendall's tau b and tau c
• Somer's d
• Spearman's rho
All of theses test are covered briefly in Chapter 18. After reading the summaries, I would argue I will need much more research into these test before I am ready to apply them to my survey results.
Research Problems
There are a few problems I can think of now, and I'm sure many more will appear as I proceed with my study.
1. Will I be able to pinpoint a topic that is researchable and not just my opinion?
2. Can I write a survey well enough to provide me with the information I need to make useful conclusions?
3. Is there a need for personal interviews?
One more item to be addressed is that the survey will be biased in favor of agents who are already on the internet. Due to the nature of my data collection, if an agent is not online, they can not participate in my survey. This is my intention because I need opinions from agents who use the internet, not agents who have never been online.
Bari Courts
Unit 7-2 - comment
Paul,
Your topic is of particular interest to me. I am a licensed insurance agent, and I do some online sales of Tax-Sheltered Annuities to school district employees. Thus, I am very versed in the school districts statistics in Ohio, Indiana, West Virginia, and Kentucky. In addition, I just competed my continuing education requirement for my insurance license online. This is the first year I have had the option of doing my continuing education online, and needless to say, I'll never be sitting through 20 hours of classroom lecture again!
I would think your survey will yield that an online continuing education option for teachers would be of great interest to teachers. I would also think you can get some valuable information from the survey regarding the types of information that would be needed.
Good luck,
Bari Courts
Unit 8-1 Peer viewed article submission guidelines
The Journal of the International Technology Education Association
This journal is dedicated to sharing ideas on how technology can help in the classroom. They ask for practical examples in each article. This journal ask for articles that are double spaced and of no more than 2000 words, but specifically ask for shorter articles if possible. Table, graphs, and artwork are permitted. Photographs are encouraged with a minimum of 300 dpi. Referenced must be properly listed at the end of each article. Five copies of the article must be submitted.
The review process is a bit lengthy. Articles are sent to at least three reviewers and feedback from the review process takes six to eight weeks. Articles accepted for publication will normally appear within one year of the acceptance date. I hope you were not planning on submitting state of the art research with this journal, it could be out of date by the time it is published.
The National School-Age Care Journal
The National School-Age Care Journal is the official journal of the National School-Age Care Alliance. Its mission is to inform and educate school-age care professionals and advocates about research and practice in the field. The Journal is peer reviewed and published semiannually. The focus of the Journal is on current research and effective, innovative practice. They do mention that book reviews and TV programming reviews are acceptable.
The Journal ask for reports to be double spaced and not to exceed 3000 words, including references. They specifically require that the submission adhere to the style outline by APA (4th edition). They also would like a brief abstract submitted of less that 400 words.
I found it interesting that the Journal states it is a peer-reviewed publication, yet there are no instructions on how many peers review your work or any kind of time-table for getting your work published.
The Academy of Management Journal (AMJ)
The AMJ publishes empirical studies relevant to all areas of management. The primary criterion for publication is that the paper makes a substantive contribution to research and theory in management. (sounds like the definition of a dissertation). There are a few rules before submitting an article. The author should attach to an e-mail message two electronic copies of your submission: one with a title page containing author-identifying contact information, one without the author-identifying title page. Each file should contain all figures, tables, and fully double-spaced references. These two files should be named AUTHOR and NO-AUTHOR, respectively. All submissions must be made in a Word 6.0/95 -compatible file format. A regular article can not exceed 40 pages. The abstract can not exceed 75 words.
This is the only Journal that stated that any hypotheses has to be explicitly identified as such.
This Journal also has an entire section dedicated to reviewers. The AMJ stresses timeliness in providing quick feedback to authors. They do not state how long is considered timely. The reviews are "double-blind" reviews, neither the author or reviewer know each others identity. The reviewer are asked to do the following when reviewing an article:
• be specific,
• be constructive,
• identify strengths,
• what is the contribution of the article,
• point out uncertain issues,
• find something nice to say.
Best's Review Author Guidelines
I threw in this magazine to show some differences between journal articles and magazines.
Best's Review magazine is published for insurance professionals. They routinely supplement their magazine with outside-authored original articles discussing specific trends, problems and issues affecting this industry. Their basic criteria for content is: Would this article interest and inform the insurance professional who reads our publication?
Some of the guidelines for submitting an article.
1. Only articles written exclusively for Best's Review will be published. Material that has been sent to other publications will not be considered.
2. Authors should explain technical terms clearly and simply. When in doubt, use a grammar checker and aim for a favorable score on a readability index.
3. Articles should be up to 1,500 words, give or take 250 words.
4. Best's Review is a business magazine, not an academic journal. We avoid footnotes, endnotes or bibliographies. Authors must cite sources (full name, location and, if necessary, a brief description of person or company) within the text. Legal citations should be limited; where they appear, they should consist of case, court and date.
5. Authors will receive five complimentary copies of the magazine in which their article appears. Reprints are available to authors in quantities of 100 or more at cost.
6. Since each issue of Best's Review is copyrighted by the A.M. Best Company, publications wishing to use the article after if has appeared in our magazine must obtain permission to do so.
Bari Courts
Unit 8-2 Ethics
In mentoring a doctoral candidate on ethics I would be sure the candidate understands that ethics is a fairly broad term. I would be sure the following concepts are covered in our discussion.
Ethics include correctly disclosing credit for writings and ideas of other people. Most people are good at giving credit to a person for their writing, but often ideas are used without proper reference.
Respondent confidentiality must also be maintained. This could be a good time to consider getting an informed consent for the participant to make sure nothing is being hidden for the participant. This also can help protect the candidate from any non-disclosure issues. In addition, the candidate must make every effort to keep the privacy of the participants confidential. Privacy laws are very strict in the United States.
A candidate must not misrepresent the results of their research and must avoid deceiving the readers and participants. I was once told in a statistics class that if you beat the data long enough, it will eventually tell you what you want to hear. Refrain from beating the data and try to make a logical, educated interpretation of what the data says. Do not beat the data to conform to what you want the results to be.
Keep in mind that while you are responsible for the ethic surrounding your research, there may be other people involved who have a different definition of ethics that you. For example, a sponsor may find you list of participants a useful marketing tool for in the future. Or perhaps a sponsor is looking for your research to supply support for a project he is backing. It is also possible an assistant is not aware of the ethical issues and thinks it is ok to give out participants names to salespeople. While you can be very diligent about your personal ethics, keep in mind that there are other people involved who could violate your ethical standard.
Bari Courts
Unit 8-1R - comment
Melissa,
I liked the information you found on the HBR. I am a big reader of the HBR. I find it the best written publication being printed at this time. I have often wondered what it took to get published in the HBR. I do not find it surprising that the HBR has such a stringent set of criteria for getting published. I particularly find the case examples included in each article to be of great use in helping me understand the authors points.
My only complaint about the HBR is the subscription price (over $100) for 6 magazines a year. I am a miser and usually read the current edition at the local library.
Your second journal, Business Strategy and the Environment, provides a stark contrast to the HBR. It list mostly superficial requirements like page requirements and file type specifications. In fact, when looking at a distance at the two publications requirements, it appears the HBR is looking at quality of the material presented and Business Strategy and the Environment is more concerned with form of the material.
Bari Courts
Unit 8-2R - comment
Kevin,
You bring up a good point I had not considered in how to handle the ethics and privacy issues in getting my research through cyberspace. I am planning on doing my research online and had planned on controlling my population by only providing the web site address to my targeted population. This presents interesting issues. It is possible someone could submit as many surveys as they want by continuing to hit the return button on the survey. Also, if I am collecting personal information online, someone could copy that information from me without proper security measures.
One idea I have for reducing duplicate entries is to track the name and IP address of each participant. I can run a quick search to see if someone with that name or from that computer (IP address) has already submitted a survey. If they have, I can either reject the survey or tag it as a possible duplicate.
Bari Courts
Unit 8-3 - Research Methodology
I intend on basing my research plan on the management research question hierarchy. I am fairly sure of what my management dilemma and management questions are going to be. I am now in the process of deciding how to develop a survey for research questions.
My data collection will consist of a mostly qualitative survey. I anticipate using questions that use the following:
simple category scale (Do you use the forms download on the Northern Life website?)
multiple choice / multi-response scale (Which of the following licenses do you hold?)
Likert scale (The internet is my main form of communication with the home office? strongly agree – strongly disagree)
multiple rating list scale (Please indicate how beneficial the following services would be on the home office website.)
I anticipate problems along the way with my survey. One issue is how to accurately identify the population size I am surveying because I am going to use the internet for the survey. I also am somewhat concerned that I can generate enough surveys to create meaningful data. In fact, I'm not sure how many surveys are even needed to create meaningful data. And finally, I need to do some work to ensure that my online survey adheres to proper ethical standards including privacy and security issues.
Bari Courts
Unit 8-3R - comment
Melissa,
You are one of the few of us in the course who are using a small survey size and personally doing each survey. Most of us are doing large mailings and hoping to get a good response. While working on my final paper for this course, I am using a dissertation that uses a small survey size. It presents some interesting options. The small survey size allows you to get more detailed information from a participant, especially on their feelings (which is hard to get on a paper survey). The small size also allows for a time lapse between surveys to better judge how answers may change over time. Because I am doing one big mailing and using the response, I am in essence giving a perspective from a moment in time, not over time.
Unfortunately, you run the risk of just a couple bad (biased, incomplete, not truthful) surveys skewing your entire results. In the dissertation I read for my paper author actually had 2 extra participants that he did not use just in case one of the original participants was not able to complete the test.
Bari Courts
Team Final Posting
Likert shortcomings
I have been strongly considering using the Likert scales for my survey (like many other people). I spent some time trying to find the shortcomings of the Likert method in an effort to see if the shortcomings would be too difficult for me to overcome. I ran across an article by Ronald Downey and Craig King (Missing Data in Likert Ratings: A Comparison of Replacement Methods) that pointed out missing data is often a big problem for Likert surveys. They strongly suggest that the best method of handling this problem is to replace the data. I found this surprising. They spend 15 pages describing the best method for replacing the data. The reason replacement is their choice it because they feel that omitting the surveys with missing data reduces the power and accuracy of the analysis. In addition, ignoring the data can lead to an underestimation of the participants score .
This article has given me some new insight and I must not ponder how I will handle omitted data should I follow through on my desire to use a Likert survey.
Bari Courts
................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related download
- research question hierarchy transcript
- vwl statistik prof dr peter schmidt
- 5 motivation in organizations
- pico paper reflection
- access and sampling in qualitative research on trust
- microsoft word apa sample paper with
- july 18 2006 pearce associates
- the management research question hierarchy is a method of
- key steps in doing social science research pursuing a
Related searches
- what is a statement of teaching philosophy
- what is a theme of a story
- research question hierarchy steps
- which is a disadvantage of using credit
- what is a philosophy of education
- research question hierarchy example
- which relation is a function of x
- what is a philosophy of nursing
- standard deviation is a measure of what
- build management research question hierarchy
- what is a statement of interest
- which substance is a product of photosynthesis