Sunday, March 29, 2009

New research shows Sports News is written at the 8th grade level!

This weekend a prospective customer had asked me about the grade levels of writing in newspapers. I didn't have an answer and after reviewing some Internet sources on the subject was still uncertain that anyone really knows the answer. I had seen people indicate everything from the 5th grade to the 12th grade depending on the source. So we did some investigation on our own.

For Measurement, I used the Flesch Readability Ease and Flesch Kincaid Grade Level. I found out that Microsoft Word has a algorithm in Spelling and Grammar that calculates the Flesch Kincaid Grade Level. Be warned this algorithm is seriously flawed. It is capped at 12th grade and does not provide indication of college level writing ability. I didn't use Microsoft's Flesch Kincaid Grade Level for this measure as the data distributions were not normal. I wound up using another software technique for the Flesch-Kincaid Grade Level scores. By the way I think Flesch Kincaid is the method the US Department of Defense uses in assessing grade level so this is a fairly well known method of determining readability of writing.

I devised a test plan using the concept of Randomized Complete Block Designs or RCBD. I looked at the following factors:

Newspaper Brand - Chicago Tribune, Tampa Tribune, NY Times and Baton Rouge Advocate.
Newspaper Section - Front Page, Metro, Sports or Business
Writer - AP Wire Story or Local Writer

There were three hypothesis to be checked.

1) Does the mean grade level vary across the newspaper brands?
2) Does the mean grade level vary within the newspaper sections?
3) Does the mean grade level vary if it is written for AP or Local?

The experiment was conducted examining 48 "online" news articles and measuring grade level as the response variable. This was enough data for a screening experiment and to have a 90% probability of detecting an effect if one was present.

Finding One
What we find is that the average newspaper article grade level is 10.62 or between the 10th and 11th grade. The minimum grade level was 6.45 and the maximum was 14.95. I could say that with 95% confidence the mean writing levels for the newspaper surveyed were between 10th and 11th grade. I could not detect any difference in the mean grade levels between the groups of newspapers. The writing between Baton Rouge, Chicago, NY and Tampa was consistent.

Finding Two
With regard to variances within papers there was a significant finding. Sports articles are written at a lower grade average than business, front page and metro section type of articles.
The average sports article is written for a 8.63 grade level. I could say with 95% confidence the mean grade level sports related articles falls between the 7th and 9th grades. The minimum grade level for a sports article was 6.45. The maximum grade level observed was 10.89.

On the flip side of this issue the mean grade level for the Business, Metro and Front Page articles was determined to be 11.28. If we ignore the lower level of writing in the sports sections, most newspaper articles are written at about the 11th grade level on average.

This begs the question; why are sports articles written at a level approximately 3 years less than the main section news articles?

Finding Three
There was no observable difference in grade levels if a story was written by a local writer or by AP.

In summary, this research was at best a screening test to preliminary answer some questions about the quality of newspaper writing. I was surprised to see the difference in sports articles and cannot really explain why sports writers would write at a lower grade level. Perhaps it is the language of sports. Perhaps it is simpler word, sentence and paragraph structure due to the nature of the games and issues. It would be interesting to do more investigations like this in the future.

Wednesday, March 18, 2009

Optimizing Your Processes with Virtual DOE

Most businesses want improvement in the operation. No doubt managers issue orders to “change this” or “tweak” that. All managers have the good intention of changing things for the better. Most of the time these efforts fall short. Usually it is only a small change with a small result. It takes courage to experiment on a real business process. The risks could be substantial if a manager fails. The tension always exists between the need to improve and the need to maintain stability in an organization. Stability wins out quite often. The breakthrough improvement never results.

Fortunately, there is a solution to this dilemma - Virtual Design of Experiments (Virtual DOE). How does Virtual DOE work? The concept is simple. You do not have to experiment on your “live” processes. A process model is developed using simulation software. The model contains all the steps and constraints of your actual business model. The model is validated and tested against the data from the actual process. Once validated the fun can begin. A focus group of process users is assembled to brainstorm how to improve the model. What changes will improve the current state model? The process users are involved in the proposed changes to the process.

Imagine if your system is controlled by 9 main factors that you can switch between high and low. What is the combination of these factors will deliver the lowest cost or most sales? What are the chances that you have happened onto the best combination by chance? To run every case there are 512 possible combinations. This is difficult in a real process, but relatively easy for a computer simulation. Using techniques of fractional factorial designs the proposed process changes can be distilled to the factors that really have an impact on your bottom line.

Using the results of the Virtual DOE trials, a clear winning solution will emerge. Using action planning and task management, steps can be taken to implement recommendations to change the process. Measurements established and controls can be placed on the process to maintain the improvement. Virtual DOE can be applied to processes of all kinds. It can be applied to complex simulations of chemical processes or complex simulations of discrete event models such as distribution networks, call centers or sales order fulfillment.

ImprovementGuru Consulting, LLC is your expert with Virtual DOE. We will assist with modeling your process. We create and validate the simulation. We will organize and conduct your brainstorming focus group. Run the experiments and produce a final report with recommendations for improvement. We don’t stop there. We assist with implementation. If needed procedures and controls will be placed on your processes to maintain your improvement. We offer a 100% guarantee of satisfaction. If we don’t make improvement, we don’t accept payment.

Wednesday, March 11, 2009

What's wrong with LinkedIn Polls?

Social networking sites such as LinkedIn are offering users to publish single question polls to people within their personal social network. The problem with this approach to surveying is that results will almost always be skewed or biased. The bias can be attributed to the question framing or respondent audience.

As an example a recent question asked, "When you shopped for groceries, what kind of bag do you use? a) paper, b) plastic, c) reusable and d) don't care."

It is easy to see that results from this type of survey may be biased for the following reasons.

1) Response Rate - You really have no control over who answers this question. How many were asked and what is the response rate. Those interested in green initiatives may be more likely to answer a question like this than those who do not care.

2) Demographics of the Target Group - The professional nature of the target audience biases the results. LinkedIn is primarily a network of professionals with a certain demographic. My NetGen kids use Facebook or MySpace and not LinkedIn. Professionals may be more likely to sympathize with green initiatives than non-professionals or those with low incomes. I've been to Wal-mart and although they offer some reusable bags for groceries the bulk of the groceries are packed in plastic. There seems to be no limit on the number of bags a customer can take home.

3) Confusion about choices. Aren't paper and plastic a reusable bag of some type. Plastic bags can be reused just as easily as some other type of reusable bag. Same goes for paper. They may not last as long but they certainly can be reuseable.

Social networking polls can be fun, however, don't use these polls for seeking market research or other important information to run your business. These polls are likely to give a biased answer with a large margin of error.

Box.Net Widget - Great Website Application

I am extremely impressed with the service from Box.Net and the widget they provide that allows a user to provide downloadable content to any website. The widget is available at http://www.box.net/widget. Once you sign up for the account and enter the data the widget generates a html script that you can add to any web page. It took me about 15 minutes to add this widget to three different websites. Once the widget is in users of your website can download materials that you want them to have access to. You can also upload additional material if needed. It is a great tool. I have a working example on my blog, my LinkedIn profile and my website http://improvementguru.com

Thursday, March 05, 2009

Fixing Broken Surveys

I came to the realization today that about 90% of my survey clients engage me after the fact. What this means is that they have completed the survey, have the data and do not have the ability to analyze and interpret the meaning of the data that has been collected. In some cases the survey or design was flawed from the beginning and conclusions about a research hypothesis simply cannot be made appropriately.

I've had some success breaking surveys into two groups and then providing analysis. In other cases we have had to augment a survey with further sampling in order to get meaningful results.

This fall I worked with a PhD candidate on her dissertation research. We were able to salvage the work that had been done, but the thinking up front could have been better.

If at all possible get someone to review a research design prior to execution. Even if you do not get someone to review it BETA test the instrument and the analysis to make sure you can analyze the results successfully and that it is meaningful toward your hypothesis. BETA testing sounds like something simple enough but it is almost always overlooked.

Do your surveys meet standards?

Be wary of internet polls and surveys conducted through popular online media such as facebook, myspace or other sites. Quality research requires that the researcher follow guidelines and subscribe to the appropriate code of ethics and standards. Three organizations publish standards for research that cover ethics.

1) CASRO - Council of American Survey Research Organizations at http://www.casro.org

and

2) American Association of Public Opinion Researchers at http://www.aapor.org

and lastly,

3) ASA or the American Statistical Association at http://www.amstat.org

As for ASA I know they have a complete section devoted to survey research methods. CASRO and AAPOR have codes relating to privacy, internet research and disclosure.

Researchers should disclose the methods used to conduct a survey. This way peer groups and other interested parties can judge the validity of the research.

Tuesday, March 03, 2009

Making the most of Surveys!

Surveys are a good tool to obtain information about your product, brand or research questions. There is a surge in survey research for businesses as new technologies and new initiatives such as green emerge or become revitalized. What is the best strategy for a survey? How should you approach it?

1) Planning and Design- I always use a 1/3 rule for my projects. About 33% of the time is spent up front in planning and design. Defining the research questions, clarifying objectives and setting up a model so that it can be analyzed using practical means. Probably the worst thing you can do is ask questions that give data in a format that cannot be analyzed easily. My students always tend to ask open ended questions? This is great for focus groups or one on one interviews but just does not work for surveys where large amounts of data are required.

2) Quantify the sample size and sample strategy. Who will be sampled and how many? Also try to figure out in advance how you will handle missing data or survey drop-outs. It happens to most surveys that respondents do drop out. There are ways to plan for and mitigate this problem.

3) Beta Test....NEVER deply a survey without first conducting a beta test to find where the flaws and problems might occur. After the bugs are worked out then go full scale with implementation. Most people want to jump to the solution and overlook this keey element.

4) The next 33% of the activity is the actual survey and collection of data. If you did parts 1-3 correctly you should have little or no problems here. This is often the fun part as your data comes streaming in.

5) Analysis and recommendations......Compile a good report which communicates the findings of the survey to management. Don't overload technical details or statistical methods. Upper management is not as concerned about this as they are the actual results and your interpretations.

There is a good book, The Total Survey Error Approach, by Herbert F Weisberg that really helps as a reference for conducting surveys. I recommend it to anyone who will be doing serious survey work.

Pilots for Patients

I wanted to talk this morning about Pilots for Patients. This is a group of volunteer pilots who in early January 2008 flew there first mission. The groups mission is to provide transportation to distant medical facilities for patients that do not have the means or ability to seek specialized treatments. What's special about Pilots for Patients is that it is dedicated to providing help to patients located specifically in Louisiana. We have access to hundreds of small airports throughout the state that makes transportation easy even for those in rural or outlying areas of the state. In one year the group since its founding the group has flown over 200 missions. It's a great organization and if you want to find out more visit there website at www.pilotsforpatients.org

Sunday, March 01, 2009

Thinking Green

It's been a couple of years since my last blog and I'm going to commit to make this a better routine as the days progress. There has been a great deal of talk lately about Green Initiatives and Green Products. My son got me started on it. The concept of doing no harm to the environment and health is back in the spotlight again, much as it was in the 70s.

There have been some good products I have come across lately in my discussions that I would like to talk about.

1) www.act2greensmart.com is producing laptop bags and backpacks from recycled PET bottles. This is a great product and I will probably buy one the next chance I get.

2) Jathropa jet fuel. You can google it but apparently they have fueled one of the engines from a 747 with a mixture of Jet A and Jathropa biodiesel.

3) A contact recently mentioned development in India of a product called NuAlgi which is a nanotechnology for causing algae blooms in lakes or waterways. The product contains elements and minerals necessary to spur growth of diatom algae. These algae then convert CO2 to Oxygen and breath life back into the lake or waterway. The other application I see is for the farming of algae for biodiesel production down the road.

I don't know if any of these except the laptop bags are really commercially viable. I am keeping my eye open for new Green products and initiatives. One thing I came across that I did not think was "Greenworthy" was something called ECO-FRIENDLY BOTTLED WATER from this website http://www.earthfriendlypr.com. I wasn't buying it! I just don't think plastic bottles are friendly to the environment. That brings us back to GreenSmart. I commend the folks at act2greensmart for recycling.