One down, two to go

This past Friday I defended (and passed!) my Master's thesis. I now have my Master's degree in Statistics.

Now I just have two other major exams this summer. These exams are my PhD written prelims. They will be over all of my course material for the past two years. I will be spending the next month preparing for the exams.

Thanks for your prayers for my defense. You can be praying for me as I prepare for these next exams.

Stapled

One, two, three, I count to myself as I snatch the pages off the printer. Sixty-three, sixty-four, sixty-five and the printer wines as it finishes. I tremble as I neatly stack the pages and make sure all of the edges line up. My hand shakes as I reach for the stapler. Click. And I feel...PEACE. My thesis is complete! I breathe in deeply, the smell of freshly printed paper filling my nostrils. I exhale slowly as I feel the relief of finishing.

Sorry for my absence in writing over the past two weeks. I have been every present at a computer, but busy writing my thesis.

Over the last few weeks, I have had the following scripture running repeated through my head:

1 Peter 5:6
Humble yourselves, therefore, under God's mighty hand, that he may lift you up in due time. Cast all your anxiety on him because he cares for you.


I am so grateful that I can go to Him with all of the things that makes me anxious and know that He cares about me.

Thank you, Jesus, for your strength and help over the past few weeks.

Thesis: In conclusion…

(Read Sunday – Wednesday’s posts first)

I hope that you have learned a thing or two about quality control, statistics, and my research. I find this to be quite fascinating when I step back and look at the bigger picture. Some days I get so involved in a small aspect of my research that I forget how much fin this is.

Although I am not particularly interested in agriculture and resource conservation, even though it is very important, I am very interested in quality control and survey error. And the things that I am working on have direct application into social statistics, an area in which I have more interest.

For now, this concludes my posts about my research. I hope that you enjoyed reading about what I do and that it didn’t bore you too much.

Thesis: So what do I do?

(Read Sunday – Tuesday’s posts first)

Without getting into the nitty-gritty aspects of my research, here is a brief exposure to my research.

I have studied the in-stream sampling method that was used to identify the quality control sample in the 2005 NRI data. Both the data collection and review are completed for this data. Thus, I am able to look at if the errors that we thought the data collectors were making they were actually making and if the sampling units that we were targeting actually contained errors. I have analyzed several different variables to determine this. The outcome is that the sampling method was doing what we expected and did a good job of identifying sampling units that contained errors.

I have then been looking at ways to improve this sampling method to be more efficient. Also, while the sampling method used in the 2005 data was good, there were also things that we didn’t expect to find and ways in which we can want to change the design. I am working on recommendations for the in-stream sampling design to be used in later NRI data collection.

Thesis: In-stream sampling

(Read Sunday and Monday’s posts first)

In-stream sampling is one sampling method used to effective identify sampling units for the quality control sample.

Things that we must consider with this sampling design are

• which sampling units we wish to target (i.e. which sampling units we think the data collectors might make mistakes on)
• what prior information we have on the sampling units to use for targeting the selection process
• how to design the sample so that the data collect cannot predict when review will occur
• temporal aspects of the sampling rate (usually we wish to oversample at the beginning and ending of the sampling process when the data collectors are more prone to make errors)

I am researching what attributes are important in designing an in-stream sampling method to achieve the desired results.

Thesis: Quality Control

(Read Sunday’s post first)

When we think of quality control, we usually think of its use in industry and manufacturing. This is an important use of quality control. But, the quality of the data that is collect in a survey is also of importance. This is what my research is about.

We are interested in knowing how well the data collectors are doing at accurately collecting the data. This information allows us to create margins of errors and indicate variability. In normal practice, a subset of the data collectors’ sampling units are identified to be reviewed. Then an expert in the topic of the survey looks at the data that has been collected and determines whether it is accurate. In many cases sampling designs are not used for the identification of the sampling units for review.

My research is to look at the sampling design that was implemented in the 2005 NRI data. My goal is to determine if this was a good design and to write the sampling design for the 2007 NRI quality control sample.

Thesis: What is the NRI?

Since I am in the middle of writing my Master's thesis and many people have questions as to what I am actually researching, I have decided to take this next week to introduce my research. My research is on in-stream sampling in quality control and its application to the NRI.

Today is an introduction to the NRI, the data set with which I work.
Monday I will explain quality control at large.
Tuesday I will explain in-stream sampling methods and their importance in quality control.
Wednesday I will explain my analyses on the NRI data.
Thursday I will try to conclude and tie things together.

I hope that this week will be understandable and let you into my world of statistical research.

In conjunction with the Center for Survey Statistics and Methodology (CSSM) at ISU, I work for the United States Department of Agriculture on the National Resources Inventory (NRI) which is conducted by the National Resources Conservation Service (NRCS). The goal of the NRCS is to provide information to the government about the conservation on U.S. private lands. The data in the NRI is a means to accomplish this goal.

The NRI assess the status, condition, and trends in soil, water, and other natural resources on all non-Federal lands. This survey is longitudinal, meaning that it is a continuous inventory over time. The lands which are surveyed are in all 50 States plus Puerto Rico, the US Virgin Island, and some Pacific Basin locations.

The NRI contains data on land cover, land use, soil erosion, prime farmland soils, wetlands, habitat diversity, and conservation practices at more than 800,000 sample sites. As you can imagine, this is a complicated data set with a plethora of information.

The data is collected by people all over the Nation who are trained to view areal photographs of the land and answer a set of questions. I do not actually collect the data, but I am working with this data after it has been coded. So essentially, I work with a data set of numbers on which I perform statistical analyses.

So that, in a nut shell, is the NRI.