lab report for class

This is a lab report for class

EDF 5400 — Lab 4 — The ?2-test
Read this first.

Note: I try to fix the numbers on the Excel import to correspond to data corresponding to the latest available year. What numbers you need to use depends on which year’s spreadsheet you are using (and whether you are using the full data or only Region 1). The letters don’t need to change from what is in SPSS, but the first number should correspond to the row that has the data names, and the last row should correspond to the last row of data (i.e., it should be the same as what is in the box above).

Data sources :
http://schoolgrades.fldoe.org/reports/index.asp

We will use the “School Grades, Basic information on all schools” data set from this web page.
If you are using a student version of Excel (which limits the number of data records you can use), you may use a subset of the data (as long as you document how you created the subset).
Alternate data file (named : SGbasic_2014-Region1)

This is an excerpt from the full data file available at the FDOE web site containing only Alachua through Charlotte County schools. The excerpt should run in the student version of SPSS (the full version requires the full SPSS available in the LRC and Library computer labs). You may use either version (or if you prefer to work with a different subset of the data, you may do that as well). Make sure your conclusions are appropriate for the data file you are using.
1. The Research Question

Charter schools offer a way to experiment with educational policies within the public school system. They are independent schools within the public school district in which district students can be enrolled by their parents. Although the founders of the charter school must present the school district or state with an educational plan when receiving their charter, they have considerable freedom for how they can set their educational policies. Policies that prove effective in charter schools can then be adopted by the regular non-charter school system.

Proponents of charter schools say that they allow parents to find a school whose educational policy matches the learning needs of their child. This is especially good for children who don’t learn well in traditional classrooms. The proponents claim charters should result in better learning outcomes.

Opponents of charter schools say that they put the students at risk by exposing them to untested educational policies, and inexperienced teachers and administrators. In particular, critics accuse the charters of trying to replace experienced union teachers with inexperienced non-union teachers, and sidestepping union policies designed to protect both students and teachers. Because of this lack of experience, the opponents claim charters should result in worse learning outcomes.

Running counter to both of these analyses is the issue of selection bias. Opponents of the schools note that charter schools serve a parent-selected population (and parental involvement is known to be an important input to educational success) and are free to kick out students who do not meet the schools behavioral or academic standards. This effect would make the non-charter schools look worse than their charter components. Proponents counter that many charters aim to deliberately serve populations that are not well served by the current system, and therefore the selection bias should run in the opposite direction.

This is a complex political issue that is likely to be debated for many years. To help evaluate charter schools, the State of Florida’s Department of Education (FDOE) publishes some data that bear on the subject. Each year the FDOE assigns a letter grade to all schools, both charter and non-charter, based on a number of factors related to students’ performance and improvement in performance. The exact criteria are complex, but we can ask a straightforward question: According to the State’s criteria (and assigned grades) can we see a difference in performance (as measured by the state-assigned grade) of the charter and non-charter schools?

In this lab you will study the relationship between two variables, Charter, a categorical variable which indicates whether a given school is a charter school or not, and School Grade, a letter grade assigned to a school for a selected academic year.
2. The data

The state publishes the annual report on school grades at http://schoolgrades.fldoe.org/ . You can download a number of reports from that location, as well as find information about how the accountability reports are compiled. The data you want can be found by following the link http://schoolgrades.fldoe.org/reports/index.asp On the site, click on the file named School Grades, Basic Information for Schools for 2014 academic year. This should download an Excel® spreadsheet containing the data. Note that the original spreadsheet has Columns A through E locked in place and Column E containing the school name is rather wide. You may need to play with the size of the window or the width of the columns to be able to see all of the columns.

[The complete data file is too big to fit into the student version of SPSS that comes with your textbook. You have two choices: (1) you can do your analyses in the LRC, library or the virtual computing lab, using the full version of SPSS, or (2) you can use an extract of the complete data. I have prepared a subset containing just Region 1 schools (which contains Leon County) on Blackboard. Other subsets are reasonable and you can do the subset extraction in your favorite spreadsheet program. Whatever method you use, you need to be clear about (a) what you did, and (b) how it affects the generalizability of your conclusions.]

SPSS® will read the data directly from the Excel format, but we should take a closer look at the data before reading it into SPSS. Open up in whatever program you usually use to read spreadsheets. You should see some notes about the data at the top of the file. Note that the data start on line 8. In order to read the data into SPSS, you need to get rid of the first 7 lines.

The easy way to do this is to simply delete the first 7 lines in your spreadsheet program and resave it in Excel format. The alternative way to do this is to tell SPSS to start reading the data at line 8.

For either method start by selecting File > Open > Data… and then change the file type in the open dialog to Excel (.xls) files. Select your data file to open it.

Next you will be presented with a dialog that asks you where in the spreadsheet the data are located. If you edited the notes out of the top of the file, the defaults are fine and you can just press “OK”. If not, then you need to adjust the range. The box marked “Worksheet” has a dropdown that says “SG: [A1:AH3279]”. That means it will take columns A through AH and rows 1 through 3279. We want to skip the first 7 rows so type “A8:AH3279” in the box marked range and then press OK. [These numbers change from year to year, and I don’t always get them correct when I edit the instruction. If it doesn’t look right, you may need to fiddle with the numbers. Note that you cannot have the file open in Excel and read the data in SPSS at the same time on a Windows computer.]

The data should now be imported. You might want to check the data view and variables view to make sure that everything was correctly imported. You should see in the second column of the variables view that almost every variable is of a “String” and not “Numeric” type.

Some of the SPSS commands you want to use only work with variables that are coded as numbers. To go from string coding to numeric coding you will need to recode the variable. To do this, select Transform > Automatic Recode …. In the Automatic Recode dialog box, select the variable you want to recode (e.g., “Charter”) and hit the arrow. It should now show you the name followed by an arrow and a bunch of question marks, (e.g., “Charter ->???????”). Type a new name for the variable in the box marked new name, (e.g., “Charter1”), and then hit the button marked “Add New Name”. Now hit OK. Repeat that as often as you need to.

3. Unidimensional Analysis

Before starting the analysis of the relationships between the variables, you probably want to look at some one-dimensional summaries. The most useful command here is Analyze > Descriptive Statistics > Frequencies …. You may also want to look at some graphical summaries. The most useful are probably Graphs > Legacy Dialogs > Bar… and Graphs > Legacy Dialogs > Pie….

Which unidimensional graphs will best help you describe the data? What tables might be useful? Remember that SPSS tables may need to be reformatted for your report.

4. Finding Association between Two Variables

Graphically, there are two approaches to showing the association between two categorical variables. The first is to simply make pie charts of one variable, paneled by the other variable. This should show changes in portion of the one variable. Another way is to work with bar charts. SPSS offers two varieties of bar chart that are useful. One is a clustered bar chart, where each combination of the two categories is given a bar, but the bars are grouped to make comparisons across variables easier. The second is the stacked bar chart, where the bar is split according the second variable. It is worth spending some time playing around with these different graphical styles. Often each one will be better for emphasizing different relationships. Pick the one that best tells your story.

The Analyze > Descriptive Statistics > Crosstabs … command provides a tabular view of the relationship. Click on the “Statistics…” button and then “Chi-squared” to get a chi-squared test for the independence of the two variables. Click on the “Cells…” button to add row, column and/or total percentages to your table. Tables with all of those percentages in them are particularly difficult to read, so you almost certainly will want to reformat them for your report.

Which crosstabs and graphs will best help you uncover the relationship between Charter status and School Grade? Which graphs best show what is happening? What statistical tests are most useful? You should look at both the most recent (2014) data as well as last year’s data (2013).
5. The Pending Schools

You have probably noticed by now that a large number of schools have received a grade of “Pending” or “I” or the grade may be simply left blank. [Depending on what time of year we do the lab, we may or may not see the “Pending” results.] This means that they have not turned in one or more of the pieces of data that are needed to calculate their grade. Does the Pending status affect the relationship between grade and charter status? Use Data > Select Cases… to remove the schools with “Pending” status. Now rerun some of your analyses. Does this make a big difference in any of the conclusions?

6. The Lab Report

Pretend that you are analyzing these results and submitting them as a paper to a journal. You will need to follow general APA style. Your report should have the following sections (you can use different names, but the contents should be there).

1. Introduction — A brief statement of the general problem you are solving and why somebody might care. [I’m doing this because it’s a homework assignment doesn’t count; be creative here.] 2. Research Question/Hypothesis — A brief statement of what question you are trying to answer and what you expect to find according to your theory. This is the scientific hypothesis and not the null hypothesis. Note that this lab has three research questions: corresponding to Sections 3, 4 and 5.
3. Data — A brief description of the data. What was the population and how were the data sampled. [This should not be a click-by-click description of what you did in SPSS, but rather make it sound like a real scientific study.] Don’t forget to cite your source!
4. Analysis — Some descriptive statistics about the data followed by the results of the formal hypothesis test. You should include a graph and table to highlight what you observed. You should include a test statistic, p-value and effect size. APA style contains recommended method for reporting the results of common tests. Check the APA style guide for the labs on Blackboard.
5. Conclusions — Describe what the statistical results mean in terms of the scientific problem posed at the beginning of the report. A significant difference is not your conclusion, but rather evidence that supports your conclusion. Speculate about whether or not you can generalize from the conclusions.

This lab is a little bit different, as we will not do the normal feedback revision cycle. The version you turn in will be the one your final score is based on, so proofread you work carefully. By now you know the kinds of things we check for in the grading.
7. Pointers on Style

Don’t use too many digits in your tables or text. SPSS will cheerfully compute means to 6 decimal places even if the original data only has one or two. Generally speaking, a mean should have one more decimal place than the original data and a standard deviation should have one more decimal place than the mean.

Finally, go back to lab one and the style guide and review the pointers on making graphs and naming variables. All of these rules will still apply.

Jiwon (Alice) Nam, a TA for 5400 in Fall of 2010, put together a brief guide to APA style. Take a look at this. You don’t need to follow APA style strictly, but you must have all of the five elements described in Section 8 and they must be easy for the reader to find to get full style points.

8. Frequently Asked Questions
1. How does the grading on this lab differ from other labs?
We will not do the usual feedback round on this lab, the version you turn in is the final version. Therefore ask yourself, does your lab have the following elements?
a. Graphical and/or tabular summaries of the distributions of the major variables.
b. Crosstabs and chi-squared test for comparing charter schools and 2014 school grades
c. Crosstabs and chi-squared test for comparing charter schools and 2014 grades without Pending grades
d. Crosstabs and chi-squared test for comparing charter schools and 2013 grades
e. Graphical displays to support b & d.
f. All of the usual style points.

2. What if I have a different opinion about charter schools than the person doing the grading?
This is a fair question, as we expect that you have diverse opinions on the subject and we would be quite surprised if all of you agreed with us on this subject. (Personally, we have a mixed opinion of charters: some are good and some are bad. We are skeptical about their value, but are willing to consider what the data show!) You will not be graded on what your final conclusion is, but rather how well your conclusion is supported by the evidence. Unfortunately, Statistics rarely gives us a final and definitive answer to policy questions, often because we have not run the ideal experiment that would provide that answer. At best it provides evidence that supports or refutes a given theory, adding substance to the debate but not settling it.

3. What is the scoring rubric for this lab?
As you don’t get the normal feedback cycle with this lab, it is reasonable to release the scoring rubric in advance. Here it is:
Style
Context 2 1.5 1 0.5 0
Text No problem with text. Minor isolated problems with text. Minor organizational issues: Document is mostly organized clearly but with
a few problems (e.g., introducing new results in conclusions) or
missing supporting pieces (e.g., references).
Problems with unclear wording. Note that this is not just poor grammar or
usage, but wording that is confusing to the point of not understanding
what the student is trying to convey.
Major organizational issues (cumulative with minor). Document is missing
supporting elements (e.g., introduction, conclusions, problem
statement) or parts are so jumbled you can’t tell what is what.
Figures & Tables No problems with figures and tables. Isolated problems with one or two figures. Problems with tables and figures: unclear or missing captions for tables and figures, using SPSS variable names in place of human readable labels. Too many digits in tables.
Mismatch between text and figures Including lots of tables or figures that are unnecessary, redundant, or not well grounded in text.
Correct Answers
Requirement 1 0.5 0
Marginal Summaries Summaries of the marginal distributions of the major variables, 2014 grades, 2013 grades and charter status. Summaries of some but not all of the variables. No unidimensional summaries: either tabular or graphical.
X2 2014 Correct chi-squared test for independence between 2014 grades and charter status.
Attempted chi-squared test but missing a required detail: X2, d.f., p or interpretation
2014 grades by charter test not performed.

X2 2014 Pending Correct chi-squared test for independence between 2014 grades and charter status without pending schools. Attempted chi-squared test but missing a required detail: X2, d.f., p or interpretation
2014 grades by charter test not performed without pending grades removed.
X2 2013 Correct chi-squared test for independence between 2013 grades and charter status. Attempted chi-squared test but missing a required detail: X2, d.f., p or interpretation
2013 grades by charter test not performed.

Graphics Pie-chart or bar plots presented for both 2014 and 2013 data (okay not done both with and without pending). Graphs to support only one of the tests (or just the two 2014 tests and no 2013 graph) No Graphs

Cross Tabs Cross tabs presented for all three tests.
Cross tabs present for some but not all tests. No cross tabs.

Note: Rubric in Turn-it-in may still use older year values. If I missed editing this in any spots, I mean for you to look at both the most recent year’s data (with and without pending) and last year’s data.
FOR YOUR ASSIGNMENTS TO BE DONE AT A CHEAPER PRICE PLACE THIS ORDER OR A SIMILAR ORDER WITH US NOW