CAHSEE data problems, gray hair
28/09/08 09:41 Filed in: CAHSEE | Data
errors
There are few jobs more thankless than
data management and, yet, data management is the necessary first
step to any sort of data analysis, including the descriptive graphs
available at 4point0schools. The CST data for instance has a row of
data for every subgroup in every grade for every CST test. The data
file is mammoth and has required me to buy more RAM for my
computer. The CAHSEE data is far more simple. It contains two tests
(English and Math) and some subtest data. There are far fewer
schools in the data file since, obviously, only high schools
administer the California High School Exit Exam. I allocated one
day to load in the new data to my system and create the 2008 CAHSEE
graphics packages we are giving away this month.
I figured I’d be done by mid-morning and that perhaps I would spend the rest of the day in the hammock or watching MI5 on DVD. (It’s a fantasy of mine that when a certain task is done, I’ll lay in the hammock, but in fact I always end up working on another project.)
On the morning in question, I had already run into a half a dozen problems just a few hours in. Some were typical – the year wasn’t showing up right on the graphs, a title was missing, the current year’s data wasn’t reading properly. But in addition to the usual, there was some serious funkiness that I could not figure out. I sat at my desk six months pregnant having suffered from pregnancy mental health problems in the past thinking, “My God, maybe I know nothing about data.” I emerged from my office on occasion to announce, “This is nuts or maybe I’m nuts.”
Of course in dealing with data, I always assume the problem is mine and it usually is. There are so many ways you can make a mistake. But by the end of the day I walked out of the office and normally would have made the announcement like a peacock (or peahen) but was so beaten and exhausted I simply muttered:
“The problem is with the state data file.”
Once I had identified the problem, I fixed it easily in my data set and moved forward with my graphs since I had already sent postcards to 1,000 high schools in the state offering freebies.
As a good citizen, I sent an email to CAHSEE technical assistance explaining the problem. The only indication I have received that there was a problem is that one week later, the data was pulled for about two days and a new data file was loaded that was 3 megabytes smaller than the original. I’ll not announce what the problem was, but it was consistent with a newer, smaller data set. It was also the type of problem that can happen to anyone. I feel a great deal of sympathy for whomever it is that works with the CAHSEE data even though they did not respond to my email. (This would have been good: “Hi Amanda. No, you’re not crazy, at least not on when it comes to data.&rdquo
All of this is to say, if you get a graph from me and you think there is something “funky” about it, let me know. Not only will I fix it if there is a problem, I will respond to your email and even thank you for the alert. Furthermore, I’ll even give you more free graphs (that hopefully will have no problems).
Data errors are extremely easy to make but with enough checks in place, you can be reasonably sure that your work is accurate. We often end up with mistakes in early drafts of our graphs, but they do have a process that they go through and, to our knowledge, no one has ever received a lemon.
I figured I’d be done by mid-morning and that perhaps I would spend the rest of the day in the hammock or watching MI5 on DVD. (It’s a fantasy of mine that when a certain task is done, I’ll lay in the hammock, but in fact I always end up working on another project.)
On the morning in question, I had already run into a half a dozen problems just a few hours in. Some were typical – the year wasn’t showing up right on the graphs, a title was missing, the current year’s data wasn’t reading properly. But in addition to the usual, there was some serious funkiness that I could not figure out. I sat at my desk six months pregnant having suffered from pregnancy mental health problems in the past thinking, “My God, maybe I know nothing about data.” I emerged from my office on occasion to announce, “This is nuts or maybe I’m nuts.”
Of course in dealing with data, I always assume the problem is mine and it usually is. There are so many ways you can make a mistake. But by the end of the day I walked out of the office and normally would have made the announcement like a peacock (or peahen) but was so beaten and exhausted I simply muttered:
“The problem is with the state data file.”
Once I had identified the problem, I fixed it easily in my data set and moved forward with my graphs since I had already sent postcards to 1,000 high schools in the state offering freebies.
As a good citizen, I sent an email to CAHSEE technical assistance explaining the problem. The only indication I have received that there was a problem is that one week later, the data was pulled for about two days and a new data file was loaded that was 3 megabytes smaller than the original. I’ll not announce what the problem was, but it was consistent with a newer, smaller data set. It was also the type of problem that can happen to anyone. I feel a great deal of sympathy for whomever it is that works with the CAHSEE data even though they did not respond to my email. (This would have been good: “Hi Amanda. No, you’re not crazy, at least not on when it comes to data.&rdquo
All of this is to say, if you get a graph from me and you think there is something “funky” about it, let me know. Not only will I fix it if there is a problem, I will respond to your email and even thank you for the alert. Furthermore, I’ll even give you more free graphs (that hopefully will have no problems).
Data errors are extremely easy to make but with enough checks in place, you can be reasonably sure that your work is accurate. We often end up with mistakes in early drafts of our graphs, but they do have a process that they go through and, to our knowledge, no one has ever received a lemon.
4point0schools.com |