old code and beginnings

… and by old, I mean really old. Like year 2008 old. Like the first R code I ever wrote.

I recently went digging in an old dropbox account for photos and to my surprise I had saved a copy of my undergraduate thesis, complete with code.

My undergraduate degree was in Psychology and I had a keen interest in psychometrics. My thesis involved an application of item response theory to ordinal data.

if you are interested, I used Samejima’s Graded Response Model to analyze responses made to the Barratt Impulsiveness Scale from a sample of first year university students

At the time, my only experience with statistical software was with SPSS. I remember teaching myself how to use it only a year prior in order to double check my intro stats homework assignments 🤓🔬📏. I think that was a good move in retrospect because I was also able to spin that experience into a paid Stats Advisory position at the Psychology Resource Center.

I digress.

The type of model I needed to fit for my thesis was only available through proprietary software. My thesis advisor recommended I look into R as he had caught wind of a package that could help. The upside was that it would be 🆓 and open source 🤑. The downside was that I was going to have to learn “syntax” 🤢. I was aware that SPSS had syntax facilities for advanced users, but it was always intimidating 😣.

But now I had to learn “syntax”. I’m laughing while typing this believe you me. 🤣 Anyway, I don’t quite recall the learning process, but I reckon it was difficult and in the end I was probably just relieved to get the job done. ✅🎓.I ended up having to write “syntax” for other supporting analyses in my thesis, namely Confirmatory Factor Analysis for assumption checking in LISREL. shudders

For historical (and sentimental reasons) I’ve included the R code in my thesis below. It marked the first steps on the path I’m on today.

From what I can tell I was using R 2.8.1 and I’m happy to see the original 📦, ltm is still active on CRAN today.

Till next time, 🍻✌

install.packages("ltm", dependencies = TRUE)
#downloads and installs the 'LTM' packageto the user’s hard drive.

library(ltm)
#loads the 'LTM' package from the user’s hard drive

attention = read.table(file.choose(),header=T)
#prompts for location of data set

grm0 = grm(attention, constrained = TRUE, Hessian = TRUE)
#fits the constrained GRM to the specified data set. The HESSIAN = TRUE argument computes standard error values.

summary(grm0)
#returns the parameter estimates

grm1 = grm(attention, constrained = FALSE, Hessian = TRUE)
#fits the unconstrained GRM and computes standard error values

summary(grm1)
#returns the parameter estimates

anova(grm0,grm1)
#performs a likelihood ratio test on the two models.

plot(grm1, type = "IIC", lwd = 2, cex = 1.2, legend = TRUE, cx = "topleft",xlab = "Latent
Trait", cex.main = 1.5, cex.lab = 1.3, cex.axis = 1.1)
#plots the IIFs

plot(grm1, type = "IIC", items = 0, lwd = 2, xlab = "Attention",
     cex.main = 1.5, cex.lab = 1.3, cex.axis = 1.1)
#plots the TIF

plot(grm1, lwd = 2, cex = 1.2, legend = TRUE, cx = "left",
     xlab = "Attention", cex.main = 1.5, cex.lab = 1.3, cex.ax
     is = 1.1)
#plots the CRCs

plot(grm1, type = "OCCu", lwd = 2, cex = 1.2, legend = TRUE, cx = "topleft",
     xlab = "Attention", cex.main = 1.5, cex.lab = 1.3, cex.axis = 1.1)
#plots the OCCs