wlai0611
commented
5 years ago I did an independent t test on the IKSIs of entries with the <=10 letter words. The 2 groups were entries with letter position 1 and entries with letter position > 1.
theDataSmall is the_data after the filtering/word processing but before we used van Selst Jolicoeur outlier elimination.
Welch Two Sample t-test
data: theDataSmall[theDataSmall$let_pos == 1, ]$IKSIs and theDataSmall[theDataSmall$let_pos > 1, ]$IKSIs t = 151.39, df = 238690, p-value < 2.2e-16 alternative hypothesis: true difference in means is not equal to 0 95 percent confidence interval: 64.98991 66.69473 sample estimates: mean of x mean of y 233.3041 167.4618
Reviewer 1 Comment 1 asked for statistical support that the IKSI of the first letter and the middle letter slowing. I tried to show that words are slower in the middle so I made a correlation of abs(letter position - (wordLength/2)) Vs. meanIKSI/wordLength
Pearson's product-moment correlation
data: subject_meansAgg$DistFromMid and subject_meansAgg$mean_IKSI/subject_meansAgg$word_lengths t = -2.5042, df = 21, p-value = 0.02059 alternative hypothesis: true correlation is not equal to 0 95 percent confidence interval: -0.7445628 -0.0839209 sample estimates: cor -0.4795348
See entropyTypingAnalysisComment1.r
I was trying to do a t test with nonmidWords Vs. midWords but was lost on 1) how to correct for inter subject variability like in a repeated measures 2) if I chose a paired t test how to correct for the unequal sample sizes (there are more nonmidWord letters compared to midWord letters)
Do you think the correlation will suffice to show mid WOrd slowing?