Discuss the difficulty and discrimination of the items and whether or not they would continue to use items on future exams

Basically none of my clinical patients understand what neuropsychology is. In fact, it took my parents several years to finally grasp the concept. The way I attempt to communicate what happens behind the scenes is as follows: We test how your brain is functioning in a number of different areas. We then compare your performance on those tests you just took with other individuals of similar age, sex, and ethnicity. So we are able to understand what your personal strengths and weaknesses are in relation to that comparison peer group. This week is all about understanding the different types of scores and how some basic formulas can be leveraged to give scores meaning. The pertinent terms for this week are raw scores and norm-referenced scores. Raw scores refer to the actual number of correct responses on a given test. So for example, eight correct responses out of 20 potential responses would be a raw score of eight. In some cases, raw scores can be very helpful by themselves. But usually, they’re just a starting point on the road of transformation. Why? Because comparing the time it takes a spunky 18-year-old to run a quarter mile with an 82-year-old World War II veteran, simply isn’t fair. Conversely, comparing the number of fascinating details present in the life story of that adolescent and veteran is also not fair. Normative data is a pool of numbers with a theoretically-driven common theme. This allows for the clinician to transform the raw scores irrespective of test-takers individual differences, into scores that matter to the person sitting in front of them. This is also a situation where being average is both socially and statistically appropriate. I hope you all have that friend who loves to bake. For me, her name is Audrey. Let’s say Audrey goes a bit overboard and brings 100 cookies to a party. There are those party givers who grabbed one-to-two throughout the
evening, some who say they are on a diet and only eat a half, and some who are off their diet during the weekend who eat maybe three-and-a-half to four. Then there are those outliers, that one person who gets an entire cookie but only eats a small crumb. Me, I will eat those cookies until the hosts have to awkwardly try to hint that it’s past their bedtime. So while the sample may only be about 20 people or so, it may still follow a normal curve. What this means is it roughly 68 percent of folks will be in the average range of cookies eaters, leaving roughly 2 percent of fine folks, such as myself on one of those tails of the curve. With this normative sample, a researcher could further examine how factors such as age, sex, aspirational BMI, etc, play a role in cookie consumption. This allows for exactly what we’re after. The ability to have a raw score of 0.25 cookies for a three-year-old girl to produce an equivalent percentile rank to a 31-year-old man who ate four cookies. Finally, the larger and more diverse the normative sample is, the more easily generalizable it becomes. Normative data allows for fair comparison of different raw scores, potentially impacted by demographic variables. We relay that data through different scores, specifically Z, T, and scale scores. Z-scores have a mean of zero and a standard deviation of one. Negative numbers refer to scores below the mean and positive scores above the mean. T-scores have a mean of 50 and the standard deviation of 10. A T score of 35 would be 1.5 standard deviations below the mean. Scale scores have a mean of ten and the standard deviation of three. So a score of 15 would equate to roughly the 97 percentile. This allows for an informed individual to quickly glance at three numbers, negative one, 40, and 7 and with appropriate context, know that they are all saying the same thing. In sum, normative data is where numbers really start to have meaning. Once the basics are well understood, this data allows for a clinician or researcher to quickly scan a page full of numbers on different scales and translate this into a clinical diagnosis amongst other possibilities. [MUSIC]