Raw Scores

Raw scores are simply the total sum of a respondent's selections before any statistical conversions are made. While serving as rudimentary level of statistical measurement, raw scores are functionally limited. In application, raw scores are used to arrive at a set of standardized scores (i.e., T scores, z scores) that can be used to compare individuals to a reference group. Researchers use raw scores to perform statistical analyses or to norm measures. Applied practitioners use raw scores to communicate performance or measurement results.

Although not universally true, raw scores typically are the sum of correct responses out of the total possible correct responses. For example, on a scale containing 10 questions, a respondent may correctly answer 8. Therefore the respondent would achieve a raw score of 8. Of course, this raw score converts rather easily to the percentage correct on the scale, in this case 80%. In this example, 80% is a representation of the respondent's raw score; however, only the number 8 is considered the actual raw score.

Educational measurement is a common application of raw score usage. Take for instance a student who achieves a raw score of 17 on an assessment. This score provides limited information without an indicator of the total possible score. If the total score was 20, one would have a better indicator of the student's performance, yet this assessment provides no information about the student's performance relative to peers (i.e., normative) or to the student's past performance (i.e., growth modeling). However, if these other scores are part of the student's ongoing classroom assessment, one would have enough information to tally the student's current performance level. Even if this particular assessment weighs more or less heavily than other assessments with a total of only 20 possible points, these scores provide enough information to allow us to create a total of points that go into the student's overall performance file.

Applications of raw scores do not always require a measure with correct or incorrect responses. Measures of interest, personality, or motivation, for example, do not contain right or wrong answers, but responses that reflect the participant's response preferences. In this example, these inventories measure a number of domains, usually through a Likert-type scale; scores on these domains are summed but do not represent a ratio of correct to incorrect responses.

Although in most cases raw scores provide enough information for typical educational purposes, they remain limited in measurement applications. Consider for a moment that we want to examine how this student performed in comparison to peers. In a class with a small number of students, it might be tempting to simply order the performance of peers from lowest to highest and perform a count of the scores until we arrive at our student's score. However, if the scores were taken from a class of five and those additional four scores were 19, 16, 14, and 17, simply ordering those scores and then counting until we arrive at our student's score becomes inadequate. If one is interested in the student's score in comparison with peers, one would consider the group's highest and lowest score on this assessment and remove the full range of possible scores from consideration. Through the use of a selected statistical metric, we are able to convert the student's raw score into a standard score and determine how it compares to the scores of peers.

...