Raw Score

Using Raw Scores in Communication Research

A raw score tells a researcher exactly how a participant rated an item or survey question, regardless of how the participant’s score related to other participants’ scores or how the participant rated other similar items. For example, if a survey question asked a participant to select the extent to which he or she agrees or disagrees with a given statement where 1 = strongly disagree to 7 = strongly agree and the participant selected 5, then 5 would be this participant’s raw score for that item.

Raw scores are commonly used to calculate central tendency (e.g., mean, median, mode) to describe data as well as in more complex statistical analyses to test hypotheses. Statistical equations typically call for raw scores unless otherwise noted. Raw scores on one item or survey question may be averaged across participants in a sample to determine an item-level mean for the sample.

Determining Data Normality

Raw scores are used to create frequency distributions, which can alert researchers to outliers or [Page 1403]errors in the data. A frequency distribution is a table that contains the number of times each value was selected across a group of participants where each participant’s raw score is represented. For example, if a researcher wanted to know how many hours people spend on social media platforms per day, a frequency distribution table would provide easy reference to how many participants in a given sample reported spending five (or any specific number of) hours on social media. An outlier in this scenario might be a person who reported spending 15 hours on social media per day when the rest of the sample ranged from 1–8 hours.

Measuring Latent Constructs

A single raw score is limited in what it can reveal about variables used to describe a sample or population. Raw scores on several survey questions, however, can inform a construct. A latent construct is an underlying theoretical concept that is made up of a series of direct and indirect observations (often raw scores) and captures something an individual survey item cannot. For instance, hope is a multifaceted construct that is made up of three dimensions or parts: self-efficacy, self-esteem, and creative problem solving. It is challenging to capture ratings of hope with just one question so researchers use questions about all three dimensions to understand hope. Raw scores on questions that ask about self-efficacy, self-esteem, and creative problem solving are measured and analyzed together to understand a participant or sample’s average ratings of hope. In this way, raw scores are used to calculate mean or average scores for full scales that represent a construct—a value also called a scale score (i.e., representing the average of raw scores on several items in one scale).

...