Reverse Scoring

A common attitude scaling format, Likert-type scaling, presents a statement and asks respondents to agree or disagree, and scores range from, for example, 1 to 5. Sometimes the same group of statements on a single measure is stated in different “directions.” That is, sometimes a 5 indicates a [Page 1433]high level of endorsement of a particular attitude, whereas on other items, a 5 means a low level of endorsement of that attitude. Before responses can be combined into a single meaningful total score, all items must be in the same direction. To accomplish this, the scores for those items that are in an opposite direction are “reversed.” High scores become low scores and low scores become high scores.

Scores are reversed in a straightforward manner that depends on the range of possible scores for the selected items. Using the common 1, 2, 3, 4, 5 format, where 1 = strongly disagree and 5 = strongly agree, one would reverse score in this way:

Reverse scoring is necessary when research instrument developers have purposefully written a group of items with some items in a different direction than others. A mix of directions in attitude (or any self-report) statements is sometimes designed to break a mental response set in the respondent or force increased concentration when responding. Reversing some questions is also thought to reduce acquiescence and boredom of respondents. However, respondents may misinterpret the test statements when the wording is reversed. It is believed that this may occur due to awkward phrasing on items that are written in reverse, such as “I don’t often read the funny papers.” Some populations with concentration difficulties, such as the elderly and children diagnosed with attention-deficit/hyperactivity disorder, have been studied with evidence of the failure to understand or attend to reverse scored items. It has also been suggested that reversed items may actually measure a construct different than the one intended by a researcher, and factor analyses frequently load items on different factors when they are worded in reverse. Consequently, additional analysis is recommended to indicate the parity of reversed items to other items on the test.

Reverse scored items on assessment scales used in personality theory measurement and clinical symptomology have found some compromise in internal consistency for reverse scored items among older adults. Translating an instrument from one language to another also raises concern, as the linguistic similarity between 2 items written in opposite directions may break down when in another language or cultural context.

The psychometric characteristics of items when reversed have also been studied. Internal consistency tends to be relatively equal when comparing reversed and original items, though it is not always the case and should be examined in each given study. Item response theory analyses have been conducted on selected personality measures to confirm the unidimensional nature of a group of items, whether reversed or not. Researchers have concluded that the different samplings of participants examined by item response theory analysis reflected differing response styles in terms of both the multidimensional nature of the questionnaire and responses to reverse scoring. This item response theory work and other studies have sometimes found that self-reported traits are more complex than reverse scoring ratings may be able to capture.

...