The reliability and validity of a manual coding system for identifying emotion
Emotion focused coping and emotional expression have been identified as potential mediators of long-term adjustment to a cancer diagnosis. Tools for measuring emotional expression in text have not been well-validated, although there is an abundance of textual data that is available to help explore emotional expression and the relationship between emotional expression and outcome measures. The purpose of this dissertation was to develop a manual coding system of emotional expression that could be compared with current programs that have been used to evaluate emotional expression. Inter-rater reliability was found to be good with the use of the manually coded system of emotional expression between both trained and blinded coders. Sensitivity and specificity statistics compared the ability of a currently used text analysis program to predict both presence and absence of emotional expression. While excellent specificity and negative predictive value were found, the program demonstrated relatively good sensitivity and poor positive predictive value. There were not many notable differences found between manual coding and computerized coding in predicting change over time in regards to outcome measures. Replicating these findings on another sample is one area of future direction. Assessing the possibility of supplementing current content analysis programs to include the capability of attending to context is another important area of future direction.