1998
DOI: 10.1080/00221309809595542
|View full text |Cite
|
Sign up to set email alerts
|

Missing Data in Likert Ratings: A Comparison of Replacement Methods

Abstract: The effects of using two methods (item mean and person mean) for replacing missing data in Likert scales were studied. The results showed that both methods were good representations of the original data when both the number of respondents with missing data and the number of items missing were 20% or less. As the numbers of missing items and of respondents with missing data increased for the person mean substitution method, a spurious increase in the inter-item correlations (and, therefore, reliability) for the… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
4
1

Citation Types

6
315
0
8

Year Published

2004
2004
2024
2024

Publication Types

Select...
8
1

Relationship

0
9

Authors

Journals

citations
Cited by 510 publications
(340 citation statements)
references
References 7 publications
6
315
0
8
Order By: Relevance
“…Missing data were examined as an initial step in data analysis. This examination was guided by the recommendations set forth by Downey and King (1998), which indicate that data should be retained if the number of participants with missing data is less than 20% of the total sample and if the missing data itself constitutes less than 20% of the total measure. With regard to the TOSCA-A, missing data were observed among 6 participants (i.e., 6.2% of the total sample) and consisted of between 1 and 8 items (i.e., 1.5% to 12% of the total measure).…”
Section: Discussionmentioning
confidence: 99%
See 1 more Smart Citation
“…Missing data were examined as an initial step in data analysis. This examination was guided by the recommendations set forth by Downey and King (1998), which indicate that data should be retained if the number of participants with missing data is less than 20% of the total sample and if the missing data itself constitutes less than 20% of the total measure. With regard to the TOSCA-A, missing data were observed among 6 participants (i.e., 6.2% of the total sample) and consisted of between 1 and 8 items (i.e., 1.5% to 12% of the total measure).…”
Section: Discussionmentioning
confidence: 99%
“…These data were retained because they involved less than 20% of participants and less than 20% of the total measures, therefore adhering to the guidelines indicated by Downey and King. In these cases, item means were used to replace the missing item values because this method is considered a good representation of the original data in the current scenario (i.e., when the number of participants with missing data and the number of items missing are 20% or less), and is preferable to the person mean substitution method because it does not artificially inflate indices of scale reliability (Downey & King, 1998).…”
Section: Discussionmentioning
confidence: 99%
“…The dissociative subtype of PTSD (DS) was used as dichotomous independent variable, and change in scores on both CAPS (pre- and post-treatment) and PSS-SR (at day 1 and post-treatment) as the dependent variable. Person mean imputation was performed when the percentage of missing data on the total number of items of a measure did not exceed 10% (Downey & King, 1998; Hawthorne & Elliott, 2005). If this was not the case, participants were excluded from analyses (Cohen, Cohen, West, & Aiken, 2003).…”
Section: Methodsmentioning
confidence: 99%
“…Totals for all categories may not sum to 506 due to missing data. We used mean substitution for handling the missing data (Downey & King, 1998). Unfortunately, the largest opposition to the new policy among the four groups was the patients, who perceived they have the highest individual influence in terms of the health reform.…”
Section: Discussionmentioning
confidence: 99%