Item Total Correlation
- Measures how well a single item relates to the overall score formed by summing all items.
- High values indicate the item aligns with the construct the set measures; low values suggest the item may be uninformative.
- Used to detect and remove or revise items that reduce a measure’s reliability or validity.
Definition
Section titled “Definition”Item-total correlation is a statistical measure that describes the relationship between individual items (or variables) and the sum of all items in a set. This measure is often used in psychological research to assess the internal consistency and reliability of survey or test items.
Explanation
Section titled “Explanation”Item-total correlation quantifies how strongly an individual item is associated with the overall score derived from all items. A high item-total correlation indicates the item is a good indicator of the construct the instrument is intended to measure. A low item-total correlation indicates the item is less related to the overall construct and may not provide useful information. Practitioners use these correlations to evaluate which items contribute to, or detract from, a measure’s internal consistency and reliability.
Examples
Section titled “Examples”Political attitudes survey
Section titled “Political attitudes survey”A survey measuring attitudes toward political issues might include items on taxes, immigration, and gun control. The item-total correlation for each question (for example, the taxes question) indicates how well that individual item relates to the overall survey score. A high item-total correlation for the taxes question suggests it is a good measure of attitudes toward taxes; a low item-total correlation for the gun control question suggests it may not be a good measure of attitudes toward gun control.
School test
Section titled “School test”A teacher’s test measuring student knowledge may include multiple-choice, true/false, and short-answer questions. The item-total correlation for each question type indicates how well that item relates to the overall test score. If multiple-choice questions show high item-total correlations, they are likely good measures of student knowledge; if short-answer questions show low item-total correlations, those items may not be providing useful information and could be removed or revised.
Use cases
Section titled “Use cases”- Identifying survey or test items that do not contribute useful information to the overall measure.
- Informing decisions to remove or revise items to improve a measure’s reliability and validity.
- Assessing internal consistency of instruments in psychological and educational research.
Notes or pitfalls
Section titled “Notes or pitfalls”- A low item-total correlation for an individual item suggests it may not be a good measure of the intended construct and could be a candidate for removal or revision.
- If the overall score for a measure shows low item-total correlations, this may indicate the measure is not reliable or valid; further investigation of contributing items is warranted.
Related terms
Section titled “Related terms”- Internal consistency
- Reliability
- Validity
- Survey item
- Test item