OToPS/Item Overlap Methodology
Appearance
< OToPS
BD Content Overlap Example
[edit | edit source]Step | Instruction | |||||||
1: Collect Items | Create a spreadsheet containing all the items for each measure. Each row should correspond to an item. See Step 1 Example at right. | |||||||
At minimum, record the (abbreviated) name of the measure, the item number (the order that the item appears on the scale), and the item stem (i.e., the text of the item). Include any additional information about the items that might be useful (e.g., scoring instructions). | Step 1 Example | |||||||
It will also be helpful to have a column representing the broader symptom category or cluster. This will help later with organizing the items. If the measure already has subscales, you can use that as the category. If it is not clear how to categorize the items, you can leave that column blank at this stage and come back later to assign categories. Consult with a content expert as needed | Measure | Item Number | Item | Category | Scoring Info | |||
Create a copy of the spreadsheet created in step 1 (could be a duplicate tab in the same file). This original will be a record of all the items on all the scales before any of the items are combined. This may be a helpful reference later in the process. | YMRS | 1 | Is your child's mood higher (better) than usual? | Elevated mood | from "absent" to "euphoric; inappropriate laughter; singing" | |||
2: Combine Similar Items Within Each Measure | Sometimes scales contain multiple items measuring essentially the same symptom (e.g., "Yells when angry" and "Screams when angry"). We need to combine these within each scale to avoid inflating the apparent amount of overlap. Go through the items on each scale and decide if two or more of the items need combining. | YMRS | 2 | Does your child's energy level or motor activity appear to be greater than usual? | Increased motor activity-energy | from "absent" to "motor excitement; continuous hyperactivity (cannot be calmed)" | ||
To combine items, you can either 1) choose one of the items to be representative of the symptom; or 2) write a new description of the symptom that captures the content of both items. For example, I might combine "Yells when angry" and "Screams when angry" with the symptom, "Yells/Screams when angry". See examples in Fried et al. (2017) and consult with content experts if not sure. If in doubt, leave them separate. | YMRS | 3 | Is your child showing more than usual interest in sexual matters? | Sexual interest | from "normal; not increased" to "overt sexual acts (toward patients, staff, or interviewer)" | |||
Once you've decided to combine 2 or more items, create a new row with the final symptom description in the "Item" column. List the item numbers for the combined items, separated by commas, in the "Item Number" column. The "Measure" column should contain the name of the measure. If there are additional columns (e.g., scoring info), try to combine that information if possible, or remove that information if necessary. Remove the rows for the items that have been combined so that only the new information remains. See Step 2 Example at right. | KMRS | 9 | Increased self-esteem and appraisal of his/her worth, power, or knowledge (up to grandiose delusions) as compared with usual level. | Grandiosity | ||||
3: Sort by Content | Stack all the rows from all the measures into a single spreadsheet (if not already organized this way). Then, work on re-organizing the rows of the spreadsheet to put similar items next to one another. This will create "clusters" of rows with similar items. You can cut/paste or drag the rows (on Google sheets). If you input category information in step 1, you could start by sorting that column alphabetically (this is particularly useful if you used constent category labels across scales). | KMRS | 10 | When you were (___), were there times that you talked very rapidly or talked on and on and couldn't be stopped?
Did people say you were talking too much? Could people understand you? |
Accelerated, pressured or increased amount of speech | |||
If some items are lacking data in the "Category" column, compare them with other items and use your judgment to decide if they measure similar content based on the wording. Consult with a content expert as needed. | KMRS | 11 | Excessive involvement in dangerous activities without recognizing the high potential for painful consequences. | Poor judgment | ||||
The "Category" field doesn't need to be filled out for every item (unless you are using this information in the final overlap analysis). However, you may find it helpful to add this information for each item to facilitate the clustering process. | ||||||||
Once items/rows are organized into clusters with similar content, put similar clusters next to one another as best as possible. | Step 2 Example -- ? double coding/review by expert in the field | |||||||
This step does not have to be perfect -- just do your best to get similar items next to one another. This will make the next step easier. | Before Combining Items 1 and 2 | Measure | Item Number | Item | Category | |||
4: Construct the Spreadsheet for Coding | Create a new, blank tab in the document. This will be the sheet that contains the data for the content overlap analysis. | IAD | 1 | Yelling | Verbal Aggression | |||
In columns A, B, and C, add column headers: "Sympom Number", "Symptom Label", and "Symptom". These columns will start empty, but you will fill them in as you code the symptoms | IAD | 2 | Screaming | Verbal Aggression | ||||
You can start in row 1, but it actually doesn't matter what row you start in. If you want to include meta-data at the top of your sheet, add these column headers below the meta-data | IAD | 8 | Arguing | Verbal Aggression | ||||
Starting in column D, add a column with the abbreviated name of each measure you're analyzing. These columns will contain the symptom overlap codes. (Note that in the Step 4 Example at right, the measure names start in column E; that difference is not important). | IAD | 5 | Throwing | Object Misuse | ||||
Leave the column immediately to the right of the last measure-name column blank. Then, copy and paste the measure-name columns you just created in the cells to the right of the blank cell. In this second set of columns, you'll code which items on each measure correspond to the symptom in that row. | ||||||||
You should now have a row of column headings starting with "Symptom Number", "Symptom Label", and "Symptom" followed by the names of each measure, a blank cell, and the names of each measure again. | After Combining Items 1 and 2 | Measure | Item Number | Item | Category | |||
The blank cell serves as a visual divider; you can make it narrower and shade it in, as in the Step 4 Example at right. Or you can get rid of it and use gridlines instead. Again, it's just a visual reminder that these two ranges contain separate data despite having the same column headings. | IAD | 1,2 | Yelling/Screaming | Verbal Aggression | ||||
5: Code the items | Returning to the sorted spreadsheet created in Step 3, go through each symptom cluster and identify distinct symptoms within each cluster. This will take some thought and judgment, but having organized the symptoms into clusters will hopefully help. There may be distinct symptoms within each cluster, or it could be the case that all the items measure the same thing. Add a description of that unique symptom to the Symptom column in the new spreadsheet you created in Step 4. | IAD | 8 | Arguing | Verbal Aggression | |||
Using the clustering spreadsheet, code the content overlap among the items across the different measures that represent that symptom. Since the items are already clustered by content, you will not need to compare each item to all the items across all the measures -- just compare it to the other items within its content cluster. | IAD | 5 | Throwing | Object Misuse | ||||
The exact coding schema may differ from project to project. In the Step 4 example, 0 = no overlap, 1 = partial overlap, and 2 = full overlap. | ||||||||
Record which items on each scale correspond to the listed symptom. If there are multiple items on the same scale, list them with commas in the cell. | ||||||||
Repeat this process until all of the items on all of the scales have been coded. | ||||||||
Step 4 Example | ||||||||
In this example, the data have already been entered and there is some additional formatting. However, this illustrates how the columns should be named for recording data about symptom overlap (the first set of measure-name columns) and item mappings (second set). |
Key for Item Overlap Tables
[edit | edit source]Number | Description |
---|---|
0[1] | No, not on scale |
1[1] | Symptom featured generally on a scale (e.g., composite/double-barrelled item) |
2[1] | Symptom specifically on scale |
3 | Symptom present, but fails to specify direction |
4 | Symptom present, but specifies degree (RMOAS example, but Fried would put in #1) |
5 | Direct observation by professional (sign) versus patient report (symptom) |
Item Overlap Subpages
[edit | edit source]Borderline personality disorder scales |
Brief general mental health screeners |
Mania Scales in kids |
Pediatric PTSD Scales |
Catatonia scales |
ADHD scales |
Pediatric depression scales |