Submit a request
Submit a request

Available Fields for Item Analysis Datasets

This page lists the statistical fields that can be calculated and retrieved using Item analysis. Specify one or more of these fields when initialising your report dataset using Data API's SET /reports/datasets endpoint.

These fields can be calculated for every Question in the dataset. Some fields are only available in a specific dataset type.

item_position

The Item's position in the Activity. This will be empty unless the the Item is in the same position in all analyzed sessions.

Only available in activity-analysis-by-question.

item_reference The reference of the Item that contains the Question.
organisation_id

The Item's source organization. This allows to disambiguate Items with the same reference found in multiple Item banks.

question_number The number of the Question within its Item. For example, the 2nd Question on an Item will have Question number 2.
question_reference The reference of the Question.
count_sessions

The number of times this Question has appeared in a session.

count_attempted The number of times this Question has been attempted by a user.
p_value The average relative score of the Question, where relative score is defined as score/max_score. For unattempted Questions, a zero score is assumed. Ranges from 0 to 1.
stddev_p_value Standard deviation of p_value.
p_value_if_attempted The average relative score of the Question, where relative score is defined as score/max_score. Unattempted Questions are not included in the calculation. Ranges from 0 to 1
stddev_p_value_if_attempted Standard deviation of p_value_if_attempted.
discrimination_index

Measures how well a user's overall score correlates with their score on this Question. Ranges from -1 to 1. Typically a discrimination index of less than 0.2 would be viewed as a poor Question and should be reviewed.

Calculated by comparing the p_value of two different populations; the users scoring in the top and bottom 27% on the whole Activity.

We first calculate the top and bottom 27th percentile of the overall Activity scores across the whole population. We then select users for the top group if their overall score is greater than or equal to the top 27th percentile score, using the same logic to select the bottom group.

We then calculate the p_value for this Question from the scores of each group of users. The discrimination index is the p_value of the top group minus the p_valueof the bottom group.

Only available in activity-analysis-by-question.

These summary fields result in a single value that applies to the whole dataset.

count_questions

The total number of Questions in the dataset.

count_questions

The total number of Questions in the dataset.

count_sessions

The total number of sessions matching the filters in the initial dataset request.

count_sessions_analyzed

The total number of sessions that contain exactly the same Questions, and were kept for analysis.

count_sessions_discarded

The total number of sessions that were discarded before analysis because they did not contain the correct set of Questions.

count_sessions_top_27p

The total number of sessions included in the top 27%, for the purposes of calculating discrimination_index.

count_sessions_bottom_27p

The total number of sessions included in the bottom 27%, for the purposes of calculating discrimination_index.

Was this article helpful?
1 out of 1 found this helpful