IntelliSurvey's proprietary data-cleansing system CheatSweep™ helps provide users with more accurate data for their surveys. Using patent-pending algorithms, CheatSweep™ looks at multiple variables to score data quality for all respondents, removing respondents from the dataset who appear to be providing fraudulent, inaccurate, or inattentive information. Enabling CheatSweep™ allows respondents who receive scores above a specified percentage or value to be marked as Possible Fraud, or "F," with their data no longer counted towards overall results.
For additional information on CheatSweep, click here. If you cannot open that link, click here to download the Word version.
Note: In order to access the CheatSweep Report applet, the survey must contain enable cheatsweep in the published version of the survey. Without that code, the applet is hidden from users.
Users must have a survey role of Survey Owner, Manager, or Editor to view the CheatSweep Report, or be assigned the add-on role of CheatSweep Viewer. Only Survey Owners, Managers, and Editors can assign other users the role of CheatSweep Viewer.
To access the CheatSweep Reports applet, click the CheatSweep button from the Reports button group on the Survey Navigation menu.
The report displays three fields - cs_good, cs_bad, and csp, described below. These fields are displayed as separate Topline tiles, and by default, appear as bar graphs.
To ensure that the latest results are being displayed, click the Refresh button in the upper right-hand corner of the header.
Fields
cs_good
A multi-select, closed-end variable, cs_good tracks a variety of factors that positively impact, or lower, the overall CheatSweep score.
cs_bad
A multi-select, closed-end variable, cs_bad tracks factors that have a negative impact, resulting in a higher CheatSweep score.
csp
Lastly, the csp (CheatSweep Probability) field contains the CheatSweep score. This score is the percent chance that a given record is a cheater. The tile reports these scores in aggregate.
CheatSweep run history
The CheatSweep Reports applet defaults to displaying the survey's CheatSweep field tiles, but users may also view the survey's CheatSweep run history by clicking View > View Run History. The CheatSweep Run History table provides a log of each time CheatSweep was run on the survey, displayed in reverse-chronological order. Icons are available above the table that allow users to page forward and backward or refresh the data to ensure the most recent CheatSweep run is shown.
The following information is included in the table. Note that each column offers a tooltip describing the column that may be accessed via the info icon .
| Column | Description |
| Run Timestamp | The date and time at which a CheatSweep scoring was performed. This column defaults to displaying the scoring history in reverse chronological order. |
| Total Count | The total number of records that were considered Completes (status "C"). |
| Score Count | The total number of records scored during each sweep. |
| Sweep Count | The number of records re-statused from "C" (Completes) to "F" (Fraudulent) during each sweep. |
| Unsweep Count | The number of records whose statuses are unchanged in a resweep. |
| Run Time(s) | The duration of the CheatSweep run in seconds. |
| Threshold | The score over which respondents are re-statused. |
| Remove Percent | The percent of all records to be re-statused, based on the highest csp score. |
| Group By | Indicates which variable scoring groups are based off of. If no variable was used to create scoring groups, this cell will be blank. |
| Group By Options | Indicates which option was used to score the groups on a given CheatSweep run. |
To return to the tile view, click View > View Tiles.
Rescoring
CheatSweep will, under some circumstances, need to be manually instructed to rerun scoring for some or all survey records. Examples of such a circumstance are:
- An Expected Country was incorrectly set.
- The attention check programming contained an error.
- A threshold value is changed.
To force a rescore, select Rescore from the Actions menu.
When selected, a dialog box appears asking for confirmation. Rescoring can cause respondents to move from status "F" (possible fraud) to "C" (complete), and vice versa. Rescores generally take only a few minutes to complete.
Creating Excel reports
Like other report types in IntelliSurvey, CheatSweep report results can be downloaded as Excel files. To download an Excel file of the results, select Excel from the Actions menu. A notification modal will let users know that the file is being prepared, with a second notification letting the user know when the Excel file has been created. Click Download to download the Excel file.
Each field will have its own page in the Excel file, displaying both a table of the results and a graphic representation of the data as well.
Editing tile viewing options
Users may edit the default viewing options for the fields shown one of two ways - either via the Gear menu (for individual tiles) or the Actions > Bulk Edit option (for all tiles).
To edit a single tile's display, click the Gear menu. From here, users may customize details such as the chart's title, labels, colors, or toggle between displaying the results as a bar chart (default) or pie chart.
For more information on the Gear menu's options, see Closed-end tile display options.
To change the current display settings for all CheatSweep Report tiles, use the Bulk Edit option. This will bring up the Configure All Tiles modal. Clicking the checkbox to the left of each setting will activate that option's selection field.
The following configuration options are available in the modal.
| Option | Values |
|---|---|
|
Display As (applies to closed-end tiles only) |
Options include:
|
|
Show Label (applies to closed-end tiles only) |
Options include:
|
| Sort By |
Options include:
|
| Colors | Modifies the colors used to render charts. For more information, see Topline colors. |
| Keep options and apply to all future tiles | If selected, these settings will be preserved for the duration of the user's session. |
Comments
0 comments
Please sign in to leave a comment.