In antwoord op N Hansen
Re: Serious bug: Guest is allowed to fill in questionnaire
deur Mike Churchward -
You're probably right about them not being able to fill them out - it is now fixed in CVS. But what about anonymous questionnaires? Should guests be able to fill them out?
mike
mike
In antwoord op Mike Churchward
Re: Serious bug: Guest is allowed to fill in questionnaire
deur Michael Penney -
But what about anonymous questionnaires? Should guests be able to fill them out?
Yes absolutely, please don't change this. Actually for an anonymous survey, authentication shouldn't even be checked ideally, if you force login as guest then many folks think the survey isn't really anonymous.
Anonymous Q's should remain publically available, we use them all the time here for anonymous publically available surveys.
Yes absolutely, please don't change this. Actually for an anonymous survey, authentication shouldn't even be checked ideally, if you force login as guest then many folks think the survey isn't really anonymous.
Anonymous Q's should remain publically available, we use them all the time here for anonymous publically available surveys.
In antwoord op Michael Penney
Re: Serious bug: Guest is allowed to fill in questionnaire
deur Don Hinkelman -
Anonymous Q's should remain publically available, we use them all the time here for anonymous publically available surveys.
How do you account for someone submitting an anonymous form multiple times? Or is this not a problem. For evaluations and standard surveys, it should not. But I imagine for a controversial issue, it may tempt some partisans to "stuff" the box.
How do you account for someone submitting an anonymous form multiple times? Or is this not a problem. For evaluations and standard surveys, it should not. But I imagine for a controversial issue, it may tempt some partisans to "stuff" the box.
In antwoord op Don Hinkelman
Re: Serious bug: Guest is allowed to fill in questionnaire
deur Michael Penney -
Q has a 'respond once' setting for anonymous, which IIRC gather checks an IP or a cookie (Mike?).
But really of course you can't. But if you force login, then of courseit isn't anonymous. Mostly we rely on the fact that these are prettylong surveys we give and most folks are far too busy to try and stuffthe box-and they aren't on that controversial of subjects that anyone is likely to have a big axe to grind.
This is certainly an issue with student evals, but since Moodle logs 'anon' responses of logged in students, it probably shouldn't be used for that in it's current form. We use it for a number of forms that aren't really surveys, but need emailing and CSV export. We have a complete set of custom reports based on various fields working with Q, at some point we'll try to release them (hardcoded table names now
).
But the main thing is folks can take them from an url without having to login, which makes the mod very useful for quickly making public forms and surveys that don't require login (in our experience, if you require a login for a survey your response rate goes down the drain). So we'll have to hack that out if it gets put in (can you comment it with something easy to find if you 'fix' this Mike?).
In general, I think anonymous surveys about controversial subjects are probably a very hard thing to do right. For instance we've built surveys for clients where the authentication was completely separate from the results, however the survey got a very low response rate and in informally checking with some of the subjects we found--quite reasonably--that they didn't believe that the results actually were separate, after all the system knew who they were before they took the survey.
Cookie setting and IP checking certainly reduce the chances of ballot stuffing, but I think it's a pretty tough thing to get right overall.
But really of course you can't. But if you force login, then of courseit isn't anonymous. Mostly we rely on the fact that these are prettylong surveys we give and most folks are far too busy to try and stuffthe box-and they aren't on that controversial of subjects that anyone is likely to have a big axe to grind.
This is certainly an issue with student evals, but since Moodle logs 'anon' responses of logged in students, it probably shouldn't be used for that in it's current form. We use it for a number of forms that aren't really surveys, but need emailing and CSV export. We have a complete set of custom reports based on various fields working with Q, at some point we'll try to release them (hardcoded table names now
But the main thing is folks can take them from an url without having to login, which makes the mod very useful for quickly making public forms and surveys that don't require login (in our experience, if you require a login for a survey your response rate goes down the drain). So we'll have to hack that out if it gets put in (can you comment it with something easy to find if you 'fix' this Mike?).
In general, I think anonymous surveys about controversial subjects are probably a very hard thing to do right. For instance we've built surveys for clients where the authentication was completely separate from the results, however the survey got a very low response rate and in informally checking with some of the subjects we found--quite reasonably--that they didn't believe that the results actually were separate, after all the system knew who they were before they took the survey.
Cookie setting and IP checking certainly reduce the chances of ballot stuffing, but I think it's a pretty tough thing to get right overall.
In antwoord op Michael Penney
Re: Serious bug: Guest is allowed to fill in questionnaire
deur Don Hinkelman -
This looks good-enough to be reliable in most circumstances--use long forms, IP-checking and avoid dangerous topics. I agree response rates are the most important thing--and anonymous surveys will allow that.
In antwoord op Michael Penney
Re: Serious bug: Guest is allowed to fill in questionnaire
deur Lars Jensen -
Hi,
If my understanding of the way the Q works is correct, then the only way you can get information about who filled out an anonymous Q (even with moodle authenticaton) is by comparing timestamps in the moodle logs with timestamps on the CSV-export spreadsheet. And this is only circumstantial information - not hard information.
There is no information in the CSV-export about ip-address, username, name, etc. if the survey is anonymous. So an anonymous survey is basically just that, even if users have to authenticate.
Suggestion: One could still improve anonymity a bit by not logging questionnaire activity in the moodle logs for anonymous questionnaires.
Lars.
If my understanding of the way the Q works is correct, then the only way you can get information about who filled out an anonymous Q (even with moodle authenticaton) is by comparing timestamps in the moodle logs with timestamps on the CSV-export spreadsheet. And this is only circumstantial information - not hard information.
There is no information in the CSV-export about ip-address, username, name, etc. if the survey is anonymous. So an anonymous survey is basically just that, even if users have to authenticate.
Suggestion: One could still improve anonymity a bit by not logging questionnaire activity in the moodle logs for anonymous questionnaires.
Lars.
In antwoord op Lars Jensen
Re: Serious bug: Guest is allowed to fill in questionnaire
deur Mike Churchward -
Okay, just to clarify some things.
As for guest responses, they are logged with the guest userid. For this reason, single completion questionnaires with guest responses wouldn't be very useful. Only the first guest login would be allowed.
So, for now, I will implement it this way:
anonymous questionnaires will allow guest access if they allow multiple submissions and are not restricted to either teachers or students (i.e. everyone can complete them).
This change is now in CVS.
mike
- 'Anonymous' questionnaires are anonymous to phpESP only. The data that is retrieved using the reporting system won't indicate anything about the Moodle user who responded. The Moodle tables that track usage will still have data about which users responded.
- All questionnaire responses are recorded in the'questionnaire_attempts' table. These include user id. Multiple vs. single attempts are enforced using this table.
As for guest responses, they are logged with the guest userid. For this reason, single completion questionnaires with guest responses wouldn't be very useful. Only the first guest login would be allowed.
So, for now, I will implement it this way:
anonymous questionnaires will allow guest access if they allow multiple submissions and are not restricted to either teachers or students (i.e. everyone can complete them).
This change is now in CVS.
mike
In antwoord op Mike Churchward
Re: Serious bug: Guest is allowed to fill in questionnaire
deur Lars Jensen -
Hi Mike,
All questionnaire responses are recorded in the'questionnaire_attempts'table. These include user id. Multiple vs. single attempts are enforcedusing this table.
This info is off limits to anyone but the moodle admin, so the only way a moodle instructor can get info about usernames is by comparing timestamps in moodle logs and on the csv spreadsheet. Would it be possible to eliminates the logs for anonymous surveys?
Now, we could enforce even more anonymity by not tracking the userid with the responses, if this is important.
Good idea for anonymous surveys!
Lars.
All questionnaire responses are recorded in the'questionnaire_attempts'table. These include user id. Multiple vs. single attempts are enforcedusing this table.
This info is off limits to anyone but the moodle admin, so the only way a moodle instructor can get info about usernames is by comparing timestamps in moodle logs and on the csv spreadsheet. Would it be possible to eliminates the logs for anonymous surveys?
Now, we could enforce even more anonymity by not tracking the userid with the responses, if this is important.
Good idea for anonymous surveys!
Lars.
In antwoord op Don Hinkelman
Re: Serious bug: Guest is allowed to fill in questionnaire
deur N Hansen -
I think there are five potential scenarios:
1-By name (must be a registered course/site user), allowed to answer only once
2-By name (must be a registered course/site user), allowed to answer more than once
3-Anonymous (must be a registered course/site user, but teacher can't see name), allowed to answer only once
4-Anonymous (must be registered course/site user, but teacher can't see name), allowed to answer multiple times
5-Anonymous (doesn't require login), by default they could answer multiple times, there really is no way around it
1-By name (must be a registered course/site user), allowed to answer only once
2-By name (must be a registered course/site user), allowed to answer more than once
3-Anonymous (must be a registered course/site user, but teacher can't see name), allowed to answer only once
4-Anonymous (must be registered course/site user, but teacher can't see name), allowed to answer multiple times
5-Anonymous (doesn't require login), by default they could answer multiple times, there really is no way around it
In antwoord op Mike Churchward
Re: Serious bug: Guest is allowed to fill in questionnaire
deur Mark Stevens -
Please keep the ability to give an anonymous public survey 
We use them a lot. (For Lars: public surveys do give a timestamp and an ip address in CSV... which can be more or less useful