Webcast questions and answers:
'Tips for Designing Great Surveys' 16 June 2009
Thank you to everyone who commented and asked questions [some were answered during the webcast].
These responses have been compiled by Dr Stephen Jenkins, MD Snap Surveys, Ken Mihill - Snap Training, Ralph Sutcliffe and Louise Hodrien - Snap SurveyShop.
» Listen to the webcast [open access until 15 July 2009]
| Delegate | ||
| Question | Answer | |
| Julie Allmayer | Why would you use check boxes rather than radio buttons for a single response question? It seems to me that you're just inviting multiple responses. | Question type: You are right that it is important to set each question response type as either single or multiple response as appropriate. In an online survey, single-response questions will, by default, use radio buttons, while multiple-response questions will use square boxes. However, both appear as square boxes in paper surveys (as in our example questions). |
| Dave Burton | Can I have your opinions about whether attitudinal questions should (a) have a "not applicable/don't know" (b) a "neither" or both. Or neither! | Attitudinal questions: If asking respondents to rate satisfaction with various services, a 'Not applicable' option is needed for those who have not used a particular service (unless you have already asked about usage of different services and routed accordingly). If you're using a Strongly agree/Tend to agree/Tend to disagree/Strongly disagree scale, the decision of whether to include a neutral option in the middle is more personal - would you prefer to force respondents to decide one way or another, or do you think a neutral/'Neither agree nor disagree' response is valid? There is no right or wrong answer, but it's a good idea to be consistent to allow comparisons to be made. For expert advice on questionnaire design and wording, please contact the Snap SurveyShop Research Team. |
| Suzanne Cagle Barbara Ford Clint McCaskill Kathleen Patterson |
Interested to learn more about the feature that allows you know how long the survey will take. | Paradata in Snap: This can be found in the Paradata options. This will add a variable called ID.time to your survey which will record duration. |
| Jean Callaghan |
Do you do any student surveys at universities? if so, can you share/advise on how British universities do them? | Student Surveys: If you're planning to conduct the survey yourself but would like some ideas on best practice, we would suggest posting a question on the Snap Forum at http://www.snapsurveys.com/forum so that other Snap users can share their thoughts. If you are interested in outsourcing a student survey, please contact Snap SurveyShop. |
| Stephanie Graf | How many segments (ex per demographics:age/gender/department...) does Snapsurvey allows when analysing the results of the survey ? | Analysis in Snap: Up to 2000 categories in any dimension. |
| Hayley Hargreaves | The pre-email slide indicates SNAP running/ administering survey (not North?Council) - yet it will be me in my organisation so will have access to the survey respondents' details. Could this be misleading? | Respondent confidentiality: The most important thing is to be completely open and honest about who will have access to the data and whether individuals will be identifiable or not. For a staff survey it is best either to employ an external contractor to ensure confidentiality, or if the survey must be run in-house then we would recommend making it completely anonymous (i.e. without unique respondent IDs). {Snap SurveyShop are experienced in running staff surveys - projects@snapsurveys.com] |
Alena Harvey |
May I ask what sort of percentage response rates we should expect from surveys. I assume it varies on how well you know the business? Say from 10,000 businesses what would you expect the percentage response rate to be if they don't all know you? I'd be sending it out hard copy and on-line via e-mail / website. | Response rates: It is impossible to estimate the response rate without more information unfortunately. It depends on many factors such as: the relationship the respondents have with the company (are they regular/occasional/non-customers, will they have heard of the company, etc.), how long the survey is, whether the introduction email or letter is well-written and professional, whether there is an incentive to take part (e.g. a charity donation / summary report of results), whether the subject matter is interesting and relevant, and who the survey is targetted at (a survey of Chief Executives for example would probably get a very low response rate). So in short it is very hard to predict - you could consider running a small scale pilot first to get an idea of response rates. |
Jann Huber Kuldeep Singh |
How do you determine the sampling size--how many to send surveys to? for example out of 10,000. | Sample size: There are various online sample size calculators you can use (just do a search online), which take into account the size of the population you are surveying (e.g. all staff in your company) and the confidence interval you would like to achieve. This will give you an idea of how many survey responses you should aim for, although you will then need an idea of likely response rate in order to determine how many people to send the survey out to. The best indication of this is to look at similar surveys you have run in the past. Another important consideration when determining sample size is any sub-group analysis you would like to carry out afterwards, to ensure you have large enough sample sizes (ideally 100+ per sub-group). And finally, budget is of course a key consideration. |
| Cathy Jensen |
Does randomizing answers on a Liekert scale make sense? | Likert scale: If a Likert scale is used to mean a single rating scale, then it does not make sense. However, if a Likert scale is used in its original sense as a set of rating scale questions, then it does make sense to randomise the sequence of the individual questions. |
| Ramona Johnson | How do you associate the survey with a database? | Databases in Snap: In Snap the database link wizard on the file dropdown menu allows you to link database contents to a survey. In real time, Snap WebHost surveys can use the contents of a database to preseed responses. |
| Helen Mayes | If a paper questionnaire has been designed in Snap and sent out in another country and the results analysed in Excel (and another language) if the Excel file is sent to us in the UK can we import the results into Snap to draw graphs etc? | Importing data into Snap: In general, Excel files can be imported as .csv files into snap for analysis. The key requirements will be that each column is headed by a cell containing the question name ( such as q4 or q4a) and that responses to multichoice and grid questions are either given as code values or as text values which exactly match the corresponding variable's code labels. |
| Rebecca Ortman Graham Russell Angela Withey |
Hi, I've had Snap (individual license) for several years and have never seen the smileys, mapping options, sliders, and other visuals. Are these available to me on the latest build, or are they limited to some other version of Snap? I am constantly updating to the newest build. Thanks very much. | Snap 10: These features will all be available in Snap 10. You will automatically be upgraded when it is released unless you are not on CoverPlan [the Annual Maintenance Contract]- in this case, please contact the Snap Sales Team. If you are not sure whether you are on CoverPlan call UK: 020 7747 8900 or US 603-610-8700 |
| Craig Parylo | Is the use of an unbalanced scale a valid method of collecting responses? | Question scaling: As a general rule, respondents expect a balanced scale for their responses, otherwise the question may be perceived as biased. |
| Jamie Studts | I wanted to inquire about the process of taking multiple surveys and compiling them into a 'master' file. I do not like copying and pasting as this causes the numbering and formatting to get messy. | Merging surveys in Snap: In Snap, the database link wizard enables you to combine the contents of several surveys, either by appending cases, or by creating combined cases based on a respondent id in the various surveys. |
| Alan Thompson | Ranked questions in Snap - how is this done please ? e.g. state 1 for most important down to 6 for least important. | Ranked questions in Snap: Set up the options to be ranked using Open First and Open Next styles, then use the SurveyPlus 'Rating Check' programme to set up the ranking system. (Go to File > Programs > Snap 9 > SurveyPlus > Rating Check). If you need any help, please contact Snap Support. |
If you want any further information please contact us.