• Summary
• Downloads
• Explanatory Notes
• Related Information
• Past Releases

PROMPT CARD GUIDELINES

While aspects of these standards will be of interest to those outside the ABS, they were developed for internal use. As such, some information contained in these standards will not be applicable to an external audience. ABS staff should refer to the Corporate Manuals database for the most recent version of these documents, as some details (names, phone numbers etc.) have been removed from the online version.

Contents

• Introduction
• Number and structure of Response Options
• Appropriate presentation of cards
• Layout
• Visual Design
• Typography
• Spacing
• Language
• Summary Checklist

Introduction

These prompt card guidelines have been designed for use in ABS Household surveys. The guidelines aim to ensure ABS prompt cards are designed according to best practice and are consistent across surveys. The guidelines have been compiled from ABS experience, international research and best practice standards used in the Office of National Statistics (UK) and Statistics New Zealand. Note - The visual examples used in these guidelines are adapted from prompt cards used in ABS surveys (except where noted otherwise), and may not reflect current wording.

Number of prompt cards

Prompt cards are used in interviews to visually present response categories to respondents, allowing them to read through the list in their own time. While prompt cards can be very useful in assisting respondents to select the correct response, they should be used only when necessary. Where there is only a small number of simple response categories (e.g. yes/no), a prompt card is unlikely to help the respondent, and is therefore unnecessary. Limiting the number of prompt cards to those essential for helping the respondent reduces the overall number of cards required at each interview. This is more practical for the interviewer and can also reduce the respondent's perceived burden of the interview.

Numbering of prompt cards

It is recommended that each module should be given an alpha-numeric code, and the cards be numbered in ascending order within each module. For example, if the first module is on personal characteristics, then this module will be called A. If the next module is on income, then this module will be B. Cards within each module will be numbered from 1 upwards (i.e. Module A cards will be numbered A1, A2, A3 etc., and Module B cards will be numbered B1, B2, B3 etc.). This is because prompt cards may need to be changed in between the dress rehearsal and the final enumeration, and it is easier to renumber cards within a module than an entire survey set. The code should appear in the top right hand corner of the prompt card.

Number/Order of Response Options

The response options on the prompt card should be comprehensive so that they cover the most commonly provided answers (as identified in pre-testing). However, this needs to be balanced with ensuring that the list is not overly long. Long lists of response options increase the risk of primacy effects (where respondents are more likely to choose the first applicable option from the top of the list, without reading the entire list). Long prompt cards can result in questions having to be repeated as respondents are more likely to forget the question after reading the card. It is recommended that unordered lists should not have more than 9 response options listed at any one time. Where there are more than 9 response options, consideration should be given to using headings to group them by theme (see 'Grouping of categories' below), or (where possible) asking a series of questions with fewer response options. However, this is not always possible due to user requirements such as the use of international standard questions or output categories.

Numbering for consistency with CATI/CAPI

If there are more than 9 response options in a multiple-choice list on a prompt card, the numbering should start at 10 rather than 1, for consistency with Blaise programming standards. This is to avoid ambiguity in the responses entered by the interviewer. For example, the key sequence 113 may be interpreted as 1-13 or 11-3. This confusion can be avoided if all values have the same number of digits.

Rating Scales

Often prompt card response options contain rating scales. Here categories need to be balanced so there are an equal number of positive and negative choices. A neutral response option should also be provided. It is also generally better if negative options are presented first in the list, especially for sensitive questions. This is to show the respondent that it is acceptable to choose a negative option, should they wish to do so. In general, three or five categories is the appropriate number of options for rating scales.

Grouping of categories

To avoid using an excessive number of unordered response options, there are two approaches that can be utilised. The first approach involves arranging the options into groups such as main and subcategories. This is outlined in the diagram below. While grouping related responses can help provide structure to the list, headings used to group lengthy response options by theme can be misleading as some respondents can be led to believe they need to choose one response per thematic group. Instructions must be clear if headings are used (see the 'Instructions' section), and the use of headings to group themed options should be consistent throughout the survey.

The second approach lists the main categories on one prompt card and uses separate cards for the subcategories in each heading. This approach is more suitable when the list of response options is too long to display clearly on one page. It is useful to add an example to the main categories so that respondents know whether the category applies to them.

'Don't know', 'Refusal' and Other Options

"Don't know" and "refusal" options should not be placed on the prompt card, even if they are placed on the pick list for the interviewer. Whether it is appropriate to include an "other" or "other (specify)" category on the prompt card is dependent on the survey itself and the nature of the topic.

Folders, booklets or laminated cards

The method used to display prompt cards will vary according to the length of the enumeration period and the number of prompt cards. For surveys which are longer than 4-5 months, and/or when there are a large number of prompt cards, it is better to keep the prompt cards in a folder with plastic sheets for increased durability. Many interviewers prefer to use a folder as it easier to keep the pages clean. Folders also allow for easier replacement of cards which need updating during enumeration when this is required. Booklets are generally used when the survey has a shorter enumeration period and there are not too many prompt cards.

One advantage of using a booklet is that it avoids the possibility of glare coming from the plastic sheets of the folder. However, folders have an advantage over booklets as bound booklets have the tendency to fall apart. It should also be possible to acquire matt finished plastic for folders.

Laminated cards with a matt finish are an alternative option, to ensure durability of prompt cards that will be re-used in the future. It is not appropriate to use high gloss finish cards, as the glare from the cards can interfere with respondents' ability to read the response options. Due to the costs associated with laminating cards, the method of presentation may not always be possible.

Repeat identical non-consecutive prompt cards

For non-consecutive questions with the same response options, duplicate prompt cards should be used rather than flipping backwards to an earlier prompt card. Flipping backwards gives the respondent the impression that the interview is not progressing. Where consecutive questions require an identical prompt card, a single version is sufficient (i.e. the respondent can stay on one card for two or more consecutive questions).

Duplicate copies of individual prompt cards

Generally, interviewers should have at least two sets of cards in case one set gets lost/damaged. However, the number of duplicate sets is also dependent on the the length the survey is out in the field. The longer a survey is in the field, the greater is the necessity to have duplicate copies, due to general wear and tear. Duplicate sets are also useful in cases where interviewers have to leave a booklet behind for telephone interviews, or if group interviews are being conducted. Two to three sets are generally sufficient, however, this will vary across surveys.

Consistency and Sequence

Two important principles should be followed when designing prompt cards: consistency and sequence. All of the type, graphic and layout conventions outlined in these guidelines should be followed consistently through all the prompt cards. Lack of consistency may lead to confusion and errors.

Portrait versus Landscape

Portrait format is recommended over landscape format. This is because no more than 70 characters per line should be used, as more than 70 characters on one line can be tiring to read. Landscape format allows more than 70 characters to be used, which makes the portrait format a better option.

Full Page Layout

A full page layout creates a spread-out arrangement of options, which is often the most attractive layout. A split-page layout, where the categories appear in two columns with a vertical line in between, is another option if there is a lengthy list of response options. However, this layout can be more difficult to read. It is generally recommended that if space is an issue, the font size is reduced (to no less than size 16 font - see "Type Size"), or categories are split into more than one page.

Use of Colour

The use of different coloured paper to distinguish and identify different prompt cards can enhance efficiency. However, the use of colour to enhance efficiency must be weighed against the cost of printing on coloured paper. For printing in-house, the current cost of an A4 sheet of white paper is 1 cent, while a coloured A4 sheet is 2 cents (this cost excludes the cost of printing text/images onto the paper). Generally, colour can be deemed useful in the following situations:

- The complexity of the survey: for complex surveys with numerous prompt cards, different colours can be used to distinguish modules or topics.

-Concurrent Surveys: It can be useful to use different coloured prompt cards for surveys that are being run concurrently with the same interviewer group.

-Household spokesperson: It has been suggested that the first prompt card which is used for the household spokesperson be printed on different coloured paper.

Caution must be used when choosing the colours to be used, as different colours elicit different reactions in respondents, and this could bias their responses. It is recommended that dark text is used on lightly coloured paper, as white paper can create a glare which can make reading difficult. Off-white or cream coloured paper is preferred, but other light coloured paper can also be considered, as long as contrast and readability are maintained. As noted above, using coloured paper will increase printing costs, so may not always be possible. In these cases, the use of white paper is acceptable.

Colour combinations that should never be used are red and green, and yellow and purple, as the combination impedes legibility.

Use of Graphics

For prompt cards which contain graphics, ensure that the graphics adequately represent the response they require. It can be time consuming and difficult for interviewers to clarify the meanings of images on prompt cards. Accordingly, images should be carefully chosen.

For increased legibility and better distinction between different items, it is recommended that graphics are printed in colour rather than black and white.

Graphics that do not have a functional purpose (i.e. to represent the response option) should not be used. This is because respondents can often confuse the difference between an image that is used as a supplement to enhance the survey, and an image that merely embellishes the survey for aesthetic purposes.

The following two images depict the problems that can arise when using graphical representations on prompt cards. Health surveys often use images of vegetables to clarify the meaning of a 'serve of vegetables.' There is the potential for respondents to read the image literally, and assume that the only vegetables being referred to are the vegetables in the picture.

Diagram 1.4 (From Limes and Lycopene: What actually is 5 serves?)

Images and Illustrations

Images should be placed above, below or to the right of the text, to ensure that the images do not confuse the eye. Text that is placed to the right of the image can make the text difficult to read, as the pattern of reading English is from left to right.

Typeface

ABS forms use one of the following four typefaces: Arial, Verdana, MS Sans Serif or Times New Roman, depending on the mode. However, they should not be mixed together on the same form. This also applies to the use of typeface in prompt cards. For short lines of prose, sans serif fonts (e.g. Arial and Verdana) are generally more readable. It is therefore recommended that Arial or Verdana be used for prompt card text.

Upper Case Text

Upper-case text is difficult to read. It is recommended that text be set in lower case, using capitals for the start of key words and the first letters of sentences. Capital letters can also be used for the letters of appropriate names, titles and abbreviations.

Whole words should not be set in capitals, as the words lose their distinctive shapes when they are capitalised, all becoming converted into rectangles. This makes reading and comprehension more difficult.

Diagram 1.7: Capitalising Letters

It is important to remember that prompt cards are not read at the same distance as one reads a book or newspaper, and as a result font needs to be larger. It is recommended that a minimum of size 16 font be used on prompt cards. This is because most partially sighted people require larger font between sizes 16 and 22. This recommendation will also suit respondents who have less than perfect eyesight or failing long distance eyesight.

• Categories :In certain situations where both main categories and subcategories are used, the main categories should be placed in a larger size, to differentiate the two. The difference should be at least 4 points larger (e.g. subcategories in 16 size font, main categories in 20 size font).
• Reduction: If an answer option has one or two words that do not fit on one line, it is recommended that the font be reduced for all options on that card to fit the option onto one line only (however, ideally the font size should be no smaller than size 16 font). In cases where type size is altered on one prompt card, it is recommended that the type face is changed on cards only where this problem occurs, where there are a small number of cards which have this problem. Otherwise the size should be changed consistently for all the cards.
• Additional information: It is recommended that any information that the respondent does not need to read, such as office information (e.g. the question module and question number), be placed in the bottom corner, in font size 10 and in grey/drop-out colour, so that it does not draw the respondent's attention away from the main text (See Diagram 2.2). Office information should only be included if it will be useful for the interviewers. Information that may be useful to include here includes question numbers, question module identifiers, the survey name (abbreviated) and the year of the survey. Including this information is useful for interviewers in case the cards get out of order or mixed up with prompt cards from another survey, and need to be re-ordered.
• Card Identifiers: Card labels or identifiers (which identify the chronological card number) should be set at 24 font size.

Use emphasis selectively. It is recommended that different typographies are not mixed together for emphasis. For example:

Diagram 1.8: Inappropriate Selective Emphasis

Emphasis of key words should only be used when it is essential for data quality. When extra emphasis is required, bold is preferable to upper case letters (Couper 2008:156). Underlined text may also be used. For example, it is recommended that the no/none response category be underlined on prompt cards, to assist respondents in identifying this option.

Justification and Alignment

Left justified text is recommended for prompt cards. This includes all main and sub-categories and all response options. Left justified text is easier to read and allows the respondent to follow the natural reading path. It also avoids awkward breaks in phrases and sentences that can occur when text is fully justified.

Prompt cards should have margins of about 2.5cm on each side of the page.

Instructions

Instructions should be positioned above the response categories and placed in bold italics type. They should be the same size as the rest of the text.

Instructions should be included in the following situations:

• If multiple response options are allowed, the instruction "You may choose more than one response" should be placed in italics at the top of the prompt card.
• If response options have been grouped into sub-categories but only one response is required, the instruction "Please choose one response only" may be placed in italics at the top of the prompt card.
• If response options have been grouped into sub-categories but one response is required from each sub-category, the instruction "Please choose one response from each category" should be placed in italics at the top of the prompt card.

This is not an exhaustive list of instructions. Any other instructions should be clearly worded and placed in italics at the top of the page. The font size used should be the same as the response options.

Line Spacing

The recommended spacing between response options is 1.5 to 2 lines, if there are no subcategories or subgroups. If there are main categories and subcategories, there should be:

• 2 lines between the main sections
• 1 line between the subcategories within each section
• 1.5 lines between the main category and first subcategory

Diagram 2.0 : Appropriate line spacing

There should be a small amount of space between the response option and its corresponding identifier or number. The two should be clearly linked for the respondent.

Diagram 2.1: Appropriate use of space

Technical Terms

Technical terms, industry-specific definitions and statistical terms should be avoided unless you have evidence that respondents understand what they mean. If you do not have this evidence, explain the term in appropriate language.

CAI Screen Layout

The interviewer prompts that say "Show Prompt Card" should be in blue interviewer instruction text, to ensure that interviewers notice the prompt before the question is read out. This is important when there are numerous interviewer prompts on screen, and the interviewer has the potential to miss the prompt.

Consistency with CAPI screen

Prompt cards are accessed in Blaise instruments via the F9 key function. Prompt cards should preferably match the CAPI version to ensure consistency between what the interviewer and respondent can see. It is also important that the order of options in the card matches the order spoken in the question.

Due to space/formatting restrictions in Blaise, when the list of response categories is large, the Blaise instrument may not have the space to include all of the examples shown on the prompt card. In these cases, it is acceptable for the Blaise version to omit some of the examples provided on the paper version.

Use of headings

• The use of headings to group options should be consistent throughout the survey.
• Instructions must be clear if headings are used to group themed responses.

• Where possible, it is preferable to not have more than 9 response options.
• If there are more than 9 response options, it may be better to group response options into subcategories and use more than one prompt card.

• If there are more than 9 response options in a multiple-choice list on a prompt card, the numbering should start at 10 rather than 1, for consistency with Blaise programming standards.

• It is recommended that each module should be given an alpha-numeric code, and the cards be numbered in ascending order within each module.

• List categories vertically from top to bottom.
• If double columns are used, provide enough space between each column and separate them with a line. Do not use triple columns.

• Use portrait format.

• Ensure enough space is left for binding cards.
• Recommended spacing when there are no subcategories is between 1.5-2 lines.
• When there are main categories and subcategories, it is recommended that 2 lines are left between the main section, 1.5 lines between a main category and the first subcategory within that section, and 1 line between the subcategories within that section.

• Default margins should be set at about 2.5 cm on all sides

• Images should be placed above, below or to the right of text.
• Images that do not have a functional purpose should not be used.
• Ensure that images are unambiguous.
• Use colour for images to make them easier to distinguish.

• Use black text against white, cream or lightly coloured paper.

• Use ABS recommended sans serif font such as Arial or Verdana.
• Use type size 16 to 22.
• Use a larger type size by 4 to 6 points for subcategory headings and instructions.
• Use type size 10 faded text for interviewer and office instructions.
• Use type size 24 bold text for card numbers/identifiers.

• Present the cards in a folder or bound booklet (depending on the length of enumeration).
• Repeat identical non-consecutive prompt cards rather than flipping backwards.
• Use A4 cards for printing.
• Prompt cards should show identical response options to those shown on CAPI screens.
• Interviewers should have at least two sets of cards in case one set gets lost/damaged.

Couper, M.P (2008) Designing Effective Web Surveys, Cambridge University Press, Cambridge.

Endersby, J (1994) "Type 5: Legibility & Readability", Desktop 1994

Schwarz, N, Hippler, H & Noelle-Neumann, E (1992) "A cognitive model of response order effects in survey measurement," in Schwarz, N & Sudman, S (eds) Context Effects in Social and Psychological Research, Springer-Verlag, New York.

Office of National Statistics, UK

Statistics New Zealand

Diagrams

Diagram 1.3 Cancer Institute of NSW (2007) NSW Health: Live Life Well, viewed on 18/2/09

<http://www.gofor2and5.com.au/DataStore/files/pdf/NSW/NSWH2&5_FS_Serve.pdf>

Diagram 1.4 Elliot, K (1997) Limes and Lycopene: What actually is 5 serves? viewed on 18/2/09

<http://www.kathrynelliott.com.au/blog/2007/01/07/what-actually-is-five-serves>

Diagram 1.5: NSW Office of Drug and Alcohol Policy (2006) Standard Drinks Guide, viewed on 18/2/09, <http://www.alcoholinfo.nsw.gov.au/__data/page/1385/FactSheet_AlcoholandConsumption.pdf>