October 2015 survey Full record

Residents' survey 2015

Full survey record

Purpose

This survey was devised by the five volunteer working groups who are compiling the planning options to present to the village.

Its purpose is to find out the areas and issues the Neighbourhood Plan should consider in more depth.

It will lead to options being presented to the village in 2016.

Publicity and invitations

The survey was open to all residents of the Parish of Harvington.

The following methods were used to publicise and invite participation:

Five large (6^ft) banners were placed on the verges of the roads by which the village is accessed. They were present throughout October,
A large (12^ft) banner was placed on the Village Hall, inviting people to come there to complete the survey,
A4 posters were placed in the village notice-boards and other 'usual places' around the village,
An announcement was made in the Village News, with key dates added to the 'Village Calendar'
Every house was give a 'pink slip' invitation, containing a Survey Access Code.

A sample invitation can be seen to contain:

The dates of the survey,
The web site,
What to do if the Internet cannot be used,
An eMail address for help and information,
A telephone help line.
The Survey Access Code - unique to each household.

The invitations were printed on stiff pink A5 paper, so that - hopefully - they would not be discarded as junk mail.

Confidentiality

eMail addresses and house names / numbers will not be made public.

People's individual option selections in the survey will also remain confidential.

Where people made written comments in the text boxes in the survey their name may be published along with their comment.

Publication of results

The analysed results of the survey have published on this web site.

The two most useful reports are:

A short overview of the results (8 pages, PDF),
The full results are presented as bar graphs (23 pages) and residents' textual contributions (34 pages).

The availability of these reports has been made public:

An eMail has been sent out to people who contributed to the survey,
The 18 people who completed the survey in the Village Hall will be given paper copies of the short overview,
The two Harvington-related Facebook pages will be given the link to these results,
Villagers in general will be notified via the Village News.

The results are available in a machine-readable format ( Turtle-RDF ) to enable independent analysis.
This 3.2Mb file contains both the definitions of the questions asked and the (anonymous) responses.

Embedded comments provide some help with understanding the schema used.

It is envisaged that anyone wishing to undertake their own independent analysis of the results might find a framework such as Sesame quite useful.

Survey Definition

The survey was only presented on-line in an interactive format, in which irrelevant questions were omitted.
There was never a paper version of the survey which could have been used.

For record purposes a listing of the survey questions has been produced.
This consists of:

The questions people were asked,
A record of the type of answer required for each question:
Passive samples of the interactive maps which people were shown,
The help text associated with each question (in brown boxes),
A textual representation of the logic used to decide which questions to omit - depending on their previous responses (in red boxes).

The answer types are:

SINGLE: The user selects just one option from the presented list.
MULTIPLE: The user may select any number of options.
PRIORITY: The user was invited to arrange any number of the options in priority order.
MAP: A map with coloured areas was presented, as well as a textual description of the corresponding areas. The user was invited to select any number of the areas in priority order.
NUMBER: The user was invited to enter a number.
TEXT: The user was invited to make free-format textual comments.

RANDOM indicates that the options were presented in random order (see positional bias below).
MANDATORY indicates that the user had to make a selection before proceeding.

Technical commentary

Avoidance of positional bias

When people are presented with long lists there is a tendency to place greater attention on the first few items.

Where questions involve selection or ranking of options in which values or judgements (rather than facts) are involved this can result in a bias towards those items near the top of the list.

To avoid this bias our survey randomised the order in which non-factual options are presented to survey users - everyone will see them in a different order.

Analysis of priorities

Within the survey there are questions and maps in which people were invited to arrange their preferences in priority order.

We analysed these questions and maps using the Condorcet voting system.

In brief, this ranked options by conducting all possible pair-wise elections and then ranking options by how many 'elections' they won against the other options - by how many 'votes' each option received.

How is personal information protected?

All the data from this survey is captured and held on a Linux server in a UK data center and then down-loaded to an analysis system in Harvington.

The registration data objects (name, address, eMail, etc.) are individually encrypted using the Threefish cryptographic algorithm.

Where passwords are required the original password is not stored or encrypted in any way. A cryptographic digest is formed using the Whirlpool cryptographic hash function, and then encrypted using Threefish.

How are eMail addresses validated?

When an eMail address is to be validated a message is sent with a link on which people are to click. This link records the address being validated, a time stamp and a brief salted Whirlpool-based digest.

When links are presented for validation the time-stamp is checked to ensure the link has not timed out and the digest is correct.

Measures are taken to protect against brute-force attempts to forge links by repeated guess-work.

Survey integrity

Great care was taken to ensure the integrity of the survey:

We first required people to validate an eMail address,
The IP address of the client system was recorded (which allowed us to get some idea of the geographical location of the user), as was the time-stamp of the successful validation,
People were then required to present a Survey Access Code. The codes were distributed by hand to people's houses
Where individuals lost or had never received codes a replacement was delivered by hand to their house.

There were a few attempts to 'hack' the survey - none successful:

A few people from elsewhere in the UK started but did not complete the eMail validation process,
A Beijing agency of the Chinese government made several requests - over successive days - to validate an address, but did no follow through.

There were no failed attempts to forge a Security Access Code.

Only one person contacted us to say that he was unable to complete the survey; this person attempted to register - using an obsolete browser - just 18 minutes before midnight on the closing day!

What was the Survey Access Code?

Every household in Harvington was given a unique Survey Access Code. This code had to be presented before the survey could be completed.

We don't record which code went to which house, we only know which of the 19 Village News 'walks' the code was delivered to.

We checked how many times each Survey Access Code was presented. More than would be expected from a normal Harvington household would have been be investigated - there were no such cases.

We have three time-stamped data which can be correlated in this investigation: the Survey Access Code, the validated eMail address and the IP address of the machine being used.

The structure of the Survey Access Code is:

A single letter indicating which of 19 'walks' in the village this was delivered to,
Two letters - different for each house in the walk,
Two letters from a salted Whirlpool digest of the first three letters to hinder and help detect code forgery.