1. Introduction
Page 5
1.1 Background
The authors of this report became involved with the costs of blindness research project during 2005. It should be noted that Jonathan Godfrey was involved before this point through his activities as an elected representative for the Association of Blind Citizens of New Zealand, and that organization's actions as part of the validation Group led by Clive Lansink. This group's work culminated in late January 2005, although Clive Lansink (as Association of Blind Citizens of NZ representative), Jonathan Godfrey (statistician), Don McKenzie (RNZFB Board Chairman), Geoff Warne and Greg Morgan (RNZFB staff) met during 2005 to discuss further improvements of the report prepared by Research and Strategy Ltd and Market Economics Ltd, hereafter referred to as the Gravitas report. As a result of the involvement with the Clive Lansink led group, Jonathan Godfrey was able to learn that more could be done with the data that Gravitas had collected on the RNZFB's behalf. An informal conversation with Paula Daye (CEO of the RNZFB) led to a more formal arrangement being created that an opportunity for Jonathan Godfrey and Deborah Brunning to extract more detailed information from the data.
Thanks are given to the staff of the RNZFB and Gravitas for their assistance in providing the data and additional feedback during the initial phases of this supplementary investigation.
1.2 Authors' credentials
Jonathan Godfrey is a lecturer at Massey University in the Statistics group of the Institute of Information Sciences and Technology. He holds a PhD
Page 6
in Statistics and has a keen interest in advocacy issues, both for the blind and wider communities. Deborah Brunning holds a Bachelor of Education degree, and diplomas in Mathematics Education and Statistics and is currently completing her master's degree in Statistics. She has worked with the costs of blindness data as part of her academic qualifications under Jonathan's supervision. This staff-student relationship has provided a unique, low-cost opportunity for the costs of blindness data to be given greater analysis.
1.3 Scope of the Report
There are two main thrusts of this report. First, a review of the opportunities and limitations for current and future uses of the costs of blindness data. Second, the authors have presented some work on estimating the true cost of some aspects of blindness by accounting for the limited financial resources of respondents. This work is supplemented by some additional analyses of the data to gain a greater understanding of the RNZFB membership.
The majority of the material behind the information presented in this report is in the form of tables, graphs, and some output from statistical software. Only the important details are presented in the body of this report, while the appendices offer greater detail for the more interested reader. Figures and tables have titles that describe their content in an attempt to make the document blind-friendly, and only those exhibits from the appendices that are thought necessary have been included in the main report. Any material that also appears in the appendices is labelled according to the numbering used in the appendices, while additional graphs and tables are numbered separately in the body of the report.
Page 7
Those readers wishing to know what work has been undertaken by the authors are advised to read the main part of the report, while the appendices remain a useful tool for those people that will need more detailed information for submissions or in conjunction with other research. Any correspondence with the authors should be directed through Jonathan Godfrey if further explanation of the material in the appendices is required.
1.4 Terms used in this report
A list of terms used in this report that may not be familiar to the reader and could lead to confusion are briefly described here.
censoring
This is when we know that a single numerical observation actually represents a range of possible values rather than a fixed point. It often arises in medical research when at the end of an experiment, a person remains alive so we say they survived at least x months. In this situation, we know they have survived at least that long, and our analysis would take this into consideration. In this way, the known length of their survival is credited to the treatment they were given, and an adjustment for the fact that they live for a bit longer is made. In this report, we say that some costs are censored as people would have spent more money on something if they had the opportunity to do so.
mean
The simple arithmetic average found by adding all values and dividing by the number of values. It is often influenced by very large (unusual) values and is often not used for this reason.
Page 8
median
The point in a data set that is in the middle. It is greater than 50% of the values and less than the other 50%. This is often used as a good substitute for the mean when values are not spread evenly. Examples in common use are house prices, and incomes.
outlier
An observation that is considered unusual as it does not seem to belong to the set of results for some reason. More formally, large outliers are defined as being greater than the point equal to the upper quartile + 1.5 times the interquartile range.
personal cost
The values used for analysis in this report are called "personal costs" which actually means they are monies spent by the person surveyed, or on their immediate behalf by their family.
population
The total group of people that survey results relate to. In this instance the population is the RNZFB membership at the time the sample was drawn from the RNZFB database.
propensity
Whatever we talk about in life, each person has a propensity to do an activity. The term is interchangeable with probability, or proportion (if expressed as a percentage). We might say that a person will walk home 60% of the time so has a propensity to walk of 0.6. This value is found using averaging over all people that are similar in some way. We might say that blind people in general walk home 60% of the time and use this as the basis for a propensity for an individual. In this presentation, it is used as an indicator of the proportion of blind or vision impaired people that incur a cost.
quartiles
Three numbers that split an ordered set of numbers into four equally sized groups. The first (or lower) quartile is the point which is greater than 25% of the numbers and less than the
Page 9
remaining 75%. The median is the middle point and splits the set into two equal halves, and the upper quartile is the point which is greater than 75% of values and less than 25% of values. The difference between the upper and lower quartiles is known as the inter-quartile range.
sample
The group of people selected to answer the survey questionnaire. They are usually selected at random, but at times, constraints are placed on the number of people from certain demographic subgroups.
survey weighting
The method used to find how many members of a population are represented by a single person in the survey. This is important when some demographic groups are over- or under-represented in the sample. Individuals in the sample that come from over-represented demographic subgroups in the population represent fewer people than do those from under-represented subgroups. These survey weights were provided by Gravitas, but have been converted to integer values for many analyses, which has introduced differences in some totals.