Appendix C: False positives in the 2006 Participation Activity Limitation Survey

Warning View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

As a new module for the 2006 PALS, the false positive module marked a new direction in the analysis of respondents to a post-censal activity limitation survey. The 2001 PALS found that approximately 25% of respondents who had indicated a limitation on the Census and were selected for PALS did not have a limitation according to the subsequent PALS interview. This attrition of the 2001 PALS sample prompted numerous questions concerning the characteristics of this group and why there is inconsistency in their reporting of activity limitations. There are several interesting characteristics of the false positive population that are discussed in this appendix, including age groups and proxy effects.

Age groups

Previous surveys on activity limitations have found a strong association between age and the prevalence of activity limitation. Similarly, there is also a strong association between the false positive rate and age, although the direction of this relationship changes with the age groups.

The most difficult age group in which to identify activity limitations is children age 0 to 4. The 2006 PALS surveys found that 61.8% of children 0 to 4 for whom a limitation was indicated on the Census and were selected for PALS did not have any limitation according to the definition of disability used by PALS. The rapid development of children within this age group typically leads to the misidentification of minor delays as limitations or disabilities.

The false positive rate drops sharply beyond the 0 to 4 age group to approximately 30% for children 5 to14 followed by a relatively steady decrease to 11.2% in 2006 for the 75 and over age group. There are, however, two interesting points to note on this declining trend. First, there is a considerable jump in the false positive rates for the 15 to 24 age group followed by a return to the steady decline. This age group marks the transition between the adult and children's versions of the questionnaires yet everyone receives the same filter questions on the Census. Thus, we are looking at the merging point of the adult and children's PALS populations and should expect some variation.

The second interesting pattern observed in the 2006 false positive rate is a dramatic shift in the distribution of the false positives. There is a considerable increase in the false positive rate for the 0 to 4 age group between 2001 and 2006, increasing 6.6 percentage points during the 5 year period from 55.2% to 61.2% in 2006. This gap slowly narrows as the age groups increase until the false positive rates for 2001 and 2006 are identical at 26.7% for the 25 to 44 age group. The opposite relationship becomes evident at the other end of the age spectrum where the false positive rate drops 5.1 percentage points from 16.3% to 11.2% between 2001 and 2006 (see Chart 6).

Chart 6 False positive rate, by age group, Participation and Activity Limitation Survey, 2001 and 2006. Opens a new browser window.

Chart 6 False positive rate, by age group, Participation and Activity Limitation Survey, 2001 and 2006

Proxy effects

Defining disability is a very difficult task for a single person. This task becomes doubly difficult when comparing the definitions of disability between two people, particularly when the two people are trying to agree on whether one of them has a disability. The person without a disability may find it easier to label someone else as having a disability compared to labeling themselves as having a disability. Following this stream of analysis, there are some interesting proxy effects in the false positive population concerning the identification of activity limitations and reasons for someone being a false positive. It should also be noted that this discussion of proxy and non-proxy effects relates to the adult population only.

The first notable difference between the proxy and non-proxy false positive populations is the 5.8 percentage point difference between the false positive rate for proxy interviews (13.6%) compared to the non-proxy interviews (19.6%). Given that much of the Census is completed by proxy, the lower false positive rate for proxy interviews in PALS suggests that there is more consistency when one or more proxies are identifying a limitation in a third person versus a proxy and a person with a mild limitation trying to agree if this limitation is sufficiently serious to be reported on a survey.

Further evidence of the inconsistency in the reporting of activity limitations between proxy and non-proxy respondents is found through an analysis of the reasons why the respondent became a false positive. The clearest evidence of this inconsistency is found by examining the proxy status and data on who completed the household's Census form for respondents who indicated that their inclusion in PALS was an error.

In cases where the target respondent had completed the Census form for their household, only 33.7% of PALS proxy interviews indicated the respondents inclusion in PALS was an error compared to 54.3% of PALS respondents who completed the PALS interview themselves. Previous PALS research has found that people with mild or cyclical limitations may report a limitation to the disability filter questions found on the Census and many other instruments on occasions when they are having a difficult time but will recover relatively quickly. Therefore, the discrepancy between the false positive rates for people who completed their own Census form but believe their inclusion in PALS was an error is likely attributable to the varying perception of the respondent's mild or cyclical limitation between the proxy and non-proxy participants. In addition, subsequent internal data capture quality assessments have revealed extremely low rates of errors, far below the error rate required to support the claims that participation in PALS was an error, indicating that participation in PALS was not a processing error on the part of the Census or PALS.

Table 13 Proxy rates and Census form completion for respondents who believe their inclusion in PALS was an error, Participation and Activity Limitation Survey, 2006. Opens a new browser window.

Table 13 Proxy rates and Census form completion for respondents who believe their inclusion in PALS was an error, Participation and Activity Limitation Survey, 2006


You need to use the free Adobe Reader to view PDF documents. To view (open) these files, simply click on the link. To download (save) them, right-click on the link. Note that if you are using Internet Explorer or AOL, PDF documents sometimes do not open properly. See Troubleshooting PDFs. PDF documents may not be accessible by some devices. For more information, visit the Adobe website or contact us for assistance.