« Previous | Contents | Next »
Listen
5. Victimisation
The estimation of victimisation is the principal purpose of the Scottish Crime and Victimisation Survey. In this chapter we look at the victimisation data collected by the face-to-face and telephone surveys in an effort to determine the extent to which the change to a telephone survey might have impacted on the estimates obtained.
At this stage we are still using data that has only had basic design weights applied. No account has been taken of the biases identified either at the sampling stage, through the exclusion of mobile-only and TPS households from the telephone survey, or at the data collection stage where both surveys will, in the various ways described above, have experienced response bias. In Appendix 2 we consider the ability of corrective weighting to take account of these.
5.1 How victimisation is recorded
Both the face-to-face and telephone surveys adopted a similar approach to identifying incidents of victimisation. Respondents were asked a series of 'screening' questions which identified whether the household/individual had experienced any incidents of any of 18 possible types of crimes within the survey reference period. In both surveys, the reference period was "since 1 April 2003".
Where respondents reported any incidents, these were followed up with a Victim Form (Victim Module in the telephone survey but we will refer to both as Victim Forms) which collects a detailed description of the incident and associated information such as whether the incident was reported to the police. To limit the burden on respondents, a maximum of five Victim Forms was completed for each respondent. In the face-to-face survey, five full victim forms were completed but in the telephone survey three full forms and two shortened forms were completed, again to reduce the burden on respondents. The limit of five forms results in some incidents not being followed up but this affects only a small number of respondents.
Each form can represent either a single incident or a series of incidents. A series of incidents occurs when the same type of offence is committed against the respondent over a period of time, probably by the same people. In a series, the incidents are both similar and related. For the purposes of analysis, series forms are weighted by the number of incidents they represent up to a maximum of five.
A key purpose of the Victim Form is to probe more fully on when incidents happened in order to determine whether the incident reported in the form actually fell within the year that is used for analysis of victimisation. For these surveys, the reference period for analysing victimisation is April 2003 to March 2004.
5.2 Victimisation in the surveys
Table 15 shows a number of comparisons of the level of victimisation recorded in the telephone and face-to-face surveys. In each case, the telephone survey records higher levels of victimisation.
- A higher proportion of respondents report any incidents (line 2)
- Respondents who report any incidents report more of them, completing more Victim Forms per person than similar respondents in the face-to-face survey (line 4)
- A lower proportion of incidents in the telephone survey are counted as eligible for inclusion in the analysis (line5) 29 but this still leaves relatively more victim forms per respondent (line 7)
- Each eligible Victim Form represents more incidents in the telephone survey (line 10), suggesting that the telephone survey has recorded higher levels of victimisation than the face-to-face survey.
Table 15 - The recording of victimisation in both surveys
| | Face-to-face survey | Telephone |
|---|
1 | Total respondents | 3,034 | 9,509 |
|---|
2 | Percentage reporting any incidents | 30% | 37% |
|---|
3 | Number of victim forms completed | 1,306 | 5,791 |
|---|
4 | VFs per 'reporting' respondent | 1.44 | 1.67 |
|---|
5 | Proportion of VFs eligible | 64% | 52% |
|---|
6 | Number of eligible Vfs | 836 | 3,006 |
|---|
7 | Eligible VFs per respondent | 0.28 | 0.32 |
|---|
8 | % of eligible VFs representing series incidents | 26% | 31% |
|---|
9 | Total incidents from eligible VFs | 1,186 | 4,572 |
|---|
10 | Incidents per eligible VF | 1.41 | 1.52 |
|---|
11 | Incidents per respondent | 0.39 | 0.48 |
|---|
5.2.1 Prevalence rates
The telephone survey records more eligible incidents per respondent interviewed but it also records a slightly different profile of victimisation in terms of the proportion of all incidents that are coded into crime categories. 30 In general, compared with the face-to-face survey, the telephone survey records more crime and this difference is particularly marked in relation to personal crime rather than property crime.
Table 16 shows the prevalence rates recorded in each survey - the percentage of respondents reporting any incidents of each type in the survey reference period - and compares the telephone survey with both the face-to-face survey and the 2003 Scottish Crime Survey. This is the last reported full sweep of the crime survey, with over 5,000 respondents compared with just over 3,000 in the face-to-face survey.
Table 16 - Prevalence rates for telephone, face-to-face and 2003 SCS
| Tel
(T) | F2F
(F) | 2003
(S) | Diff
(T-F) | | Diff
(T-S) | | Diff
(F-S) | |
|---|
Household crime |
|---|
Motor vehicle vandalism | 4.7 | 4.5 | 5.8 | 0.2 | | -1.1 | * | -1.3 | * |
|---|
Property vandalism | 3.5 | 2.7 | 4.4 | 0.8 | * | -0.9 | * | -1.7 | * |
|---|
Theft from motor vehicle | 2.6 | 2.6 | 3.0 | 0.0 | | -0.4 | | -0.4 | |
|---|
Attempted theft of/from mv | 0.6 | 0.5 | 1.1 | 0.1 | | -0.5 | * | -0.6 | * |
|---|
Theft of motor vehicle | 0.3 | 0.5 | 0.7 | -0.2 | | -0.4 | * | -0.2 | |
|---|
Bicycle theft | 0.8 | 0.8 | 1.2 | 0.0 | | -0.4 | * | -0.4 | |
|---|
Housebreaking | 2.6 | 1.8 | 2.9 | 0.8 | * | -0.3 | | -1.1 | * |
|---|
Other household theft | 3.3 | 3.2 | 3.3 | 0.1 | | 0.0 | | -0.1 | |
|---|
Any household crime | 15.5 | 14.1 | 18.3 | 1.4 | | -2.8 | * | -4.2 | * |
|---|
Personal crime |
|---|
Petty assault | 4.1 | 2.1 | 2.4 | 2.0 | * | 1.7 | * | -0.3 | |
|---|
Serious assault | 0.5 | 0.4 | 0.8 | 0.1 | | -0.3 | * | -0.4 | * |
|---|
Sexual offences | 0.4 | 0.2 | 0.0 | 0.2 | | 0.4 | * | 0.2 | * |
|---|
Theft from person | 1.2 | 0.3 | 0.5 | 0.9 | * | 0.7 | * | -0.2 | |
|---|
Robbery | 0.4 | 0.2 | 0.4 | 0.2 | | 0.0 | | -0.2 | |
|---|
Other personal theft | 2.3 | 1.7 | 2.2 | 0.6 | * | 0.1 | | -0.5 | |
|---|
Threats | 2.9 | 1.8 | 1.6 | 1.1 | * | 1.3 | * | 0.2 | |
|---|
Any personal crime | 7.2 | 4.0 | 5.4 | 3.2 | * | 1.8 | * | -1.4 | * |
|---|
| 9,509 | 3,034 | 5,041 | | | | | | |
|---|
* Indicates difference significant at 95% confidence limits. Eligible incidents only
There are three important points to note from this:
- the telephone survey generally records higher prevalence rates than the face-to-face survey. Where differences are significant, the telephone survey is consistently higher than the face-to-face survey but the prevalence rate for household crime is not significantly different. The rate for personal crime is greater by over 60%.
- the telephone survey records lower prevalence of household crime than the 2003 Crime Survey - a drop of 15%. Prevalence of personal crime is 33% higher than that recorded in 2003.
- the face-to-face survey also records lower prevalence of household crime than the 2003 Crime Survey although the drop is much larger at 23%. Personal crime is 26% lower than in 2003.
5.2.2 Incidence rates
Incidence rates express the number of eligible incidents recorded in the surveys as a number of incidents per 10,000 units of population. These units can be households, adults, car owning households or bicycle owning households depending on the offence. The rate itself is the product of multiplying the number of eligible incidents by a grossing factor equal to the number of 10,000s of units in the population.
Table 17 - Comparison of incidence rates - telephone, face-to-face and 2003 Scottish Crime Survey
| Tel (T) | F2F (F) | 2003 (S) | Sig diff
(T-F) | Sig Diff
(T-S) | Sig Diff
(F-S) |
|---|
Motor vehicle vandalism | 785.4 | 768.0 | 825.9 | | | |
|---|
Property vandalism | 709.6 | 550.4 | 830.5 | | | |
|---|
Theft from motor vehicle | 356.2 | 309.8 | 407.8 | | | |
|---|
Attempted theft of/from mv | 76.2 | 65.9 | 125.7 | | | |
|---|
Theft of mv | 35.8 | 62.6 | 90.9 | | | |
|---|
Bicycle theft | 124.9 | 85.7 | 131.9 | | | |
|---|
Housebreaking | 345.4 | 263.7 | 397.5 | | | |
|---|
Other household theft | 512.4 | 431.8 | 499.2 | | | |
|---|
Petty assault | 705.9 | 420.2 | 435.4 | | * | |
|---|
Serious assault | 69.0 | 49.3 | 114.8 | | | |
|---|
Sexual offences | 56.2 | 24.6 | 1.8 | | * | |
|---|
Theft from person | 122.7 | 32.4 | 59.4 | | | |
|---|
Robbery | 50.7 | 15.8 | 48.4 | | | |
|---|
Other personal theft | 283.9 | 248.3 | 260.9 | | | |
|---|
Threats | 523.8 | 283.1 | 174.0 | * | * | |
|---|
Any household crime | 2,945.7 | 2,537.9 | 3,309.2 | | | * |
|---|
Any personal crime | 1,232.4 | 765.92 | 1,094.8 | * | * | |
|---|
| 9,509 | 3,034 | 5,041 | | | |
|---|
Comparison of the incidence rates shows that even though they broadly reflect what was seen in terms of prevalence rates, none of the differences in the household incidence rates are statistically significant when offences are considered separately. The rates for 'any household crime' shows that the telephone survey is not significantly different from either of the two face-to-face surveys but the difference between the face-to-face survey and the 2003 survey is significant.
Apart from 'threats', the personal incidence rates recorded by the telephone and face-to-face survey are not significantly different although the overall 'any personal crime' category is significantly higher in the telephone survey. Compared with the 2003 survey, the personal incidence rates in the telephone survey are significantly higher for petty assault, sexual offences, personal theft and threats. Overall, 'any personal crime' is higher in the telephone survey in comparison with both the face-to-face survey and the 2003 survey.
5.3 Recorded crime
Trends in recorded crime from 2002 and 2003/4 tend to confirm that while the face-to-face survey appears to be understating household crime, the telephone survey appears to overstate personal crime (compared with 2003 SCS figures). In summary:
- the face-to-face survey shows a fall in vandalism of 20% compared with a fall of 10% in the telephone survey and an increase of 9% in recorded incidents of vandalism
- the telephone and face-to-face surveys both show a fall in vehicle crime of 25% and 30% respectively, which is broadly consistent with the trend in recorded crime showing a fall of 19%
- the fall in bicycle theft of 30% recorded by the face-to-face survey is not consistent with the trend in recorded crime, which is static. The fall of 13% recorded by the telephone survey is more consistent with recorded crime
- housebreaking in the face-to-face survey shows a fall of 29% and the telephone survey a fall of 5%, against a recorded decrease of 17%
- recorded crimes show a drop of 4% in personal theft (which combines theft from the person and other personal theft) while the face-to-face survey shows a drop of 13%. The telephone survey shows an increase of 12%
- recorded incidents of assault rose by 4% whereas the face-to-face survey shows a fall of 15%. The telephone survey shows an increase of 41%
- robbery has fallen by 15% in the recorded crime statistics whereas the telephone survey shows an increase of 5% and the face-to-face survey shows a fall of 67%.
5.4 Other factors affecting victimisation in the telephone survey
In the preceding chapters we discussed three features of the telephone survey that might be having an influence on victimisation rates. These were:
- the exclusion of no-phone and mobile-only households from the survey coverage
- an administrative error that screened out TPS subscribers from the survey sample
- a substantially higher refusal rate than the face-to-face survey, which opens the possibility that people refusing to participate had different experiences of victimisation.
The face-to-face survey gives an indication of the victimisation experience of no-phone and mobile only households. TPS subscribers were not screened out of the samples from January 2005 and we have been provided with data from January with TPS subscribers flagged, allowing analysis of the experience of TPS subscribers. The June-September 2004 achieved interviews have also been flagged to indicate whether respondents took part in the main sweep of fieldwork or whether their interview took place as part of the re-contacting of respondents who had initially refused. The victimisation experience of these 'soft' refusals gives an indication of the extent to which the experience of refusers might differ from more willing participants.
5.4.1 No-phone and mobile-only households
With only just over 200 respondents without a fixed telephone line, for the purpose of this analysis, no-phone and mobile-only have been collapsed into a single group and compared with households with a fixed telephone line (and possibly also a mobile phone). The results are shown in Table 18, which shows that there are no significant differences in the victimisation of the two groups in the face-to-face survey. It might be, however, that in a larger sample such as the annual sample for the telephone survey, that differences such as those recorded here for motor vehicle theft, housebreaking and serious assault would be significant.
Table 18 - Comparison of prevalence rates among households with and without a fixed telephone line
| No landline | Landline | All | Chi-sq | Sig |
|---|
Motor vehicle vandalism | 4.2 | 4.6 | 4.5 | 0.09 | |
|---|
Property vandalism | 3.0 | 2.7 | 2.7 | 0.09 | |
|---|
Theft from motor vehicle | 2.3 | 2.6 | 2.6 | 0.11 | |
|---|
Attempted theft from motor vehicle | 0.4 | 0.5 | 0.5 | 0.12 | |
|---|
Theft of motor vehicle | 0.8 | 0.5 | 0.5 | 0.40 | |
|---|
Bike theft | 0.8 | 0.8 | 0.8 | 0.00 | |
|---|
Housebreaking | 3.0 | 1.7 | 1.8 | 2.21 | |
|---|
Other household theft | 3.0 | 3.2 | 3.2 | 0.02 | |
|---|
Petty assault | 2.0 | 2.2 | 2.1 | 0.03 | |
|---|
Serious assault | 1.0 | 0.4 | 0.4 | 1.94 | |
|---|
Sexual assault | 0.0 | 0.1 | 0.1 | 0.29 | |
|---|
Theft from the person | 0.5 | 0.3 | 0.3 | 0.18 | |
|---|
Robbery | 0.0 | 0.2 | 0.2 | 0.43 | |
|---|
Other personal theft | 1.7 | 1.9 | 1.9 | 0.01 | |
|---|
Threat | 4.0 | 1.6 | 1.8 | 5.89 | * |
|---|
Any household crime | 14.7 | 14.0 | 14.1 | 0.10 | |
|---|
Any personal crime | 5.0 | 4.2 | 4.3 | 0.23 | |
|---|
Base - household | 265 | 2,769 | 3,034 | | |
|---|
Base - individuals | 202 | 2,832 | 3,034 | | |
|---|
5.4.2 TPS respondents
Comparison of prevalence rates for TPS and non- TPS respondents from January 2005 shows only one significant difference - the rate of serious assault among non- TPS respondents is significantly higher than among non- TPS. Even the broad categories of 'any household crime' and 'any personal crime' show no significant differences.
Table 19 - Prevalence rates among TPS and non- TPS subscribers in January 2005
| Non- TPS | TPS | All | Chi-sq | Sig |
|---|
Motor vehicle vandalism | 6.3 | 7.1 | 6.6 | 0.50 | |
|---|
Property vandalism | 5.2 | 5.4 | 5.3 | 0.03 | |
|---|
Theft from motor vehicle | 3.1 | 2.0 | 2.7 | 2.52 | |
|---|
Attempted theft from motor vehicle | 1.2 | 1.2 | 1.2 | 0.00 | |
|---|
Theft of motor vehicle | 0.3 | 0.7 | 0.4 | 2.81 | |
|---|
Bike theft | 0.9 | 1.6 | 1.1 | 2.14 | |
|---|
Housebreaking | 2.6 | 2.7 | 2.6 | 0.01 | |
|---|
Other household theft | 3.7 | 3.8 | 3.7 | 0.01 | |
|---|
Petty assault | 3.8 | 4.3 | 4.0 | 2.72 | |
|---|
Serious assault | 1.1 | 0.1 | 0.7 | 8.04 | * |
|---|
Sexual assault | 0.3 | 0.1 | 0.3 | 2.70 | |
|---|
Theft from the person | 0.9 | 0.9 | 0.9 | 1.86 | |
|---|
Robbery | 0.9 | 0.6 | 0.8 | 2.23 | |
|---|
Other personal theft | 3.2 | 2.7 | 3.0 | 2.02 | |
|---|
Threat | 4.3 | 3.5 | 4.0 | 2.19 | |
|---|
Any household crime | 18.6 | 19.5 | 18.9 | 0.32 | |
|---|
Any personal crime | 9.9 | 8.7 | 9.5 | 2.12 | |
|---|
Base | 1463 | 891 | 2,356 | | |
|---|
Again, we would caution that this may be a consequence of only having access to one month's sample - only 2,356 cases - and that a larger sample may show significant differences.
5.4.3 Refusal in the telephone survey
The June-Sept 2004 sample was flagged to show whether respondents had taken part in the initial round of fieldwork or whether they had been interviewed as part of a refusal sample that was re-contacted in the hope of converting them to interviews. To the extent that the experience of initial refusers is different from more willing participants, this gives an indication of the impact of the telephone survey's high refusal rate on victimisation.
Table 20 shows that on the two broad measures of victimisation - any household and any personal crime - refusers have significantly lower prevalence rates than those who took part in the initial fieldwork. Several individual categories show significant differences and in all but one (theft of a motor vehicle) refusers have lower prevalence rates than main respondents. Combining some of the smaller categories that are not separately significant, such as motor vehicle and property vandalism into any vandalism shows that the refusers also have significantly lower prevalence rates.
The implication of this is that 'hard' refusers - those who were firm refusals at the outset or who refused to take part in the recontact survey - might be expected to have at least similar prevalence rates as the refusers who were converted, and perhaps lower still.
Table 20 - Prevalence rates among initial refusers and main survey participants - June-Sept sample
| Main | Refusers | All | Chi-sq | Sig |
|---|
Motor vehicle vandalism | 4.8 | 3.8 | 4.7 | 1.72 | |
|---|
Property vandalism | 3.6 | 2.4 | 3.5 | 3.30 | |
|---|
Theft from motor vehicle | 2.8 | 1.3 | 2.6 | 6.88 | * |
|---|
Attempted theft from motor vehicle | 0.6 | 0.2 | 0.6 | 2.11 | |
|---|
Theft of motor vehicle | 0.3 | 0.8 | 0.3 | 5.80 | * |
|---|
Bike theft | 0.8 | 0.7 | 0.8 | 0.17 | |
|---|
Housebreaking | 2.6 | 2.3 | 2.6 | 0.25 | |
|---|
Other household theft | 3.3 | 3.6 | 3.3 | 0.24 | |
|---|
Petty assault | 4.3 | 2.4 | 4.1 | 7.21 | * |
|---|
Serious assault | 0.5 | 0.1 | 0.5 | 2.61 | |
|---|
Sexual assault | 0.4 | 0.3 | 0.4 | 0.10 | |
|---|
Theft from the person | 1.2 | 0.8 | 1.1 | 1.00 | |
|---|
Robbery | 0.5 | 0.1 | 0.5 | 2.42 | |
|---|
Other personal theft | 2.4 | 2.1 | 2.3 | 0.29 | |
|---|
Threat | 3.0 | 2.2 | 2.9 | 1.67 | * |
|---|
Any household crime | 15.7 | 13.1 | 15.5 | 4.17 | * |
|---|
Any personal crime | 7.5 | 4.6 | 7.2 | 9.93 | * |
|---|
Base | 8,693 | 870 | 9,509 | | |
|---|
The main conclusion to be drawn from this is that refusal from the telephone survey is biased and that non-victims are less likely to participate in the survey. This tends, overall, to inflate the survey's measure of victimisation and, taking account of the victimisation rates show in Table 17, seems to inflates measures of personal crime more than household crime.
5.5 Conclusion
Analysis of victimisation is limited by the small number of cases of victimisation arising in the face-to-face survey, which makes it difficult to identify any significant differences between the telephone and face-to-face surveys and between the face-to-face survey and the 2003 SCS data. Compared with the 2003 survey, the prevalence and victimisation rates for 'any household crime' in the face-to-face survey are lower. The rates for the telephone survey are higher for personal crime.
Analysis of the exclusion of households without a landline and households subscribing to TPS does not show any bias in terms of measures of victimisation although given the small samples involved, we cannot conclude that this means these have no effect, just that no effect is observed in this data.
Refusal to the telephone survey does appear to have a significant effect on victimisation, biasing the victimisation rates upwards. It is possible that the effect shown above is weaker than might actually be happening in practice i.e. if these refusers have lower rates than non-refusers, it is possible that the rates among those who could not be converted are lower still.
To the extent that the telephone survey is biased in favour of victims, this would be particularly problematic since it suggests that there can be no scope to weight the telephone survey to correct this bias. In Appendix 2 we outline two approaches to developing a weighting solution although we conclude that neither has been successful.
« Previous | Contents | Next »