How much trust can we have in responses about individual participation in crime as an offender gathered by social surveys?

How much trust can we have in responses about individual participation in crime as an offender gathered by social surveys?

Social surveys are described within the academic literature as consisting of systematic, structured questions. These can be used in the form of both self-completion questionnaires and face to face interviews (Payne and Payne 2004). Lynch and Addington (2010) note that most data that is currently used within criminological research derives from the collection of figures from social surveys such as self-reports of offending and victimisation studies. Lynch and Addington (2010) expand that the collection of self-reports of offending are used to identify predictors of offending and establish theories as to why certain individuals have a propensity to offend, whilst others do not. The aim of this assignment is to determine if responses from the participants of social surveys can be deemed as trustworthy and valid. For the purpose of this assignment, social surveys shall refer to self-reports of offending including self-completion questionnaires and face to face interviews.

It is suggested in the work of Thornberry and Krohn (2000) that when conducting surveys of self-report offending, there is no choice of method that is significantly superior. However, one may suggestargue that some methodological approaches may produce different results to others, and therefore affect the trustworthiness of responses. When conducting face to face interviews, it is argued by DiCicco-Bloom and Crabtree (2006) that a sense of rapport can result in respondents answering freely and honestly. This can result in offenders providing details of offending that may not occur under other circumstances. However, face to face interviews are also suggested to have negative effects on responses. Bryman (2008) notes that responses may be biased due to social desirability, in which participants answer in accordance to what they believe is the correct response. The second methodological approach of self-completion questionnaires arguably attempts to overcome such issues. Self-completed questionnaires arguably reduce the possibility of interviewer effect (Junger-Tas and Marshall 1999). Self-completed questionnaires allow for privacy and anonymity of results. This is argued to positively affect results, with respondents feeling more inclined to admit criminality (Junger-Tas and Marshall 1999). However, issues may arise in relation to literacy and competency (Bryman 2008). Bryman notes that illiteracy may result in misunderstanding and the input of incorrect data. As interviewers are not there to prompt, little can be done to prevent the affect upon validity. CASI (Computer Assisted Self-administered Interview) is arguably seen as the method with greatest strengths, and highest rates of admittance of offending (Thornberry and Krohn 2000). Privacy remains, whilst personal headphones allow questions to be read to participants, whilst answers are confidentially entered into a laptop (Thornberry and Krohn). Issues of reliability and consistency of responses can be tested through the use of computer aids. Roberts and Horney (2010) suggestargue that computers are able to automatically cross reference responses to assess validity and consistency. It is contendedargued that such methods increase the trustworthiness of social survey results.

Age is claimedargued to be an influential factor in the validation of responses. Kollmar et al (1998) note that adolescents are less inclined to share private information with adults, with the preference of confiding in peers. One may argue that this unwillingness may be related to the idea of getting into trouble with authorative figures. However, Junger-Tas and Marshall (1999) contradict this view. They imply that adults are most likely to lie or withhold key data as they perceive as having more to lose, whilst adolescents are more willing to admit criminality. Age is also identified by Krohn et al (2010) to affect trustworthiness. Krohn et al argue that consistency is affected due to longitudinal approaches. As perceptions change with age, so will responses and consistency, whilst panel effects occur based on previous answers. However, Loeber and Farrington (1994) champion the use of longitudinal approaches.

The seriousness of offences is also suggested to be a significant factor in validity of responses (Thornberry and Krohn 2000). It is proposedsuggested that respondents may be more willing to answer truthfully about minor offences as opposed to serious offences (Junger-Tas and Marshall 1999). As minor offences are often committed in high frequency, Huizinga and Elliot (1986) suggest that they are over reported. Respondents are said to estimate occurrences rather than recount actual offences (Lynch and Addington 2010). These false accounts affect lambda and the patterns of offending (Lynch and Addington 2010). In order to obtain valid and consistent results, academics aim to bind offences to time periods, and use recall aids in order to prevent telescoping of offences (Huizinga and Elliot 1986, Lynch and Addington 2010). Significant life events are used as memory cues, with specific time periods targeted (Roberts and Horney 2010). However, time periods of reference are suggested to be far too spatial, resulting in problems of memory, and factual recall (Lynch and Addington 2010). Lynch and Addington exemplify the Cambridge Youth Study, which ask 32 year olds to recall crimes from the age of 16. One may argue that such a large time period will affect the validity of responses. Roberts and Horney (2010) note that there is no universally accepted time span, with many researchers deciding time periods to complement their research. Lynch and Addington (2010) highlight further issues with the trustworthiness of responses, by stating that follow up questions fail to exclude unimportant and trivial data. They exemplify offences that do not occur in the correct time period which can result in both under and over reporting.

The aim of this assignment has been to provide a brief yet critical overview of the trustworthiness of responses from social surveys. Many issues have been highlighted such as recall, methodology and participant characteristics. However, many solutions and recommendations for improvement are available within the body of literature to improve the validity and replicabilty of responses. Lynch and Addington (2010) note that cross referencing of answers with official police statistics improves the trustworthiness of responses, whilst Huizinga and Elliot (1986) argue that binding events can improve recall and validity. Validity and replicabilty are argued to be crucial when assessing participant responses, with tests available to determine such concepts (Thornberry and Krohn 2000). Tests include the test-retest approach alongside the Rasch model which determines propensity and appropriateness (Thornberry and Krohn 2000). Thornberry and Krohn (2000) highlight the difficulty of providing absolute tests of validity and consistency of responses in self-report offending surveys, with no gold standard survey available to set precedent. However, it is concluded by Thornberry and Krohn (2000) that the collection of self-report offending statistics is valid for the research purposes of providing etiological and detailed data. One may therefore conclude that although problems in relation to responses exist, measures are put in place to assess and assure the validity and trustworthiness of participants’ responses.