外文文献翻译

The effects of subjective norms on behaviour in the theory of planned behaviour: A meta-analysis

Mark Manning*

University of Massachusetts, Amherst, Massachusetts, USA

A meta-analysis investigated the effects of perceived injunctive (IN) and descriptive (DN) norms on behaviour (BEH) within the theory of planned behaviour (TPB) in a sample of 196 studies. Two related correlation matrices (pairwise and listwise) were synthesized from the data and used to model the TPB relations with path analyses.Convergent evidence indicated that the relation between DN and BEH was stronger than the relation between IN and BEH. Evidence also suggested a significant direct relation between DN and BEH in the context of TPB. A suppressor effect of IN on DN in its relation with BEH was also noted・ Moderator analyses indicated that the DN-BEH relation was stronger when there was more time between measures of cognition and behaviour, when behaviours were not socially approved, more socially motive and more pleasant: results were mixed in the case of the IN-BEH relation. Results imply that IN and DN are conceptually different constructs・

As social beings, normative pressure inevitably affects our behaviour・ Social nonns influence the way we dress, how we vote, what we buy, and a host of other behavioural decisions.Social psychologists have been exploring the influence of social norms on behaviour for decades・ From AschM and Milgram s conformity- experiments (Asch, 19S6;Milgram, Bickman, & Berkowitz, 1969) through recent work by Cialdini and colleagues(Cialdini, Reno. & Kallgren, 1990; Reno, Cialdini, & Kallgren, 1993), a substantial body of evidence has demonstrated that people conform to the judgments and behaviours of others.

In experiments conducted by Cialdini and his colleagues (Cialdini et al., 1990; Reno et al., 1993), participants inferred behavioural norms for littering from environmental cues and acted in accord with these norms. The results highlight the fact that perceptions of norms, ratber than actual norms, can affect behaviour・ Tlie relation between perceived norms and behaviour has received much empirical support (Borsari & Carey, 2003; Campo, Brossard. Fnizer. Marchell, Lewis, & Talbot, 2003; Gomberg, Schneider, & Dejong, 2(K)I; Grube, Morgan, & MeGree, 1986; Okun, Karoly, & Lutz,2002; Riniai & Real. 2005). However, one ofthc most influential models for predicting behaviour, the thcor>*of planned behaviour (TPB; Ajzcn, 1991), posits that rather than a direct relation between norm and behaviour, perceived nortns influence behaviour indirectly by way of behavioural intentions. Investigating the perceived norm-behaviour relation in tlic context of this theory offers insight not only into the strength of the relation, but also into the extent to which perceived norms may directly influence behaviour counter to theoretical expectations.

The present study used mcta-analytic path analyses to examine, the relation between two types of perceived norms (injunctive (IN) and descriptive (DN) norms; described below) and behaviour in the context of the TPB (Ajzcn. 1991). The investigation explored the direct effects of IN and DN on behaviour as well as factors that may moderate the effect of subjective norms (SN) on behaviour・

The theory of planned behaviour

According to the TPB, the immediate antecedent of behaviour is the intention to pertbrm the behaviour (Figure 1). This behavioural intention is in turn a function of three major determinants: attitude towards the behaviour, perceived SN pertaining to the behaviour, and perceived degree of control over engaging in and ctJmpleting the behaviour (perceived behavioural control).

The formation of attitudes (ATT), SN and perceived behavioural control (PBC) are respectively functions of behavioural beliefs, normative beliefs and control beliefs that a person holds with regards to the behaviour・ Concerning ATT, the set of accessible beliefs that a person holds about the outcome of a behaviour will determine the evaluation of the behaviour, and thus influence the strength and direction of the ATT towards the behaviour.SN are a function of the normative beliefs that people relevant to the individual are perceived as having towards tbe behaviour coupled with the motivation of the individual to comply with the expected notins of these relevant persons・ PBC is a function of the perceived factors that will influence the ability to engage in the behaviour coupled with the perception as to whether or not these factors will be present.

In short, the TPB holds that favourable ATT, SN. And perceptions of control will lead to favourable intentions to engage in a given behaviour. Actual control over engaging in the behaviour is itself an important determinant・ To the extent that individuals realistically appraise the amount of control that they have over the behaviour, the measure of PBC; can serve as a proxy for actual control. Perceived control is expected to have a moderating effect such that intentions will be reflected in actual behaviour to the extent that perceived control is high.

The TPB has been applied successfully to a wide range of behaviours accounting for a sizable amount of variance (Armitage & Ckmner, 2001: Bamberg, Ajzen, & Schmidt,2003; Hardeman. Johnston. Johnston, Bonetti, Wareham, & Kinmonth. 2002; Povey.Wellens, & Conner, 2001; Rise. Thompson. & Verplanken, 2003). Regarding the SN construct, the theory holds that the effect of SN on behaviour is fully mediated by behavioural intentions・ In other words, SN are not expected to have a direct effect (DE)on behaviour but instead influetice behaviours indirectly through their effect on intentions.

Descriptive and injunctive norms

Two types of SN can be distinguished. IN are social pressures to engage in a behaviour based on the perception of what other people want you to do whereas DN are social pressures based on the observed or inferred behaviour of others・ Tliis distinction has been empirically supported (Cialdini et al .,1990; Deutsch & Gerard.1955; Grube et al., 1986; Larimer & Neighbours, 2005; Larimer. Turner, Mallett. & Geisner, 2004; Reno et al.,1993; Rhodes & Courneya, 2003; White, Terry, & Hogg, 1994). Within the TPB, the SN construct was originally conceptualized as an injunctive norm (Ajzen, 1991). More recently, however, Ajzen and Fishbein (200S) have recommended including both types of normative measures in constructing planned behaviour stirveys・ DN and IN will therefore be considered separately in the analyses to follow.

Subjective norms-behaviour relation

In reviewing the SN construct in the planned behaviour context, Conner and Armitage(1998) have noted the lack of predictive power of the IN construct when predicting intention. Due to the paucity- of studies including DN in the planned behaviour context,conclusions regarding DN in this context are sparser. Recently, several investigators have included DN as predictors of intentions in the planned behaviour model (PBM;Fekadu &

Kraft, 2002; MCiMUlan & Conner, 2(K)3; Okun et al.. 2002: Sheeran & Orbell, 1999b). Rivis and Sbeeran (2003) conducted a meta-analysis of DN in the planned behaviour context. Their analysis, based on 18 studies, demonstrated a significant relationship between DN and intention when controlling for otlier variables in the TPB.In that, these previous studies have investigated the effects of SN on intentions, to date,no planned behaviour mcta-ana lysis has explored the potential for differences in the effects of SN on behaviour in the planned behaviour context.

Deutsch and Gerard (1955) have suggested that DN and IN refer to different sources of motivation. Regarding DN, it has been shown that perceptions of behaviours of others lead one to behave in similar manners (Asch, 1956;Milgram et al., 1969). Descriptive normative information functions as a heuristic with regards to behavioural decisions offering cues as to what is appropriate behaviour iii a given situation (Cialdini et al., 1990; van Knippenberg, 2000). IN on the other iiand operate more through the role of motivation to comply with social sanctions (Ajzen, 1991;Lapinski & Rimal, 2005). To the extent that individuals are motivated to comply with perceived behavioural expectations of relevant referents, they avoid social sanctions・

Though several studies have looked at the effect of one or botli types of norms on particular behaviours, there has yet to be a single meta-analytical review that compares the relationship between the two types of norms and behaviours across a spectrum of behaviours. Consequently, on a general level it is unknown whether one type of norm has a stronger effect on behaviour than the other it may be hypothesized that DN have a stronger effect on behaviour than IN because DN are activated in the immediate behavioural situation. Furthermore, processing of DN for behavioural decisions may require less cognitive effort relative to the processing of IN, in that DN may rely more on heuristic than systematic informatioprocessing・ Perhaps, this advantage contributes to efficient behavioural decision・making in line with descriptive normative information. In fact, researchers have shown that conditions that facilitate the use of heuristic information-processing lead participants to act more in line with DN (Hertel, Neuhof, Theucr, & Kerr, 2000). It is expected therefore, that DN will have a stronger effect on behaviour relative to IN.

Direct effect ofSN on behaviour

The TPB posits that the relationship between SN and behaviour is fully mediated by behavioural intentions (Ajzen, 1991; Ajzen & Fishbein, 1973)・ However, a number of planned behaviour studies that have included normative constructs as a behavioural predictor have found direct effects of SN on behaviour (Christian & Abrams, 2004 -Study 2; Christian & Arm让age, 2002; Christian, Armitage, & Abrams, 2003; Okun et al.,2002; Trafimow & Finlay, 2001). In most research with the TPB, the effect of the normative component on intentions has received most attention (Armitage & Conner,2001; Rivis & Sheeran, 2003) while the potential for a DE of SN on behaviour has received little empirical or meta-analytical scrutiny.One reason to explore, the potential for a DE may be the hypothetical nature under which most people report cognitions pertaining to behaviour in planned behaviour studies・ Hypothetical contexts may not accurately reflect the relations between cognitions and behaviours that are evident in real behavioural contexts (Ajzen, Brown, & Carvajal, 2004). Furthermore, when an individual reports an intention to engage in a particular behaviour in one instance, that behavioural intention may be subject to change from the instance it is formed to the moment when an opportunity for behavioural engagement arises (Ajzen, 1991).For example, in the classic linn (1965) study, hotel managers expressed little intent to allow Chinese couples to stay in their hotels, however allowed them to do so when the instance arose・ It is less likely that perceptions of norms related to the behaviour will change over time. Consequently, there is the potential for reported normative perceptions to have stronger relations with behaviour compared with relations between reported behavioural intentions and behaviour. This may be reflected in the presence of a DE of SN on the particular behaviour. The present meta-analjtical synthesis provides the opportunity* to gauge the potential for a direct relation between SN and behaviour in the context of the TPB.

Variation in the magnitude of the SN^ehaviour relationship

The possibility of a DE of SN on behaviour within the TPB implies that there are two ways in which SN can affect behaviour. There can be the theoretically posited indirect effect on behaviour mediated through intentions, and there may be a DE on behaviour. The total effect therefore is the sum of these two effects・ In accord with the prediction that DN have a stronger relation with behaviours compared to the IN-behaviour relation, it is expected that the total effect of DN on behaviour is greater than the total effect of IN on behaviour. In addition to predicted differences between DN and IN in their effects on behaviour, there is the potential for differences in the magnitude of the effect within each type of norm. Compatibility* between measures of cognition and behaviour and the time between measurement of cognitions and behaviour are expected to lead to differences in the magnitudes of the effects of SN on behaviour. Additionally, the potential moderating effect of three further variables will be explored;the level of social approval of the behaviour, the extent to which social motives underlie behaviour, and the extent to which a behaviour is uselial versus pleasant may all contribute to variance in the relationship between norms and behaviour.

Compatibility

Elements of a particular behaviour can be defined in terms of the behavioural target, the action involved in the behaviour, the context in which the behaviour is performed, and the time at wliich it is performed. The relationship between cognitive predictors of a particular behaviour and engagement in the behaviour will be stronger if behavioural elements and cognitive assessment of the behaviour are compatible (Ajzen, 1996; Ajzen & Fishbein. 1977). That is to say., for instance, that if an investigator would tike to pretlict someone's propensity* to exercise 3 days a week for half an hour, measures should assess cognitions regarding exercising 3 days a week for half an hour rather than cognitions to be healthy, or some other general cognition regarding exercise・ Tenned the ”principle of compatibility0, it holds that measurements of planned behaviour variables must be compatible with the target behaviour in terms of target, action, context, and time. Given the effect of compatibility and the magnitude of the correlations between planned behaviour variables and behavioural measures, it is expected that studies where the cognitive and behavioural measures are fully compatible will feature stronger relations between SN and behaviour. It is also expected that among studies where measures are more compatible, the intention mediated relation between SN and behaviour will be stronger than any unmediated relation, in line with theoretical dictates, whereas among studies that are less compatible there will potentially be greater direct effects of SN on behaviour.

Time interval between measures of SN and behaviour

According to Ajzen ( 1991 ). cognitive precursors of behaviour that are measured closer to the target behaviour should be more predictive of behavioural engagement. Due to motivational considerations, measures of the intention to engage in a particular behaviour will vary as a function of proximity to behavioural engagement (Bandura & Schunk. 1981; Kamiol & Ross, 1996; Steel & Konig, 2006) in that tlie ftirther in the future is the potential behavioural engagement, the less predictive are intentions to engage in this behaviour. As Ibe relation between stated intentions and actual behaviour decreases over time, the potential exists for SN to be relatively more predictive of behaviour. This potential is evident in light of the argument outlined above wherein SN pertaining to a behaviour are less likely to change over time compared to behavioural intentions. As such, it is expected that as the time between measurement of cognitions and behaviour increases, SN will be reflected to a greater extent in actual behaviour.Furthermore, as the relation between intentions and behaviour diminishes, it is likely that the DE of SN on behaviour will be stronger as more time passes between measures of cognition and behaviour.

计划行为理论

根据TPB理论，行为的直接前因是执行行为的意向。这种行为的意图主要包括三个因 素：态度的行为，知觉的SN有关行为，从事和完成控制的认知度行为(知觉行为控制)。

态度(ATT) , SN和知觉行为控制(PBC)的形成分别由于行为的信念，规范信念和控 制信念，是人持有对待行为的。有关ATT,访问信念，一个人拥有一个行为的结果将决定行 为的评价，从而影响态度对于行为的强度和方向。SN形成于规范信念，有关的人个别被视 为对行为的动机，加上有个人遵守有关人士预期的规范。PBC是知觉因素的作用，这将影响 从事行为的能力以及这些因素呈现与否的观感。

简而言之，TPB认为，良好的ATT, SN,和控制的看法会导致有利的意图从事某一行为。 从事实际控制行为本身就是一个重要的决定因素。在某种程度上，个人实事求是地评价了控 制他们的行为的量，PBC的测量可以作为一个实际控制代理。预计知觉控制有一个调节作用， 这样的意图将反映在实际行为知觉控制，程度是高的。

TPB已成功地应用于广泛的行为占相当数量的方差。关于SN结构，该理论认为，对SN 对行为的影响完全中介行为意图。换句话说，SN预计不会对行为有直接影响(DE)的行为， 而是通过他们对意图的影响间接影响行为。

描述和强制性规范

SN可以区分为两种类型。IN是基于感知其他人要你做什么的从事行为的社会压力而DN 是基于别人被观察或被推断的行为的社会压力。这个区别得到实证支持。根据TPB, SN结构 最初概念化作为强制令规范，然而最近Ajzen and Fishbein (2005)建议包括两种类型的规范 措施在建设规划的行为调查中，DN和IN将因此被认为是单独分析。

主观规范行为的关系

在审查计划行为方面的SN结构中，Conner and Armitage (1998)已经注意到在预测意图 时预测能力的缺乏。由于包括在计划行为方面的DN研究的不足，在这方面的DN的结论是稀少 的。近日，一些研究者已经把DN作为意向的预测列入计划行为模型，进行DN在计划行为方面 的meta分析。他们的分析，根据18项研究，表现出在TPB中控制其他变量时DN和意图之间的 关系。在这，这些以前的研究已经考察了SN对意图的影响，迄今，没有计划行为meta分析探 讨了 SN在计划行为方面对行为的影响的潜在差异。

Deutsch and Gerard (1955)有建议DN和IN涉及不同来源的动机。关于DN,它已被证明 对他人的行为的看法导致一个类似的举止行为。描述规范性信息可以作为行为决定的启发, 在特定情况下提供线索。IN另一方面，通过角色的动机以符合社会的制裁。在某种程度上， 个人以符合所指有关的知觉行为的期望为动机，他们避免了社会的制裁。

虽然一些研究着重一种或两种类型对特定行为的规范的影响，目前尚未有是一个单一的 meta分析审查，比较两种类型的整个频谱的行为规范和行为之间的关系。

因此，一般水平，一个规范类型对行为是否比其他类型具有更强的影响是未知的.可以 推测该DN相比于IN对行为有更强的影响，因为DN启动在即时的行为情况。此外，DN对行为决 定的处理可能需要相对较少的认知努力相较于IN的处理，DN可以更多地依靠启发式信息处 理。也许，这一优势有助于有效的行为决定和描述规范信息。事实上，研究人员已经证明， 方便使用启发式信息处理这一条件引导参加者采取更多行动与DN,因t匕预计，DN将对行为有 更强的影响相较于IN。

SN对行为的直接影响

TPB认为SN和行为之间的关系完全通过行为意图断定。然而，一些计划行为研究，包括 规范行为的预测结构的研究发现，SN对行为的直接影响。在TPB大多数的研究中，规范部分 对意图的影响已受到到很大关注，而SN对行为的影响的DE潜力只有一点经验或meta分析的审 议。探索的一个原因，根据大多数人在计划行为研究中的认知行为报告，DE的潜力可能是假 设性质。假设背景下，可能无法准确反映在实际行为背景下是明显的认知和行为之间的关系。 此外，个体在一个实例中有意从事特定行为，在行为参与的机会出现时，该行为意图可能会 从形成时刻的实例受到改变。例如，在经典的林恩(1965)的研究中，酒店经理表示没有意 图让中国夫妇留在他们的酒店，但当这样做的实例出现吋则允许。有关行为准则的观念会随 时间而改变，这是不太可能的。因此，报告规范性的看法与行为有更强的关系是有可能的， 与报告行为意图和行为之间的关系相比。这可能体现在SN对特定行为的直接影响的存在。本 meta分析的合成提供了衡量在TPB背景下SN和行为之间直接关系的潜力的机会。

变化的幅度的的SN行为关系

在TPB理论下SN对行为的直接影响的可能性意味着有两种SN可以影响行为的方法。对 行为理论上假定的间接影响可以通过意向被介导，有可能有对行为的直接影响。因此，总的 效果是这两种效应的总和。与预测一致，与IN相比，DN与行为的关系更强，据预计，DN 对行为总的影响比IN对行为总的影响更大。此外DN和IN对行为影响预测的分歧，有可能 是数量级的差异对每个类型的规范作用。认知措施和行为之间的兼容性以及认知措施和行为 之间的的时间，预计将导致SN对行为的影响程度的差异。此外，另外三个变量的影响潜在 的干扰，将被探讨：社会认可的行为水平，在何种程度上的社会动机的基础行为，以及行为 是有用与愉快的，可能都有助于规范和行为之间的关系方差的程度。

兼容性

一个特定的行为的元素可以被定义在行为目标方面，参与行为的行动，行为执行的背景 以及它执行的吋间。一个特定行为的认知预测和参与行为的关系将会更强，如果行为元素和 行为的认知评估是兼容的。也就是说，例如，如果研究者想预测别人每周3天一个半小时的 锻炼的倾向，措施应评估认知方面锻炼一个半小时，每周3天，而不是对健康的认知，或一 些其他一般的锻炼认知。被称为“兼容性原则”，它认为，计划行为变量的测量必须与目标, 行动，背景，和时间方面的目标行为兼容。由于计划行为变量和行为措施之间兼容性和相关 性幅度的影响，预计认知和行为的措施是完全兼容的研究，SN和行为之间的关系将更强。 还预计，措施越兼容的研究，SN和行为之间的意向介导的关系将是比任何中间人的关系更 强，而理论支配，众研究中越不兼容的SN对行为的影响将有可能更直接。

SN措施和行为之间的时间间隔

根据Ajzen (1991),行为的认知前兆测量越接近目标行为前兆，行为参与应该是越预 测性的。由于动机的考虑，从事特定行为的意图的措施会变化为和行为参与的接近，在未来 的进一步是潜在的行为参与，从事这种行为的意图是具有较少预测性的。随着时间的推移, 确定意图和实际行为之间的关系下降，SN是相对较预测性行为的可能存在。这种潜力是显 而易见的，以上其中SN所列行为有关的参数是不太可能随着时间的推移改变的，与行为意 图相比。正因为如此，它预计，由于认知措施和行为之间时间增加，SN将反映在更大程度 上的实际行为。此外，意图和行为之间关系的下降，随着认知措施和行为之间更多吋间的传 递，可能SN对行为的直接影响会变得更强。

08国贸3班