International Journal of Accounting Information Systems 8 (2007) 1 16

Revisiting fit between AIS design and performance with the analyzer strategic-type
Emilio Boulianne
John Molson School of Business, Concordia University, Montreal, Quebec, Canada H3G 1M8 Received 15 January 2006; received in revised form 31 October 2006; accepted 26 December 2006

Abstract In considering the Miles and Snow analyzer strategic-type, this paper re-examines the relationship between strategic choice, accounting information systems design (AIS), and business-unit performance. This paper also examines the construct validity of the variables considered in the study. In response to previous studies that called for triangulation of variables with other sources, the validity of the results have been improved by using a multi-method, multiple-respondent approach for business strategy classification and business-unit performance assessment. Results of a research survey and secondary data analysis of 88 Canadian business units suggest that for prospector strategic-types, and to a lesser extent for defender strategic-types, broad-scope AIS is associated with higher performance. Managers of prospector and defender units need external, non-financial, and future-oriented information for decision-making, which represents an evolution in defenders' information needs. Results also support the assumption that information needs of analyzer strategic-type units differ from those of defender units. 2007 Elsevier Inc. All rights reserved.
Keywords: Accounting information systems design; Business strategy; Analyzer strategic-type; Performance; Multimethod; Multiple-respondent; Contingency fit

1. Introduction This paper aims to revisit our understanding of the relationship between business strategy, accounting information systems design (AIS), and business unit performance. Accounting theoreticians have claimed that, from a contingency perspective, the AIS must fit the unit's

This article received the Best Paper Award at the 2006 European Conference on Accounting Information System. Tel.: +1 514 848 2424x2754; fax: +1 514 848 4518. E-mail address: Emilio@JMSB.Concordia.ca.

1467-0895/$ - see front matter 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.accinf.2006.12.001

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

strategic-type to achieve performance (Simons, 1987; Chenhall and Langfield-Smith, 1998). Only a few studies, however, have investigated this essential relationship, limiting what is known about the association between AIS design and strategy (Langfield-Smith, 1997). The lack of studies in this area is surprising as studies including business strategy are considered to be among the most important new literature in contingency research (Chenhall, 2003). Inappropriate design and implementation of information systems has been very costly to firms in terms of time and money, suggesting the need for a careful examination of the fit between strategy and AIS design.1 For instance, previous research indicates that broad scope AIS has a positive effect on performance for firms operating under dynamic/hostile business environment and adopting an innovative strategic profile, while this same AIS setting (broad scope) may have a negative effect on performance for firms operating under stable/friendly business environment and adopting a defensive strategic profile (Gul, 1991; Abernethy and Guthrie, 1994). Previous studies investigating the business strategy/AIS link with Miles and Snow's (1978, 1994) strategy typology forced respondents to categorize their units as either defender or prospector (for instance see Simons, 1987). Such a forced and dichotomous classification between defender and prospector does not accommodate units that have pursued an analyzer strategic-type.2 Recent studies have consistently reported that when respondents have a choice between defender, prospector, or analyzer strategic-types to describe their units, they often choose the analyzer (see Collins et al., 1997; Sabherwal and Chan, 2001). Miles and Snow expected that information needs for managers of analyzer units would vary from those of prospector and defender strategic-types, although they did not describe how these information needs would differ. In the present study, the analyzer is included as a strategic-type. Examining the relationship between business strategy and AIS scope with the analyzer strategic-type allows respondents to choose the business strategy profile that may more closely correspond to their unit, and yields results that shed some light on the analyzer's information needs for decision-making; in other words, how analyzers' information needs differ from those of defenders and prospectors? To the researcher's knowledge, no other study has examined the strategy/AIS relationship with analyzertype businesses. It is important for research to have a better understanding of analyzer's information needs for decision-making since an appropriate fit between strategic-type and AIS design is a prerequisite to achieve higher business performance (Abernethy and Guthrie, 1994). The present research initiative aims also to encourage future studies to incorporate the analyzer strategic-type. Recent research indicates that defenders' information needs may have evolved, which leads us to examine whether they have changed and, if so, in what direction. This type of investigation is important for the AIS and contingency research literature as business units change their strategy to deal with new business environments, thus we need to reexamine the AIS design to include information that is pertinent for decision-making. This paper extends the Ismail and King (2005) study in the following ways: 1) Ismail and King restricted their study to manufacturing firms only, while the present study includes both manufacturing and service industries; 2) they adopted the moderation perspective to measure fit between variables, while the present study uses the matching perspective and 3) Ismail and King obtained subjective measures from a survey with one respondent per unit while in the present
AIS design is defined here as the scope of information used by managers for decision-making. While other strategic typologies have been conceptualized, Miles and Snow's (1978, 1994) is one of the most widely used, including in the field of accounting research.
2 1

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

study both subjective and objective measures are obtained from survey research and secondary data via two respondents per unit. A limitation of prior research stems from the ways that business strategies are categorized and business performance is measured. Sole reliance on subjective assessments has been criticized and a recommendation has been made for triangulation of strategy and performance variables with objective measures and multiple respondents (Govindarajan and Gupta, 1985; Young, 1996). One way to address this issue is to refer to the concept of convergent validity (the extent to which assessment of different measurement methods produces correlated measures Brownell, 1995) by using a multi-method and multiple respondent approach for strategy classification and performance assessment. It is important to obtain triangulation of the measures investigated since this approach reduces the threat of response bias and enhances the construct validity of variables. Drawing upon contingency research in accounting, hypotheses are developed in this study regarding the fit between business strategy and AIS design. The hypotheses are empirically tested by gathering data from a sample of business units. Two questionnaires were sent: the first asked for objective data and was to be answered by the highest position in finance, while the second collected subjective assessments and was to be answered by highest management position. The final sample consists of 88 pairs of usable questionnaires. The results suggest that for prospector strategic-types, and to a lesser extent for defenders, broad-scope AIS is associated with higher performance. Managers of prospector and defender units need external, non-financial, and future-oriented information for decision-making. This represents an evolution in the information needs of defenders. Results also support the assumption that information needs of analyzer differ from those of defenders. The paper is organized as follows: Section 2 presents the hypotheses and the research model; Section 3 describes the methodology; Section 4 reports the results. The last, Section 5, discusses the findings and limitations of the study, and suggests directions for future research. 2. Hypotheses development and research model Miles and Snow's theory of strategy provides the foundation of the research model as it posits higher performance for units that achieve fit among strategic and contextual factors. For example, defenders' AIS is characterized as lacking information regarding new product development and market opportunities while prospector's AIS hold information on innovation, latest product/ service development. Mismatch between AIS design and business strategy results in lower performance. Miles and Snow theory of strategy, the AIS scope construct, and empirical results are drawn together to build the hypotheses. The following sections elaborate on these relationships. 2.1. Definition of constructs 2.1.1. Business strategy Prior research suggests that business strategy is a key driver of accounting systems attributes (Simons, 1987; Abernethy and Guthrie, 1994; Chong and Chong, 1997; Banker et al., 2004). Although there are several different strategy typologies, such as Mintzberg (1973), Porter (1985), and others, Miles and Snow (1978, 1994) provide a pertinent theoretical framework for identifying the characteristics of information that fit specific business strategies. Miles and Snow are very clear in their statements that the control system of a firm should be congruent with its strategy (Simons, 1987, 359).

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

Miles and Snow formulate three viable strategic-types: defenders, prospectors, and analyzers.3 Defenders deal with a stable environment, propose few products and services, and preserve their market share through cost leadership. Prospectors deal with a dynamic environment, offer many products, and are innovators. Compared with prospectors and defenders, analyzers' business units are more heterogeneous. Analyzers sustain a stable core of products and services while attempting to be leader for some products based on concepts already introduced by prospector-type businesses; in doing so, for a set of products, analyzers act as second in on the market. Dealing with both efficiency and innovation objectives requires dual manufacturing settings and various control mechanisms, making the management of analyzers more critical and problematic. 2.1.2. Accounting information system (AIS) scope Information provided to managers by AIS can be characterized in many ways. Accounting researchers have paid specific attention to three characteristics: focus, quantification, and time horizon (IMA, 1995). Focus refers to the extent to which information focuses internally, that is, on the organization (e.g., a business unit's productivity rate) or externally on factors that relate to the organization's environment (e.g., economic conditions). Quantification refers to financial information (e.g., net profit margin) or non-financial information (e.g., employee absenteeism rate), the latter being expressed in non-monetary terms (Mia, 1993). Time horizon refers to historical, ex post information (e.g., last year's profit), or to future-oriented, ex ante information (e.g., probability estimates of a future project outcome). These three information characteristics define the AIS scope construct. Traditionally, AIS design has been limited to financial information internal to the organization with an historical time horizon. A narrow-scope AIS therefore consists of internal, financial, and historical information only. On the other hand, in addition to internalfinancialhistorical information, a broad-scope AIS consists of external, non-financial, and future-oriented information. The MAS scope construct has been used in several accounting studies (Gul and Chia, 1994; Chong and Chong, 1997; Ismail and King, 2005), among others. 2.1.3. Business unit performance We chose business unit performance as the criterion variable in examining the fit between business strategy and AIS scope. To justify this unit of analysis, we took into account that the Miles and Snow's typology has been designed to study business units. In addition, the level of responsibility at which the AIS design normally resides is at the business unit. While we recognize that business unit performance may be a multi-dimensional construct, we focus here on its financial dimension. 2.2. Hypotheses development This section covers the developments of hypotheses along with the research model as illustrated in Fig. 1. 2.2.1. Match between prospector strategic-type and AIS scope Theoretical propositions can be developed for prospectors based on their entrepreneurial, administrative, and engineering problems (Miles and Snow, 1994). Prospectors deal with a
Miles and Snow (1978, 1994) also identified a dysfunctional type, the reactor, which illustrates a lack of coherent arrangement. According to Daft and Weick (1984), the reactor is not really a strategy.
3

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

Fig. 1. Research model: Hypothesized relationship between business strategy, AIS scope, and business unit performance.

dynamic environment and offer many products and services through monitoring the environment. Since solutions to entrepreneurial problems reside in broad monitoring of the external environment, managing the business place is considered critical for decision-making. Gordon and Miller (1976) have suggested that in a dynamic environment, accounting systems should include more non-financial information and more forecasts. With regard to administrative and engineering problems, where solutions reside in the development of broad-based information systems and a low degree of products routinization, Simons (1987, 1990) finds that prospectors appear aggressive in their scanning of competitors, need non-financial information, and incorporate forecast information in their reports. Consistent with this, Abernethy and Guthrie (1994) find that broad scope AIS has a more positive impact on business unit performance for prospector than for defender units. As a result, external, non-financial and future-oriented information, which corresponds to broad-scope information, is considered more appropriate for managers of prospector-type businesses. Based on the previous discussion, the first hypothesis replicates prior studies and is stated as follows: H1. Business units with a prospector/broad-scope configuration have a higher performance than units with a prospector/narrow-scope configuration. 2.2.2. Match between defender strategic-type and AIS scope There are two competing literatures that deal with the defender/AIS scope relationship which leads to a defender/narrow-scope link and a defender/broad-scope link. On one hand, prior research indicates that defenders sustain a stable environment. This tends to be compatible with reliance on historical information, since typical accounting systems provide historical financial information and deal with stable production processes (Govindarajan, 1988; Brownell and Merchant, 1990). The defenders' focus on cost supports the importance of financial information for decision-making. Defenders also propose few products, which may reduce the need to scan the external environment (Simons, 1990). Historical, financial, and internal information, corresponding to a narrow AIS scope, appears to be suitable for managers of defenders, which suggests a defender/narrow-scope link.

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

On the other hand, with the advent of developments in accounting such as benchmarking and activity-based management, among others, recent research suggests that performance may improve for defender units that adopt information systems that include broad-scope information. For example, Miles and Snow (1994) report that defenders now make more use of state-of-the-art information systems. More specifically, since ERP systems may integrate financial, non-financial, internal and external information, Banker et al. (2001) found that units that follow a cost strategy, which corresponds to defender strategic-type,4 are more likely to use enterprise resource planning (ERP) systems. ERP systems are therefore more attractive to units that compete on cost (Banker et al., 2001). Nicolaou and Bhattacharya (2006) report an association between ERP-adopting firms and superior performance. Based on recent research, non-financial and external information appears to be appropriate for managers of defender-type businesses, suggesting a defender/broad-scope link. Studies suggesting a defender/narrow-scope association were conducted prior to the 1990s, before the advent of new management accounting techniques. Implementation of sophisticated low-cost information systems permits analyses that were not previously accessible. The use of information technology to support accounting processes may render broad-scope AIS more valuable to defenders. Although the current business environment encourages the use of broad-scope AIS, defendertype units have been identified as slow to adopt changes such as the introduction of new information systems. Simons (1987) and Gosselin (1997) have found that prospectors are more open than defenders to the adoption of new accounting techniques. In short, defenders play it safe in introducing only changes that have demonstrated clear success elsewhere. We should therefore observe a higher performance for the defender/broad-scope configuration, reflecting the benefits of using broad-scope AIS. Based on the previous discussion, we state the second hypothesis as follows: H2. Business units with a defender/broad-scope configuration have a higher performance than units with a defender/narrow-scope configuration. 2.2.3. Analyzer strategic-type and AIS scope According to Miles and Snow, attaining adequate control systems for the analyzer is more complex than for defender and prospector strategic-types. To achieve higher performance, managers of analyzers have to deal with different control systems. Miles and Snow comment on the analyzers as follow: in functional units, control systems are centralized and budget-oriented to encourage cost-efficient production (but) in product and project groups, control systems are decentralized and results-oriented so as to enhance effectiveness (1978, pp. 7677). The use of management control systems by analyzers seems therefore to be more contextual, which explains why their information needs are more difficult to describe than those of defenders and prospectors (Chong and Chong, 1997). Sim and Teoh (1997) point out that for analyzers, cost efficiencies are highlighted in established units, whereas innovations are supported in newer markets. Therefore, it appears that organizational life cycle stages may matter for analyzers. For example, Moores and Yuen (2001) find that units with products at the birth stage are associated with narrow information scope usage, whereas units with products at the growth stage are associated with broad-scope information usage. Analyzers deal with a mix of new and mature products and services and with efficiency and innovation objectives, suggesting a kind of dual information attention.
Some argue that defender and cost leadership strategic-types, from Miles and Snow (1978) and Porter (1985) respectively, illustrated similar characteristics. For a discussion on this correspondence, see Chrisman et al. (1988).
4

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

To our knowledge, no other study has examined the strategy/AIS scope link with the analyzer-type and, in the absence of literature, it is difficult to favor a particular configuration. In short, information needs for analyzer's managers are expected to be different from those of defenders and prospectors. Based on the previous discussion, we state the following two hypotheses: H3. Business unit performance differs between analyzer/broad-scope and analyzer/narrowscope configurations. H4. Analyzers' information needs are different from those of defenders and prospectors. 3. Research methodology Some have criticized prior research that used a contingency approach, suggesting that researchers were neither cautious nor coherent when using the notion of fit (see Venkatraman, 1989; Gerdin and Greve, 2004). Different perspectives of fit reflect different theoretical interpretations and require different statistical tests. Since the research interest is defined as identifying the configurations of business strategy/AIS scope associated with higher performance, fit-as-matching is the underlying theoretical representation and analysis of variance is the corresponding statistical test (Venkatraman, 1989). 3.1. Sample Survey research with managers and secondary data analysis with a financial database were employed. The plan to use a multi-method approach in order to improve construct validity required collection of various types of data. As managers are reluctant to permit disclosure of objective data (see Chenhall, 1997),5 we worked in relationship with a professional accounting association to endorse the survey and increase participation rate.6 To reinforce generalization, we used the association's member directory to pre-select a set of units from both manufacturing and service industries. To select the final sample, we first contacted members by telephone and asked whether they were organized as a business unit, which is the level of responsibility at which the information system normally resides. As business units with less than 100 employees are recognized as unlikely to have clearly attributed fields of responsibilities (Brownell and Dunk, 1991), we targeted units of 100 employees or more. We explained the nature of the study to those units that fulfilled these criteria, and sent two questionnaires to the units that agreed to participate. The first questionnaire asked for financial and non-financial data, and was to be answered by the person who occupied the highest position in finance; some of these data were collected to obtain a relative ranking through a financial database. The second questionnaire was designed to collect subjective assessments on performance and to categorize strategy and information scope. Since managers are generally most familiar with these variables, this second questionnaire was to be
5 For example, Chenhall states: In the first instance it was planned to collect objective data on divisional profitability such as growth in sales and return on assets. However, this was unsuccessful as many of the chief managers were reluctant to permit the disclosure of such data. As an alternative, perceived measures of performance were used (1997, p. 196). 6 The professional association is the Certified General Accountants (CGA) of Canada, where CGA-Quebec is the association located in the province of Quebec, Canada, with more than 8000 members. CGA-Quebec is affiliated with CGA-Canada, which brings together 60,000 members and students.

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

answered by the person who occupied the highest management position. The questionnaires were reviewed for clarity.7 Initially, 500 firms were contacted by telephone, and 380 agreed to participate in the study and received questionnaires. Both questionnaires were required to be fully completed from the same business unit to count as one observation. To reinforce the importance of participation, three telephone reminders were conducted at intervals of 2 weeks, 4 weeks, and 6 weeks. Questionnaire responses were received from 128 units. Responses from 38 of these units were eliminated either because the questionnaires were incomplete, or because they were completed by only one of the two required respondents. The sample thus consisted of 90 pairs of usable questionnaires, for a response rate of 24%, a rate similar to other studies that have used multiple respondents.8 The main reasons mentioned for non-participation in the study were confidentiality concerns and the requirement of two respondents per unit. The highest position in finance respondent profile was a comptroller who held a bachelor's degree in commerce with an accounting designation, and had an average age of 42 years. The highest management position respondent profile was a general manager who held a bachelor's degree in commerce, and had an average age of 45 years; they had held their current positions for an average of 8 years and had been employed by the same firm for an average of 14 years. At the business-unit level, the average number of employees was 156, with average revenues of 22 million Canadian dollars. The sample of business units consisted of 48 manufacturing (53%) and 42 service (47%).9 Including all variables, a t-test on manufacturing vs. service showed no significant differences between the groups. We tested the data for normality, outliers, and non-response bias. Two outliers were identified and deleted, resulting in a sample of 88 to analyze. T-tests on late versus early respondents suggested that non-respondents bias was not present. 3.2. AIS scope To assess AIS scope the six-item instrument developed by Chenhall and Morris (1986) was used which has also been frequently used in previous accounting research. Respondents were asked to indicate the extent to which information items are used for decision-making in their unit. In the present study, information items reported used by managers were available through the AIS. A seven-point Likert scale ranging from very seldom used to very often used was employed. The six items, two questions for each of the three information characteristics focus, quantification, and time horizon were summed to assess the scope of information; the higher the score, the broader the AIS scope. To categorize broad-scope and narrow-scope groups, we sorted the scores in descending order for each business unit, then created two groups based on the median; those above the median were defined as broad-scope (n = 46) and those below the median were defined as narrow-scope (n = 42), converting a quantitative variable to a categorical one.10 The Cronbach coefficient alpha for the instrument is 0.65, and a factor analysis confirms the unidimensional nature of the construct.
Two academics and an adviser in linguistics reviewed the questionnaires. When using a multiple respondent approach, the response rate decreases accordingly. For example, Chan et al. (1997) and Sabherwal and Chan (2001), who used multiple respondents, obtained response rates of 19% and 7% respectively. 9 The sample of 90 business units is similar in size and industry to the 500 contacted. The units were in pulp and paper, textile, transformation, construction, industrial products, food products, retailer, wholesaler, leasing and dealers. 10 With 88 observations, one would expect each group to have 44 observations, but some observations in the middle had the same score; we then split the group at the following observation.
8 7

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

3.3. Business strategy Strategy was assessed with Miles and Snow's (1978, 1994) typology based on Snow and Hrebiniak's (1980) definitions. This typology has been widely used in research on various industries and has shown reliability and validity (Shortell and Zajac, 1990; Doty et al., 1993). The management respondents were presented with a short description of a defender, a prospector, and an analyzer unit, and asked to choose which description most closely fit their business units as compared to other units in the industry. The 88 business units were classified as follows: 26 defenders (30%), 23 prospectors (26%) and 39 analyzers (44%). This higher percentage of analyzers is consistent with previous research (Collins et al., 1997; Sabherwal and Chan, 2001), among others. 3.3.1. Examining convergent validity for the business strategy construct To provide support to the business strategy classification obtained above, we used objective indicators as employed in previous studies (see Thomas et al., 1991; Ittner et al., 1997). Indicators used were: the ratio of marketing expenses to revenues; the number of products offered; and the number of new products introduced over the 3 three years. As prospector units are more market focused and externally oriented than defender units, prospectors were expected to score higher on marketing expenses to revenues ratio than defenders. Prospectors have constantly growing areas and offer many products, whereas defenders focus on a stable environment and propose few products, which suggests that prospectors would have the greater number of products on offer. Finally, prospectors have specific aptitude in product research, are innovators, and are estimated to have a higher number of new products introduced than defenders. From the information obtained from respondents, we calculated the three ratios discussed above and performed t-tests to determine the expected differences. Since no assumptions exist for analyzers in regard to these three ratios, we did not include that strategictype in this analysis. As indicated in Table 1, results of t-tests for each indicator show statistically significant differences between the two strategic-types. Prospectors have higher scores for all three indicators, supporting the classification obtained with the questionnaires and thereby improving the validity of the business strategy variable. 3.4. Business unit performance Business unit performance is represented by three indicators return on assets, net profit margin, and revenue growth.11 These indicators are commonly used and recognized as key dimensions of performance. Management respondents were asked to classify their business units' performance compared with that of their peers, using a 4-point scale. We thus obtained an assessment ranging from 4 (best) to 1 (worst) for each indicator. To corroborate the management's assessments, the position in finance respondents were asked to provide financial data with which to calculate the three indicators, annual revenue for size classification, and the SIC code for industry classification. Performance was then estimated relative to size and industry. The calculated financial indicators were then compared to a database in which the data are organized by quartile (Financial Performance Indicators for Canadian
We used the following definition: Return on assets = (net profit + interest expense) / total assets, Net profit margin = net profit / total revenue, and Revenue growth = (sales current year sales previous year) / sales previous year.
11

10

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

Table 1 Results of t-test of each indicator supporting the business strategy's classification Indicator Marketing expenses/revenues Number of products Number of new products Strategic-type Defender Prospector Defender Prospector Defender Prospector N 26 23 26 23 26 23 Mean 0.014 0.031 8.27 17.78 5.46 13.27 Standard deviation 0.012 0.077 17.32 22.93 7.08 14.14 Standard error mean 0.002 0.015 3.40 5.41 1.39 2.95 F 4.843 3.962 4.098 Sig. .033 .049 .053

Business from Statistics Canada). We classified the relative position of the three indicators for each business unit and obtained an assessment that also ranged from 4 (best) to 1 (worst). For example, a business unit that reported $25 million of revenue, an SIC code of 1234, and a return on assets (ROA) of 5%, would be assigned a score of 3 if the database shows that, for this specific size and industry, a ROA ranging from 0% to 10% corresponds to quartile 3. Both strengths and weaknesses of methods and data sources are considered when using a multi-method approach. Furthermore, requiring two respondents per business unit reduces the threat of functional and response bias (Huber and Power, 1985). 3.4.1. Examining convergent validity for business unit performance construct To support the business unit performance variable we used a multi-method approach based on different measurements, as we did for the business strategy variable. Convergent validity is present when the correlations are positive and significant. For the three indicators (return on assets, net profit margin, and revenue growth), the correlations between the two assessments, subjective and financial data, are respectively +0.50, p b 0.01; + 0.38, p b 0.01; and + 0.39,

Table 2 Descriptive statistics of variables Mean S.D. Min Max Theoretical range

Panel A: Subjective assessment obtained from managers AIS scope 26.3 5.5 12 42 642 Business unit performance: Return on assets (ROA) 3.06 0.98 1 4 14 Net profit margin (NPM) 3.02 0.93 1 4 14 Revenue growth (RevGrowth) 2.90 1.03 1 4 14 Performance (ROA + NPM + RevGrowth) 8.9 2.6 3 12 312 Business strategy (categorical variable): Defender, n = 26; Prospector, n = 23; Analyzer, n = 39. Total n = 88 Panel B: Financial data obtained from position in finance respondents Return on assets (ROA) 9.53 6.59 Net profit margin (NPM) 5.13 5.39 Revenue growth (RevGrowth) 6.25 17.37 Panel C: Relative positioning found with the financial database Return on assets (ROA) 2.78 0.79 Net profit margin (NPM) 3.05 0.80 Revenue growth (RevGrowth) 1.46 0.50 ROA + NPM + RevGrowth 7.31 1.54

4.66 2.67 38.5

31.39 34.5 71.2

n.a. n.a. n.a.

1 1 1 3

4 4 4 12

14 14 14 312

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116 Table 3 Correlation matrix ROA ROA NPM RevGrowth Performance Scope (categ.) Scope (conti.) 1.000 0.898 0.536 0.917 0.182 0.207 NPM 1.000 0.561 0.925 0.181 0.216 RevGrowth Performance Scope (categ.)

11

Scope (conti.)

1.000 0.801 0.106 0.168

1.000 0.177 0.223

10.000 0.811

1.000

Pearson correlation is significant at the 0.01 level, significant at the 0.05 level, (two-tailed), n = 88. ROA = Return on assets; NPM = Net profit margin; RevGrowth = Revenue growth; Performance = ROA + NPM + RevGrowth; all indicators are subjective assessments. Scope (categ.) = AIS scope as a categorical variable i.e., broad-scope vs. narrow-scope; Scope (conti.) = AIS scope as a continuous variable.

p b 0.01, which indicates convergent validity of methods. These results are similar to those of Dess and Robinson (1984) and Venkatraman and Ramanujam (1987). With the multi-method, multiple respondent approach, our objective was to increase our confidence in business strategy and business unit performance primary measures with other sources, and the results suggest that we obtained convergent validity for both measures. The three indicators return on assets, net profit margin, and revenue growth were summed to form the performance variable, with a Cronbach alpha of 0.85. Table 2 provides descriptive statistics of variables and measures, whereas Table 3 provides the correlation matrix for the variables of interest. As anticipated, the three indicators that form the performance variable return on assets, net profit margin, and revenue growth are positively and significantly correlated at the 0.01 level. The strategy variable is not included in the matrix because strategy is a categorical variable in the present study. 4. Results We used ANOVA to test the hypotheses. The three strategy types prospector (n = 23), defender (n = 26), and analyzer (n = 39) and the two AIS scope categories narrow-scope (n = 42) and broad-scope (n = 46) results in a 3X2 matrix. The sample size of 88 exceeds the minimum number of observations required for this type of analysis.12 The mean performances for the six configurations are reported in Table 4, Panel A, whereas ANOVA results are presented in Table 4, Panel B. Results are statistically significant, suggesting that performance depends on a match between business strategy and AIS scope. 4.1. Hypothesis 1: Match between prospector strategic-type and AIS scope Our first hypothesis is that units with a prospector/broad-scope configuration have a higher business unit performance than units with a prospector/narrow-scope configuration. Prospector/
First, the sample size should be 10 times the number of items present in the largest construct (Tabachnick and Fidell, 1989), and in our study the AIS scope is the largest construct with six items, so the minimum sample is 60 (10 6). Second, for each independent variable there must be at least 30 respondents (Pedhazur, 1982). In our study, we have two independent variables, so the minimum sample required is 60. Studies that investigated the business strategy/AIS link had the following sample size: Simons (1987), n = 76; Abernethy and Guthrie (1994), n = 49; and Chong and Chong (1997), n = 62.
12

12

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

Table 4 ANOVA results for the relationship between business strategy,AIS scope, and business unit performance Panel A: Mean of the configuration's performance Prospector Narrow-scope Broad-scope Mean = 8.00; S.D. = 3.38 n = 8 Cell 1 Cell 2 Mean = 10.40; S.D. = 1.45 n = 15 Defender Mean = 7.64; S.D. = 2.56 n = 14 Cell 3 Cell 4 Mean = 9.08; S.D. = 2.07 n = 12 Analyzer Mean = 9.15; S.D. = 3.05 n = 20 Cell 5 Cell 6 Mean = 8.74; S.D. = 2.40 n = 19

Mean = average of the configuration's performance; Min. = 3, Max. = 12 (theoretical and actual range); S.D. = standard deviation; n = # of unit; n total = 88. significance levels are based on a two-tailed test. Post Hoc tests (LSD): H1: Cell 2 N Cell 1, statistically significant (0.033); H2: Cell 4 N Cell 3, not statistically significant (0.151); at one tailed, significant at 0.08; H3: Cell 6 Cell 5, not statistically significant (0.611); H4: Cell 5 N Cell 3, statistically significant (0.090). Panel B: ANOVA results Main effects Tests of between-subjects effects; dependent variable: Performance Source Corrected model Intercept Strategy Scope Error Total Corrected total Type III sum of squares 34.045 666.996 14.605 14.693 552.398 7589.000 586.443 df 2 1 1 1 85 88 87 Mean square 17.023 666.996 14.605 14.693 6.499 F 2.619 102.634 2.247 2.261 Sig. .079 .000 .138 .136

Computed using alpha = .05; R2 = .058 (Adjusted R2 = .036). Significance levels are based on a two-tailed test.

broad-scope configuration obtained a mean performance of 10.40, compared with 8.00 for prospector/narrow-scope configuration, leading us to support Hypothesis 1. Moreover, as reported in Table 4, this mean difference is statistically significant (sig. = 0.033). Our assumption was that broad-scope information would be essential to managers of prospector units and results suggest that success in prospector-type units requires external, non-financial, and future-oriented information. The results support Simons (1987) and Abernethy and Guthrie (1994) studies, among others. 4.2. Hypothesis 2: Match between defender strategic-type and AIS scope Our second hypothesis is that units with a defender/broad-scope configuration have a higher business unit performance than units with a defender/narrow-scope configuration. Defender/ broad-scope configuration obtained a mean performance of 9.08, compared with 7.64 for defender/narrow-scope configuration. Recent developments in management accounting techniques that assist the decision-making process, coupled with the use of sophisticated information systems may explain these results. Using a one-tailed significance level, having directional hypotheses, the mean performance between the two configurations is significant (sig. = 0.08). Given that we found two competing literatures dealing with defenders, we anticipated the outcome of a lower degree for the relationship. As defenders have been recognized as slower to introduce new information systems and to adopt new management accounting techniques

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

13

(Gosselin, 1997), results seem to indicate that a time lag may exist before a set of such units adopts these new systems and techniques (Miles and Snow, 1994). Such a set of defender units may move slowly to broad-scope information systems while continuing to use narrow-scope AIS, which is associated with weaker performance. The present results, with a time-lag perspective, are in line with Ittner, Larcker et al. (2003) who report that units using a broad set of financial and non-financial measures obtain higher market performance a forward-looking indicator while results with accounting measures (such as sales growth) show a weaker association. Accordingly, results suggest partial support for Hypothesis 2. 4.3. Hypothesis 3: Analyzer strategic-type and AIS scope Our third hypothesis is that business unit performance differs between analyzer/broad-scope and analyzer/narrow-scope configurations. The results show that mean performance for analyzer/ broad-scope is 8.74, whereas the mean performance for the analyzer/narrow-scope configuration is 9.15; the mean difference between the two groups is not statistically significant (sig. = 0.611). Managers of analyzers are reported as dealing with different control systems to achieve costefficiency, innovation, markets development, etc. These results support Miles and Snow's assumption that attaining adequate control systems for analyzers is more complex as performance does not differ between broad-scope and narrow-scope groups. 4.4. Hypothesis 4: Analyzers information needs are different Our fourth hypothesis is that analyzers information needs are different from those of defenders and prospectors. From the results of Table 4, Panel A, we observe that performance improves with broad-scope, except for the analyzer-type, where performance is slightly higher with narrowscope. For example, the results show that mean performance for analyzer/narrow-scope configuration is 9.15, the mean performance for defender/narrow-scope is 7.64, and the mean difference between these two groups is statistically significant (sig. = 0.9; two-tailed test). This supports the idea that analyzers' information needs are different from those of defenders. Sabherwal and Chan (2001) state that analyzers consider important information systems usage for monitoring and controlling operational efficiency, an information type associated with narrowscope. The above results suggest partial support for Hypothesis 4. 4.5. Additional analyses We performed additional analyses on our results. First, hypotheses were tested with AIS scope as a categorical variable (narrow vs. broad-scope categories); we also performed an ANOVA with AIS scope as a quantitative (continuous) variable and obtained better results (F = 3.419, sig. = .037, R2 = .074). Second, we found no significant difference in performance between defenders, analyzers, and prospectors, thereby supporting the idea of the three viable strategic-types (Zahra and Pearce, 1990). Finally, using revenues as a measure for size, we found no significant difference in size between the three strategic-type units. 5. Discussion, limitations, and conclusion This paper aimed to reexamine the relationship Strategy/AIS scope/Performance with the analyzer strategic-type and enhance the validity of variables in research. It is a core assumption in

14

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

accounting research that an AIS that provides relevant information to managers, in line with strategy pursued, is associated with higher performance (Merchant and Simons, 1986; Dehning and Richardson, 2002). The present study supports this relationship since results indicate that for prospector strategic-types broad-scope AIS is associated with higher performance. In other words, managers of prospectors need external, non-financial, and future-oriented information. For defender strategic-types, the results supported their recent evolution toward the usage of a broader set of information for decision-making. This current development explains why the defender/ broad-scope relationship is actually weaker than the prospector/broad-scope. For analyzers, results supported the idea that their information needs are different from those of defenders and more complex. Furthermore, we increased the validity of our results by using a multi-method approach. As previous studies have called for triangulation of variables with other sources, we utilized different assessments to support our primary measures of business strategy and performance. Also, we used multiple respondents in the data collection, an approach that reduces the threat of response bias compared to the use of a single respondent. From the three variables examined in the present study, only AIS scope was not triangulated. In the future, the information scope classification could be strengthened by asking respondents to report on the types of information systems that are used in their units. (For example, the presence of sophisticated devices such as executive information systems corresponds to broad-scope AIS.) Moreover, other constructs such as organizational life-cycle stages should be included in research models to explore further contingency relationships. The present study has two main limitations. First, the sample used to test our hypotheses is small; a larger sample would increase the statistical validity and add to the generalizability of results. Second, performance is a multi-dimensional construct and we covered only the financial perspective. Despite these limitations, this study makes a contribution in revisiting the relationship between AIS design and strategic choices with the analyzer strategic-type, and a methodological contribution in using a multi-method, multiple respondent approach. For many years, accountants have been involved in the analysis and design of information systems for decision-making. With the advent of integrated AIS, the function of information producer for accountants is more challenging; they must closely examine the match between the business strategy pursued and the information scope provided by the AIS. If mismatches are identified, accountants should actively participate in the design of AISs. Acknowledgement This paper is based on my dissertation completed at HEC-Montreal, Canada. I thank the members of my dissertation committee, Suzanne Rivard, Michel Vezina, Claude Laurin and Michel Guindon from HEC-Montreal, and Alain Pinsonneault from McGill University. I acknowledge the comments of participants at the 2006 ECAIS Conference held in Dublin, Ireland, from Professors Uday S. Murthy and Michel Magnan, and anonymous reviewers of IJAIS. Thanks to CGA-Canada, CGA-Quebec, SAP-CAAA Research Grant Program, and the Lawrence Bloomberg Chair in Accountancy at Concordia University for their financial support. References
Abernethy MA, Guthrie CH. An empirical assessment of the fit between strategy and management information system design. Account Finance 1994;34(2):4966.

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

15

Banker RD, Janakiraman SN, Konstans C, Slaughter SA. Economic drivers of decision to use ERP systems: an empirical investigation. Working paper presented at the 2001 AAA Annual Conference Meeting held in Atlanta, USA, August; 2001. Banker RD, Chang H, Pizzini M. The balanced scorecard: judgmental effects of performance measures linked to strategy. Account Rev 2004;79(1):123. Brownell P. Research methods in management accounting. Coopers and Lybrand accounting research methodology, Monograph 2, Australia and New Zealand; 1995. p. 3. Brownell P, Dunk AS. Task uncertainty and its interaction with budgetary participation and budget emphasis: some methodological issues and empirical investigation. Account Organ Soc 1991;16(8):693703. Brownell P, Merchant KA. The budgetary and performance influences of product standardization and manufacturing process automation. J Account Res 1990;28(2):38897. Chan YE, Huff SL, Barcaly DW, Copeland DG. Business Strategic Orientation, Information Systems Strategic Orientation, and Strategic Alignment. Inf Syst Res 1997;8(2):12550. Chenhall RH. Management control systems design within its organizational context: findings from contingency-based research and directions for the future. Account Organ Soc 2003;28:12768. Chenhall RH. Reliance on manufacturing performance measures, total quality management and organizational performance. Manag Account Res 1997(8):187206. Chenhall RH, Langfield-Smith K. The relationship between strategic priorities, management techniques and management accounting: an empirical investigation using a system approach, accounting. Organ Soc 1998;23(3):24364. Chenhall RH, Morris D. The impact of structure, environment, and interdependence on the perceived usefulness of management accounting systems. Account Rev 1986;61:1635. Chong VK, Chong KM. Strategic choices, environmental uncertainty and SBU performance: a note on the intervening role of management accounting systems. Account Bus Res 1997;27(4):26876. Chrisman JJ, Hofer CW, Boulton WR. Toward a system for classifying business strategies. Acad Manage Rev 1988;13(3): 41328. Collins F, Holzmann O, Mendoza R. Strategy, budgeting, and crisis in Latin America. Account Organ Soc 1997;22(7): 66989. Daft RL, Weick KE. Toward a model of organizations as interpretation systems. Acad Manage Rev 1984;9(2):28495. Dehning B, Richardson VJ. Returns on investments in information technology: a research synthesis. J Inf Syst 2002;16(1): 730. Dess GG, Robinson RB. Measuring organizational performance in the absence of objective measures: the case of the privately held firm and conglomerate business unit. Strateg Manage J 1984;5:26573. Doty DH, Glick WH, Huber GP. Fit, equifinality, and organizational effectiveness: a test of two configurational theories. Acad Manage J 1993;36(6):1196250. Financial Performance Indicators for Canadian Business, Statistics Canada, last edition. Gerdin J, Greve J. Forms of contingency fit in management accounting research: a critical review. Account Organ Soc 2004;29(3/4):30326. Gordon L, Miller D. A contingency framework for the design of accounting information systems. Account Organ Soc 1976:5976. Gosselin M. The effect of strategy and organizational structure on the adoption and implementation of activity-based costing. Account Organ Soc 1997;22(2):10522. Govindarajan VJ. A contingency approach to strategy implementation at the SBU level. Acad Manage J 1988;31(4): 82853. Govindarajan VJ, Gupta AK. Linking control systems to business unit strategy: impact on performance. Account Organ Soc 1985;10(1):5166. Gul FA. The effects of management accounting systems and environmental uncertainty on small business managers' performance. Account Bus Res 1991;22(85):5761. Gul FA, Chia YM. The effects of management accounting systems, perceived environmental uncertainty and decentralization on managerial performance: a test of three-way interaction. Account Organ Soc 1994;19(4/5):41326. Huber G, Power D. Retrospective reports of strategic-level managers. Strateg Manage J 1985;6:17180. Institute of Management Accountants, 1995. Developing Comprehensive Performance Indicators, Statement no. 4U, March, 57 pages. Ismail NA, King M. Firm performance and AIS alignment in Malaysian SMEs. Int J Account Inf Syst 2005;6:24159. Ittner CD, Larcker DF, Rajan MV. The choice of performance measures in annual bonus contracts. Account Rev 1997;72(2):23155. Ittner CD, Larcker DF, Randal T. Performance implications of strategic performance measurement in financial services firms. Account Organ Soc 2003;28:71541.

16

E. Boulianne / International Journal of Accounting Information Systems 8 (2007) 116

Langfield-Smith K. Management controls systems and strategy: a critical review. Account Organ Soc 1997;22(2):20732. Merchant KA, Simons R. Research on control in complex organizations: an overview. J Account 1986:183203. Mia L. The role of MAS information in organizations: an empirical study. Br Account Rev 1993;25:26985. Miles RE, Snow CC. Organizational strategy, structure and process. New York: McGraw Hill; 1978. Miles RE, Snow CC. Fit, failure, and the hall of fame. New York: The Free Press; 1994. Mintzberg H. Strategy making in three modes. Calif Manage Rev 1973:4453. Moores K, Yuen S. Management accounting systems and organizational configuration: a life-cycle perspective. Account Organ Soc 2001;26:35189. Nicolaou AI, Bhattacharya S. Organizational performance effects of ERP systems usage: the impact of postimplementation changes. Int J Account Inf Syst 2006;7:1835. Pedhazur EJ. Multiple regression in behavioral research. NY: Holt, Rhinehart and Winston; 1982. Porter ME. Competitive advantage: creating and sustaining superior performance. New York: The Free Press; 1985. Sabherwal R, Chan YE. Alignment between business and IS strategies: a study of prospectors, analyzers, and defenders. Inf Syst Res 2001;12(1):1133. Shortell S, Zajac E. Perceptual and archival measures of miles and snow's strategic types: assessment of reliability and validity. Acad Manage J 1990;33(4):81732. Sim AB, Teoh HY. Relationship between business strategy, environment and controls: a three country study. J Appl Bus Res 1997;13(4):5773. Simons R. Accounting, control systems and business strategy: an empirical analysis. Account Organ Soc 1987;12(4): 35774. Simons R. The role of management control systems in creating competitive advantage: new perspectives. Account Organ Soc 1990;15(1/2):12743. Snow CC, Hrebiniak LG. Strategy, distinctive competence, and organizational performance. Adm Sci Q 1980(25):31736. Tabachnick BG, Fidell LS. Using multivariate statistics. 2nd edition. 1989. New York. Thomas AS, Litschert RJ, Ramaswamy K. The performance impact of strategy manager coalignment: an empirical examination. Strateg Manage J 1991;12:50922. Venkatraman N. The concept of fit in strategy research: toward verbal and statistical correspondence. Acad Manage Rev 1989;14(3):42344. Venkatraman N, Ramanujam V. Measurement of business economic performance: an examination of method convergence. J Manage 1987:10922 (Spring). Young SM. Survey research in management accounting: a critical assessment. CGA Canada Research Foundation, Research methods in accounting: issues and debates, research monograph #25, vol. 3. 1996. p. 5568. Chapter. Zahra SA, Pearce II JA. Research evidence on the MilesSnow typology. J Manage 1990;16(4):75168.