The Late Show: The Effects of AfterHours Accessibility on the Performance of Small Retailers...

How should small retailers best leverage their limited resources for maximum effect? In this article, we address this question as it concerns small store hours of operation. Drawing on niche theory, we advance a strategy of extraordinary accessibility as a source of competitive advantage uniquely

available to small firms. We then report results from our study of the hours of operations of over 300 small community hardware stores. We conclude by reflecting on what our findings mean for small retailers facing the prospect of competing alongside giant scale-advantaged firms.

A new era of large retailers leaves the small retailer increasingly disadvantaged in virtually every facet of operation. Whether it be advertising budgets, inventory reserves, or staff sizes, large firms are predisposed to glean any and every advantage of scale. This means small retailers must act more strategically than ever before.

One decision area concerns the allocation of the scarce resource of management time. As numerous studies report, management time is arguably the scarce resource since it makes possible all other decisions and tactics. However, a retail correlate of management time--store hours of operation--has not received much attention from either academics or practitioners. This lacuna suggests an implicit two-fold assumption: (1) store hours of operation are essentially a given; (2) store hours do not effectively differentiate one competitor from another.

What this means in actual practice is straightforward. If a competitor increases the number of hours or number of days it is open for business, rivals are assumed to have no choice but to follow suit. However, given the resource disparity between small and large retailers, accepting this assumption can be dangerous for small firms-extended hours may not necessarily generate sufficient sales to offset the incremental expenses incurred.

Another tactic appears available, however. Rather than trying to match the operating hours of its large retailer competition, the small firm, by virtue of its more informal mode of operation, can instead make itself available to customers for after-hour emergencies. This strategy, built on a logic of scarcity, involves being informally accessible to customers if and when emergency strikes, regardless of whether the store is formally closed. Compared to the logic of scale and scope, a scarcity-based strategy seeks to complement, rather than duplicate, large competitor practice.

The option of scarcity-motivated hours of operation poses three interesting questions not usually considered by small business researchers. First, do small retailers utilize such access strategies? Second, if implemented, are these strategies utilized in place of, or together with, more traditional scale- and scope-based policies? A third question concerns retail performance: do hours of operation, whether based on scale, scope, or scarcity strategies, make any discernible difference to small firm performance? To date these questions have been overlooked. While the practice of scope-related practices (Sunday shopping) was studied by Barnes and Chopoorian (1987), the practice of tactics based on scale (evening) and scarcity (after-hour) have not been considered. Our article therefore seeks to address these questions for two reasons. First, we seek to assist small retail business in most effectively allocating its limited resources. Second, on a more theoretical level, we desire to advance understanding concer ning the nature of effective small retailer niche strategy.

We approach the task in five steps. First, we discuss the nature of the large firm's comparative advantages over the small firm. Second, we review the basic range of response options available to small players. Third, we integrate these response options with insights from niche theory to advance a set of performance-related hypotheses. Fourth, we report results from our recent study on small firms and their policies regarding hours of operation. We conclude with a discussion of the implications of our findings for small retailers seeking to realize a sustainable competitive advantage in the presences of large retailers.

Literature Review

Scalar Advantage and Big Box Operations: Demise of the Small Retailer?

The past decade has witnessed the entry and increasing dominance of a new breed of "big box" retailer. As Ehrenfeld (1995) notes, in retail contexts as diverse as home improvement, toys, stationary, sporting goods, and funeral services, "category killers" gain dominant market share, often at the expense of small retailers. The reasons for this new breed of giant's success are easily discerned. The logic of efficiency seeks to reduce transaction costs wherever and however possible (Williamson 1975). One route to increased cost savings, and hence to lower cost advantage (Porter 1980), is by increasing transaction lot sizes, and thereby realizing reductions in cost per unit.

However, not all firms are equally predisposed to reap the benefits of pairing increased scale with decreased cost per unit. Compared to smaller competitors, large firms enjoy a differential advantage in procuring and distributing large quantities of goods. Furthermore, such advantage is not confined to inventory flows. Concentration of buying power partnered with preferential advertising and facility rental rates and large-scale human resource capabilities creates a formidable adversary. Taken together, these value chain components (Porter 1985) set in motion a volume-fueled retail steamroller. Given this resource assemblage, the large firm's logic of extending the scale and scope of store hours becomes clearer--scale-based inflow needs to be paired with scale-based outflows. Such interdependent scalar advantage also prompts observers such as Ehrenfeld (1995) to ask whether this new breed of giants spells the end of the traditional "Mom 'n Pop" operation.

The Small Retailer's Response Options: In-Kind, Not-in-Kind, and Evasion

We propose that competitors have three response options given a rival's strategic offensive. Two options involve counterattack; the third, nonresponse. Non-response simply means not countering the initiator's parry with any type of counter-thrust. Counterattack can be bifurcated into qualitatively similar and dissimilar responses. Qualitatively similar responses involve responding "in-kind" (with "kind" defined by the attacker).

A ready example is price cutting. Nonresponse consists of not countering the cut in any way. In-kind response involves matching a competitor's price cut with an equivalent or greater cut. In contrast, not-in-kind response involves countering in one or more ways that are qualitatively dissimilar to the initiator's thrust. In the case of a price cut, a "not-in-kind" response might mean emphasizing firm-specific after-sale service capabilities to justify a premium price.

In the case of a firm increasing the scale and/or scope of operating hours, nonresponse means not implementing any scale- or scope-related increase in operating hours. A qualitatively similar response involves matching (or bettering) the hours of operation. A qualitatively dissimilar response involves maintaining current formal hours of operation, but also initiating some value-added service which compensates or offsets the competitive move of the rival. One such policy might involve being accessible for customers' late-night emergency needs.

An important strategic question arises from this discussion: Is performance related to the choice of competitive response? Furthermore, given the possibility of implementing both qualitative and quantitative tactics, is performance enhanced by implementing both tactics in tandem? Seminal work on the nature of the niche in competition provides several illuminating insights into these questions. To these insights, and their implications for small firms, we now turn our attention.

Response Options Reconsidered: Insights from Niche Theory

In 1934, ecological researcher E. F. Gause articulated the principle of mutual exclusion. Simply stated, this axiom proposes that two species with similar ecological demands cannot simultaneously exist in the same region. The principle expanded Darwin's seminal work on evolutionary process in which he concluded "competition will generally be most severe between those forms which are most nearly related to each other in habits, constitution and structure" (1909, p. 114). Hardin (1960, p. 1294) summarized the underlying logic for Gause's hypothesis in this way:

If two species...occur together in the same habitat in -the same region, eat the same types of food and have the same other ecological requirements, then they should compete with each other, and since the chance of being equally well adapted is negligible, one of them should eliminate the other completely.

Gause's observation implies that in order to survive, every organism needs to possess one or more differentiating characteristic that distinguishes it within its environment. In support of this assertion, Nicholson (1933) commented on the necessity of each species possessing a comparative interspecies advantage: "For the steady state [of coexistence between two or more species] to exist, each species must possess some advantage over all other species with respect to some one, or group, of the control factors to which it is subject."

One survival strategy that has been observed as a feasible response to the demands of mutual exclusion is the niche strategy Considered at its most general level, the niche is defined in the context of both (1) the role filled by a particular species, that is "the role of an organism in an ecological community involving...its way of life and its effect on the environment" (Gove 1986, p. 1525); and (2) the particular habitat the species occupies, that is "the distributional relation of a species to a range of environments and communities" (Whittaker, Levin, and Root 1973, p. 321). A role-based perspective concentrates on the functional needs served by the particular species in its setting, whereas a habitat-based perspective focuses on the broader systemic pattern of interdependent conditions (that is, food, shelter, and competition) under which a species is able to flourish.

Applying Niche Theory to the Organization: Work to Date

The niche concept is also relevant in the social sciences where it has been examined from the perspectives of both role and habitat. While organizational theory (Hannon and Freeman 1977; Aldrich 1979) has focused on the sustainability implications of different environmental habitats, works in strategy (Miles and Snow 1978; Porter 1980; Miller and Friesen 1984) and marketing (Henderson 1983; Linneman and Stanton 1991; Weinstein 1994) have centered on role-based target-marketing niches.

In their conceptualization of niche theory, Hannon and Freeman described the niche as "that area in constraint space...in which the population out-competes all other local populations" (Hannon and Freeman 1977, p. 947). By definition, then, a niche consists of "all those combinations of resource levels at which the population [italics added] can survive and reproduce itself" (Hannon and Freeman 1977, p. 947). Aldrich (1979) extended conceptual work on the niche by providing a fuller description of the dynamics of comparative firm behavior within, and between, different niche resource scenarios. In defining the niche as "an abstract resource space consisting of a unique combination of resources (information, access to materials, customers, and so on) that permit a firm [italics added] to survive there" (1979, p. 40), he extended the relevance of the niche concept to the individual firm. In the strategy literature, Miles and Snow (1978) and Miller and Friesen (1984) discussed the niche in terms of different co nfigurational gestalts that facilitate competitive advantage. The concept of the individual firm finding a niche was also articulated by Porter (1980) as the focus strategy alongside the two other generic strategies of low cost and differentiation.

In the marketing literature, a number of initiatives are devoted to understanding the challenges of niche-based marketing (Linneman and Stanton 1991; Raynor 1992; Dalgic and Leeuw 1993; Weinstein 1994). Linneman and Stanton (1991) proposed five distinct bases on which to establish a niche: product/service characteristics, customer-service, distribution channels, targeted communications, and price. Examples of each form of niche strategy are readily available. An example of niche creation through a unique product/service strategy is found in Kodak's premium line of color-print film (Ansberry 1988). A customer-service niche is illustrated by Hilton Hotels' special "day-camp" offering for children. Linneman and Stanton cite Walgreen's mail-order service for prescription drugs as an example of a distribution-channel niche strategy. A communication-centered niche strategy includes Audi's efforts at "lifestyle marketing" through the creation of sportenthusiast niches (La Flamme 1988). Finally, a price-based niche strategy is exemplified by cruise lines' use of deep discounting for customers booking fares just shortly before departure.

Applying Niche Theory to the Retail Arena: Night as Nocturnal Habitat

Niche theory is also relevant to the issue of how small shopkeepers respond to giant retailers. First, niche theory's emphasis on distinctiveness suggests that unwitting attempts by small competitors to match larger rivals' initiatives may preclude realizing one or more bases for inimitable differentiation. This renders the smaller entity only more vulnerable to Gause's theorem of mutual exclusion.

A second reason arises from the nocturnal transformation of the retail marketplace. While large retailers might extend the scale or scope of shopping hours, the vast majority still nonetheless close, with the specific time often dictated by head office policy. Such policies often also preclude individual store managers from re-opening the store in response to the emergency needs of individual customers.

However, customer need is no respecter of corporate policy. Emergencies are emergencies precisely because of their severe and unpredictable nature. However, their overall frequency may constitute but a minuscule percentage of overall customer demand. Hence, the marginal increase in profits from late night emergencies may not justify being open on a 24-hour, seven-day-a-week basis. However, regardless of overall sales volume, these emergency needs do occur. Therefore, given the ever-recurring reality of emergency need, and the concomitant reality of large retailers often being constrained from servicing such need, a short-lived window of retail opportunity opens each night, closing again the next morning.

The value of each sale during this time period can be estimated as the present value of the sales arising from extraordinary access plus the net present value of the stream of marginal increases in future sales from client good will plus gains attributable to increases in the store's reputation (Fombrun and Shanley 1990). However, each sale's marginal value is also a function of its marginal cost. This suggests that the small business owner has two options: either increase marginal sales and/or reduce marginal costs. Given the irregularity of such sales, extending formal hours to increase after-hour sales appears unwise. However, offering customers what we term extraordinary accessibility, perhaps by the store manager living in close proximity to the store and/or providing customers with his or her home telephone number, involves few costs but nonetheless facilitates rapid response to customer distress.

Hypotheses Implications of Restricted Hours of Operation

The previous discussion suggests that small retailers can realize a defensible niche by being extraordinarily accessible to customers. Besides this assertion, two other related hypotheses warrant mention. First, we propose that stores not implementing operating hour tactics of either scale, scope, or scarcity essentially forfeit opportunity to satisfy customer needs. Hence we propose that restricted hours of operation result in inferior performance compared to stores attempting more extended, expanded, and extraordinary accessibility. Thus:

[H.sub.1]: Small firm performance will be negatively related to restricted hours of operation.

Implications of Extraordinary Accessibility: Singular and Additive

Our second hypothesis arises from our distinction between in-kind and not-in-kind responses. By integrating this distinction with Darwin's assertion of competition being "most severe between those forms which are most nearly related to each other" (1909, p. 114), we predict that increases in the scale or scope of operating hours are comparatively limited in generating superior performance. At worst, an increase in formal hours of operation may do no more than incur additional operating costs. This assertion is supported by Klemperer and Padilla (1997). "Small shops," they observe,"cannot open on Sundays unless their owners either sacrifice their leisure time or hire part-time employees (which is both costly and risky for a small shop when the law makes layoffs expensive) while department stores can simply establish rota working systems"(p. 484).

A slightly more optimistic prospect sees undiscerning adoption of such practices only increasing sales volume by one-sixth (in the case of Sunday shopping). However, we hypothesize a different outcome linked to qualitatively dissimilar responses leveraging competitor scarcity. By making its services extraordinarily accessible, the small firm can potentially satisfy a time-specific and time-limited opportunity not easily addressed by larger, less flexible large entities. Therefore we hypothesize:

[H.sub.2]: Small firm performance will be positively related to extraordinary (late night) accessibility.

But what of those firms simultaneously deploying more than one of these strategies of scale, scope, and scarcity? The principle of additivity suggests an incremental benefit. Thus:

[H.sub.3]: Small firm performance will be highest for those firms that combine both extraordinary (late night) accessibility and conventional (Sunday and evening) hours of operation.

Summary: Performance Implications of Cloistering, Copying, and Complementing

In summary, each hours-of-operation policy carries with it an implicit performance correlate. Non-response to the giant competitor's offensive (that is, hours of operation characterized by neither scale, scope, nor scarcity strategies) suggests poor performance by reason of nonengagement. Response to the giant on the giant's terms suggests mediocre performance by reason of the small store's nondistinctiveness. However, nonconventional response, that is, being extraordinarily accessible to customers, suggests superior performance by reason of nonconformity. By virtue of its smaller size and more flexible mode of operation, the small retailer adds value for the customer by essentially "breaking curfew" and thereby satisfying time-specific needs not addressed by large rivals.

Methodology Industry Selection

We selected the retail hardware industry (SIC 5251) for testing these hypotheses for three reasons. First, the industry has a long history of fragmented competition. As Thompson (1992) notes, the $100 billion do-it-yourself market, of which retail hardware constitutes but one segment, consists of tens of thousands of small independent retail dealers. Given our earlier assertion of small firms being inherently advantaged in responding to customers' after-hour emergencies, it was important to select a retail context with a preponderance of small firms.

The second reason is based on the recent evolutionary changes this industry has undergone. Several industry observers (Miller 1992; National Home Center News 1992;Do-it-yourself Retailing 1994) document how the entry of several large chain players (such as Home Depot, Lowes, Builders Square, and Hechingers) has altered the competitive dynamics for many small incumbents. The presence of these giants provides an interesting research opportunity insofar as they facilitate study of whether the small firm's nocturnal flexibility can effectively differentiate it from larger competitors.

A final rationale for selecting this industry relates to the nature of the products and services offered by hardware stores. Compared to many other retail sectors, hardware stores feature many more products and services of an emergency-related nature. Whether it be an electric fuse or a bathroom plunger, the hardware store's wares are often needed on an emergency basis. While this aspect of the hardware retail business may limit the generalizabilty of our study, it nonetheless serves as a good starting point for understanding the importance of extraordinary accessibility as a source of small firm advantage.

Identification of Sample; Development and Pretesting of Research Instruments

We contacted 1,169 small hardware stores in the major U.S. metropolitan areas of Atlanta (GA), Chicago (IL), Kansas City (MO), Long Island (NY), Miami (FL), Minneapolis-St. Paul (MN), and San Diego (CA). Each store was asked to participate in a structured telephone interview and to complete a mail- administered survey designed and validated in accordance with Dillman's Total Design Method (Dillman 1978). The instruments were pretested with the assistance of Cotter Canada Hardware and Variety Cooperative Incorporated (the Canadian headquarters of the U.S. parent True Value trade name franchise buying group). A representative from Cotter Canada assisted by providing a list of ten hardware stores in the central Canadian province of Manitoba for a mail-administered pretest. The Cotter Canada representative also assisted by contacting each of the stores to request their advance commitment for the study. Of the ten stores, nine completed and promptly returned the survey instrument.

The Cotter Canada representative also assisted by assessing the validity of respondents' answers. This was carried out by comparing the nine trade name franchisees' reported sales and margin levels to the performance information they had earlier provided to Cotter Canada as part of their annual trade name franchise reporting requirements. All nine of the reported margin responses were validated by the Cotter Canada representative, while eight of the reported sales levels were validated, resulting in an overall validity rating of 94.4 percent. In an effort to protect the confidentiality of the respondents, the Canadian office was only permitted to view respondent data in aggregate, thereby upholding the survey's promise of individual store confidentiality.

Response Rates

Of the total sample of 1,169 stores, 340 (29.1 percent) were inaccessible (they had either ceased operations or were unavailable to answer the initial telephone call), 62 (5.3 percent) were incorrectly categorized as retail hardware stores, and 110 (9.4 percent) were in operation and correctly classified as retail hardware stores but nonetheless refused to participate in the study Of the 677 that were correctly classified as retail hardware stores, 370 (31.6 percent of the total sample) agreed to cooperate but failed to follow through by returning their completed survey, while 307 (26.3 percent of the total sample) followed through by sending back their completed surveys. In terms of the total sample, the 307 respondents represent 26.3 percent; in terms of the 787 stores eligible to participate in the study, the 307 responses represent 39.0 percent; and in terms of the 677 stores that agreed to participate, they represent a response rate of 45.3 percent.

Operationalization of Variables

Small retailer performance: sales volume, velocity, and margin. Given the distinction between volume-, velocity- and margin-oriented retail performance (Mason, Mayer, and Ezell 1988), we included three distinct and complementary operationalizations of performance. Volume-oriented performance was operationalized as 1994 sales; velocity-oriented performance as 1994 sales divided by store square footage; and margin-oriented performance as profit margin on 1994 sales after subtracting for cost of goods sold, all operating costs, and taxes. Volume, square footage, and margin operationalizations all utilized scales developed with the assistance of a retail hardware industry expert.

Sunday, evening, and extraordinary accessibility selling practices. The survey included three items measuring the store's conventional (Sunday and evening) and nonconventional (midnight) selling practices. Respondents answered the question "Is your store open on Sundays?" with an unqualified "Yes" or "No." We did not require respondents to specify exactly how many hours their store was open on Sundays. In a separate item, respondents reported the number of evenings their store was open during a typical week. A third item asked respondents to estimate along a five-point scale how often they opened the store up after hours in response to a customer's plea for help with a late night emergency. An answer of one meant never, three sometimes, and five frequently.

External control variables: customer and competitor characteristics. During the qualifying telephone interview, each respondent was asked to identify the zip code areas comprising the critical mass of their store's trading areas. Census data on each of these zip codes was then located and weighted by each area's respective population. Using the most recent US Census data (US Census 1990), we included the habitat-related attribute of weighted per capita income (designated as RESIDENT AFFLUENCE). Likewise, using the most recent U.S. Retail Census data (Economic Census 1992), we also controlled for the number of retail establishments identifying themselves within the SIC 52 category.

Internal control variables: firm size, age, and collective membership. Each respondent identified the store's square footage from a set of scaled store square footage options, while age was reported in number of years the store had been in operation. Given the importance of buying groups in the retail hardware industry (Hardy and Magrath 1987), we also included a categorical variable for trade name franchise membership.

Analysis

Our analysis proceeds in three stages. First, we provide a descriptive profile of the sample and an aggregate summary of Sunday, evening, and extraordinary selling practices. Second, we test the three selling strategies individually across the three performance operationalizations; accordingly, our regression models are grouped by dependent variable. Third, we examine the combined impact of the three access-related tactics on performance.

Descriptive Profile of Sample and Selling Practices

The size and age profiles of the responding stores are presented in Table 1. Stores with less than 5,000 square feet accounted for just over half of our sample, while our average store had been in operation for almost four decades.

Over two-thirds of the respondents (210 or 68.6 percent) reported being open on Sundays, while 96 (31.4 percent) reported being closed. Contingency table analysis reported statistically significant (p=0.0004) differences among cities. Stores in New York, Minneapolis, and Kansas City were most likely to be open; counterparts in Atlanta and Miami were least likely to be open (see Table 2).

In terms of number of evenings open for operation, two basic clusters of stores were discerned. Almost half reported offering no evening hours, while just under 40 percent were open at least five evenings per week (see Table 3). Evening selling varied significantly across the seven cities (see Table 4). Stores in Minneapolis-St. Paul were open more evenings than stores in five other cites, with stores in Long Island open less frequently than stores in four of the other cities (see Table 5).

The mean score for extraordinary accessibility was 2.26 out of five. However, the relative frequency of this practice was negatively skewed (see Table 6), with almost two-thirds of respondents never or infrequently opening up for late night customers.

No statistically significant differences in extraordinary accessibility were observed across the cities (see Table 7). This suggests that in each of the local markets surveyed, a minority of store managers consistently perceived servicing midnight emergencies as a worthwhile retail activity.

Hypothesis Testing: Preliminary Findings

Given three performance operationalizations and three core independent variables, nine regression models were tested. Table 8 reports correlation coefficients for variables utilized in the regression models.

Before discussing the hypothesis-specific findings, a number of interesting supplemental findings warrant mention. The first finding concerns the external environmental control variables. While competitor density is positively related with sales volume and velocity, the relationship with margin is negative. These findings support the earlier articulated logic of niche theory insofar as fewer rivals mean the firm faces less competition for a viable niche. Likewise, while less economically endowed areas are often perceived as less desirable retail venues, our results suggest that such areas nonetheless offer the possibility for higher margin transactions. This may be due to either the absence of other competitors or to the residents' reduced prospects for accessing outlets in more munificent locales.

The three internal control variables also add insight. While store size is positively related to sales and negatively to sales per square foot, it has no apparent relationship with profitabllity. This suggests that small store managers may be mistaken if they assume increased size means improved survival prospects. While firm age appears unrelated to either volume or velocity, it is negatively related to margins. This suggests that many of the older stores in our sample were either deriving income from some other sources, or perhaps were able to insulate themselves from the need to generate higher margins, perhaps through owning their store's building. Also interesting was the finding that membership in a buying group had a positive effect on sales volume and velocity but only at the apparent cost of profit margin. This suggests the possibility that buying group membership is related to subtle but significant shifts in small firm strategy and operation.

Sundays, Evenings, and Extraordinary Accessibility: Performance Implications

The three regression models that were run on the relationship between Sunday opening and the three performance operationalizations (Models 1, 4, and 7 in Tables 9, 10, and 11, respectively) revealed that Sunday openings have no significant relationship with any of the three operationalizations. While this essentially refutes the Sunday-related element of our first hypothesis, it also provides some support for Kiemperer and Padilla's (1997) assertion that small firms are ill equipped to realize advantage from expanding their operating scope to include Sundays.

Regression models on the three performance operationalizations showed no significant relationships between evening hours of operation and any of the three operationalizations of performance (Models 2,5, and 8). Taken together, these first six models suggest a grim assessment--extending the formal scale and scope of a small retail outlet's hours of operations delivers minimal benefit and may only reduce already scarce profits.

Regression models on extraordinary accessibility revealed an interesting pattern (Models 3, 6, and 9). While non-significant relationships were apparent with both sales volume and velocity, a different outcome was apparent concerning sales margin. Making one's store more accessible to late night emergency need was positively related to superior profit margins. This provides support for our second hypothesis that small firm performance is positively related to extraordinary accessibility.

Tactics in Combination: Frequency and Performance Implications

Policies such as those studied here are often not implemented piecemeal, but in tandem with one or more other tactics. In order to explore this possibility, we categorized each firm in terms of Sunday, evening, and midnight selling practices. Stores open five or more evenings were categorized as "high," and those open four or fewer as "low." Stores scoring four or higher on the extraordinary accessibility scale were categorized as "high," those scoring two or three as "low," and those scoring only one as "none." Table 12 reports the comparative frequency of the 18 cells.

Two general patterns warrant mention. First, just under 10 percent of the respondents had highly restricted hours of operation, that is, they were neither open Sundays, evenings, nor after hours. Another 13 percent differed from this first group only by virtue of occasionally opening up for late-night emergencies. At the other end of the continuum, almost one-third (31 percent) of the stores reported high levels of retail accessibility; that is, offering Sunday, evening, and extraordinary access.

A final set of tests incorporating four interaction variables across the three hours-of-operation variables explored the possibilities of small firm performance being enhanced by combining both extraordinary (late night) accessibility and conventional (Sunday and evening) hours of operation. Table 13 reports results from the models using the three performance measures of sales volume, velocity, and margin (Models 10, 11, and 12, respectively).

Two interesting patterns are apparent. First, more appears to only be less. Simultaneously implementing all three tactics results in a negative interaction with both sales volume and velocity performance measures. This suggests the possibility that expanded hours of operation only transfer existing sales rather than create new sales. Therefore, unless there is a clear and defensible market-based rationale for increasing the small store's hours and days of operation, our study suggests such moves are ill-advised. Second, extraordinary accessibility appears to offer a defensible route to greater profitability. Even accounting for the combined impact of all three tactics extraordinary accessibility demonstrates a positive and significant effect on small firm profitability. Why might this be so? We consider this question in our concluding discussion.

Extraordinary Accessibility as Small Retailer Advantage: Rationale and Agenda

We began this article observing that small stores are fundamentally disadvantaged when competing head-to-head against large retailers that seek to leverage their size whenever and however possible. Large retailers are able to buy products in larger quantities at lower prices, gain preferential advertising and rental rates, and hire and deploy human resources for round-the-clock machine-like operation (Mintzberg 1989). The small firm knows no such advantages. Rather, it is intimately acquainted with scarcity and limitation (Welsh and White 1981). This means small firms must utilize caution and prudence in best leveraging their comparatively limited resources. This size-based contrast also suggests that small firms need to exercise wisdom in deciding whether and how to respond to the competitive tactics of large firms.

The ramifications of small stores recognizing and acting within their unique resource set are important for a number of different facets of operation, including the often overlooked area of hours of operation. Drawing on work in niche theory, we propose that superior small firm performance results from executing tactics qualitatively different from those of large players. In the case of hours of operation, we assert that small firms are comparatively advantaged in offering extraordinary accessibility to their customers.

Building on niche theory's distinction between role and habitat, we also assert that a short-lived but significant habitat-based opportunity predictably recurs each night. Consumers characterized by non-ambivalent need (Otnes, Lowrey, and Shrum 1997) face severely restricted retail options when less informally accessible establishments close. However, for entrepreneurial operators, one store's closing can mean another's opening. Results from this investigation support our assertion: while Sunday and evening hours of operation make no discernible contribution to performance, extraordinary accessibility offers a unique and not easily followed route to superior small firm performance.

This finding both challenges and expands existing work on customer need categorization. While classic consumption typologies include such categories as convenience, preference, shopping, and specialty goods (Copeland 1923; Holbrook and Howard 1977; Stell and Donoho 1996), or experience, integration, play, and classification (Holt 1995), our findings suggest the possibility of an additional category--distress. Analogical support for this category can be found in Stern's (1995) work on consumer myths. Using Frye's taxonomy of myth-making, she proposed a four-cell categorization of consumer narratives: comedy, romance, irony, and tragedy. Her cell of tragedy is of particular import here, as extraordinary accessibility relates not to tragedy, but its aversion.

Extraordinary Accessibility: A Research Agenda

Clearly, more work on the tactic of late night accessibility is called for. We envision no less than six key issues arising from this initial exploration. The first issue concerns intra-organizational correlates of extraordinary accessibility. Is a retailer's product and service mix significant in determining the frequency of late night appeals, or are personal selling behaviors oriented to the formation of firm-specific customer relationships more important? Likewise, the second issue concerns the intra-organizational location-specific characteristics. Are some types of stores more likely to be prevailed upon, in part, because of the nature of their physical locale? Third, how effective is extraordinary accessibility in the face of 24-hour-a-day competitors? The firms studied here faced large rivals open seven days a week, but not 24 hours a day. How probable are gains from extraordinary accessibility in the face of one or more perpetually accessible rivals? A fourth issue concerns the inter-sector replicab ility of our findings. Is extraordinary accessibility as a retail strategy confined to retail hardware, or can a bookstore, clothing shop, or bakery also realize improved performance by making itself accessible to after-hour customer need? A fifth issue concerns the distinction between deliberate and emergent strategy (Mintzberg 1989). To what extent is the practice of extraordinary accessibility a deliberate choice rather than just an incidental happening? A final issue concerns the underlying reason for the observed performance advantage. One thesis posits superior margins arising from exploiting customers in distress. However, another hypothesis is also tenable: building on Gouldner's (1960) principle of social reciprocity, could extraordinary accessibility's positive relationship with performance suggest the laying of a nighttime foundation for daytime allegiance?

Conclusion: The Logic of Letting Sleeping Giants Sleep

Although the big box category-killing superstore is a fact of life, its presence need not be a fact of death for the small scale-disadvantaged competitor. The discerning small retailer can find one or more ways to complement and thereby compete alongside the large player. One such tactic involves being extraordinarily accessible to customers. This strategy recognizes the predictable and concurrent occurrence of two phenomena: large stores closing for the night and customers having after-hour emergencies. Seeing this as opportunity for what Schelling termed a "precarious partnership" built on an "incomplete antagonism" (1980, p. 15), the alert small player can address customers' emergency needs. Building on Welsh and White's (1981) observation that a small business is not a "little big business," we conclude by likewise recommending that the small store consider the merits of being a "little awake" store.

Dr. Litz is associate professor of marketing at the Faculty of Management at the University of Manitoba at Winnipeg, Canada. His research interests focus on the nature of small incumbent and niche strategy and family firm management.

Dr. Stewart is at Ohio State University in Columbus, Ohio. Her research interests center on international business strategy and organizational learning.

(*.) The authors gratefully acknowledge the funding provided by the United States Small Business Administration-Office of Advocacy (contract #SBA-8124-Oa-94) and the University of Pittsburgh 1994-9 Central Research Development Fund (grant #5.31883) for this research project.

References

Aldrich, H. (1979). Organizations and Environments. Englewood Cliffs, NJ: Prentice- Hall, Inc.

Ansberry, C. (1988). "Kodak Stakes Out the High End of the Market with New Color Film," The Wall Street Journal, October 5, 16.

Barnes, N. G., and J. Chopoorian (1987). "Small Retailers and Sunday Sales," Journal of Small Business Management 25 (April), 40-46.

Copeland, M. T. (1923). "The Relation of Consumers' Buying Habits to Marketing Methods," Harvard Business Review 1 (April), 282-9.

Dalgic, T., and M. Leeuw (1993). "Niche Marketing Revisited: Theoretical and Practical Issues," paper presented at the annual meeting of the Academy of Marketing Sciences Proceedings, Miami Beach, FL, May.

Darwin, C. (1909). The Foundations of the Origin of the Species. Cambridge, UK: University Press.

Dillman, D.A. (1978). Mail and Telephone Surveys: The Total Design Method. New York: John Wiley and Sons.

Do-It-Yourself-Retailing (1994). "Competing Head-to-Head: In their Own Words, Hardware and Home Improvement Retailers Share their Competitive Strategies," January, 46-54, 69.

Ehrenfeld, T. (1995). "The Demise of Mom and Pop?" Inc. 17(1), 46-48.

Fombrun, C., and M. Shanley (1990). "What's in a Name? Reputation Building and Corporate Strategy," Academy of Management Journal 33(2), 233-258.

Gause, E.F. (1934). The Struggle for Existence. Baltimore, MD: William and Wilkins.

Gouldner, A. (1960). "The Norm of Reciprocity: A Preliminary Statement," American Sociological Review 25(2), 161-178.

Gave, P.G. (1986). Webster's Third International Dictionary. Springfield, MA: Merriam-Webster Inc.

Hannon, M., and J. Freeman (1977). "The Population Ecology of Organizations," American Journal of Sociology 82(5), 929-964.

Hardin, G. (1960). "The Competitive Exclusion Principle," Science 131, 1292-1297.

Hardy, K., and A. Magrath (1987). "Buying Groups: Clout for Small Businesses," Harvard Business Review 63 (September-October), 16-18 20,22,24. Henderson, B. (1983). "The Anatomy of Competition," Journal of Marketing 47(Spring), 7-11.

Holbrook, M.B., and J. Howard (1977). "Frequently Purchased Nondurable Goods and Services," in Selected Aspects of Consumer Behavior. Ed. R. Ferber. Washington, DC: National Science Foundation, 189-222.

Holt, D. (1995). "How Consumers Consume: A Typology of Consumption Practices," Journal of Consumer Research 22(June), 1-16.

Klemperer, P, and A.j. Padilla (1997). "Do Firms' Product Lines Include Too Many Varieties?" RAND Journal of Economics 28(3), 472-488.

La Flamme, T.A. (1988). "Audi's Lifestyle Marketing Plan Targets Athletes of Upscale Sports," Marketing News, August 15, 1-2.

Linneman, R.E., and J.L. Stanton (1991). Making Niche Marketing Work. New York: McGraw-Hill, Inc.

Mason, J.B., M.L. Mayer, and H.F. Ezell (1988). Retailing, third edition. Plano, TX: Business Publications, Inc.

Miles, R., and C. Snow (1978). Organizational Strategy, Structure, and Process. New York: McGraw-Hill Publishing Company.

Miller, C. (1992). "Big Chains Battle for Market Share in Home Improvement," Marketing News 26(20),1, 10-11.

Miller, D., and P Friesen (1984). Organizations: A Quantum view. Englewood Cliffs, NJ: Prentice-Hall, Inc.

Mintzberg, H.(1989). Mintzberg on Management. New York: Free Press.

National Home Center News (1992). "Elmont Store Wreaks Havoc on Local Outlets: Survivors Ponder Future as Home Depot Expands on Long Island, N.Y," December 14, 160, 162, 164.

Nicholson, A.J. (1933). Journal of Animal Ecology 2 (Supplement), 132-178.

Otnes, C., T. M. Lowrey, and L. J. Shrum (1997). "Toward an Understanding of Consumer Ambivalence," Journal of Consumer Research 24(June), 80-93.

Porter, M. (1980). Competitive Strategy. New York: The Free Press. --- (1985). Competitive Advantage. New York: The Free Press.

Raynor, M. (1992). "The Pitfalls of Niche Marketing," Journal of Business Strategy March-April, 29-32.

Schelling, T.C. (1980). The Strategy of Conflict. Cambridge, MA: Harvard University.

Stell, R., and C. Donoho (1996). "Classifying Services From a Consumer Perspective," Journal of Services Marketing 10(6), 33-44.

Stern, B.B. (1995). "Consumer Myths: Frye's Taxonomy and the Structural Analysis of Consumption Text," Journal of Consumer Research 22(September), 165ff.

Thompson, R. (1992). "There's No Place Like Home Depot," Nation's Business 80(2), February, 30-33.

US Census (1990). 1990 Census of Population and Housing Summary, Tape 3B, CD9O-3B-01 through 03. Washington, D.C.: US Government Printing Office.

US Economic Census (1992). 1992 Economic Census Zip Code Statistics, CD-EC92-2A. Washington, D.C.: US Government Printing Office.

Weinstein, A. (1994). Market Segmentation: Using Demographics, Psycho-graphics and Other Niche Marketing Techniques to Predict and Model Customer Behavior. Chicago, IL: Probus Publishing Company.

Welsh, J.A., and J.F. White (1981). "A Small Business Is Not a Little Big Business," Harvard Business Review 59(4) 18-32.

Whittaker, R., S. Levin, and R. Root (1973). "Niche, Habitat, and Ecotope," American Naturalist 107(955), 321-338.

Williamson, O. (1975). Markets and Hierarchies: Analysis and Antitrust Implications. New York: Free Press.

                   Distribution of Firms by Size and Age
Variable              Category                        Frequency Percent
Size (square footage) Less than 2,000 sq. ft.             40     13.2
                      Between 2,000 and 5,000 sq. ft.    121     39.8
                      Over 5,000 sq.ft.                  143     47.0
Age in years (s.d.)   37.95 years (24.84)
              Frequency of Sunday Opening across Seven Cities
                         (percent in parentheses)
City          Open Sundays Closed Sundays Total
Total sample      210            96        306
                 (68.6)        (31.4)     (100)
  Atlanta          23            23        46
                 (50.0)        (50.0)
  Chicago          56            31        87
                 (64.4)        (35.6)
  Kansas City      13            4         17
                 (76.5)        (23.5)
  Miami            8             10        18
                 (44.4)        (55.6)
  Minneapolis      50            13        63
                 (79.4)        (20.6)
  New York         51            9         60
                  (85)          (15)
  San Diego        9             6         15
                  (60)          (40)
Chi-square = 24.88 with 6 df
p = 0.0004
                   Frequency of Number of Evenings Open
Number of Frequency Cumulative         Cumulative
Evenings            Frequency  Percent  Percent
    0        148       148      50.2      50.2
    1         5        153      1.69      51.9
    2        11        164      3.73      55.6
    3         5        169      1.69      57.3
    4         7        176      2.37      59.7
    5        77        253      26.1      85.8
    6        30        283      10.2      95.9
    7        12        295      4.07      100
                    Comparative Frequency of Number of
                     Evenings Open across Seven Cities
City        Expected Cell Mean Cell Count
Atlanta           2.133            45
Chicago           2.635            85
Kansas City       3.294            17
Miami             1.389            18
Minneapolis       4.373            59
New York          0.518            56
San Diego         2.067            15
F = 14.277
p[less than or equal to] 0.0001
                           Post-Hoc Analysis of
               Inter-City Differences in Number of Evenings
Contrasting Pair        Difference Standard Error Probability
Minneapolis - Atlanta     2.23955      0.4632      0.000930
Minneapolis - Chicago     1.73759      0.3966      0.004677
Minneapolis - Miami       2.98399      0.6302      0.001349
Minneapolis - New York   -3.85502      0.4366      0.000000
Minneapolis - San Diego  -2.30621      0.6767      0.075053
New York - Atlanta       -1.61548      0.4685      0.068267
New York - Chicago       -2.11744      0.4028      0.000174
New York - Kansas City   -2.77626      0.6481      0.006452
                   Frequency of Extraordinary Accessibility
                         Cumulative         Cumulative
Score          Frequency Frequency  Percent  Percent
1 (never          114       114      37.7      37.7
2                 81        195      26.8      64.6
3 (sometimes)     53        248      17.5      82.1
4                 20        268      6.62      88.7
5 (frequently)    34        302      11.3      100
           Comparative Frequency of Extraordinary Accessibility
City               Mean (standard deviation)  n
Total sample                 2.25            306
                            (1.32)
  Atlanta                    2.19             46
                            (1.42)
  Chicago                    2.12             88
                            (1.20)
  Kansas City                2.70             17
                            (1.79)
  Miami                      2.22             18
                            (0.87)
  Minneapolis                2.26             63
                            (1.25)
  Long Island (NY)           2.50             60
                            (1.51)
  San Diego                  1.64             14
                            (0.74)
ANOVA result:
F= 1.3349
p = .2413
                         Correlation Coefficients
                  Volume Velocity Margin Income 52 firms  Size   Age   Group
Sales volume
  (total amount)   1.000
Sales velocity
  (sales/sq. ft.)  0.541    1.000
Profit margin
  on sales                 -0.025  0.051  1.000
Per capita
  income                    0.159  0.092 -0.190  1.000
Number of
  "52" firms       0.100    0.090 -0.119 -0.005  1.000
Store size
  (square feet)    0.541   -0.370 -0.082  0.107  0.030    1.000
Store age
  (years)          0.057    0.057 -0.074  0.064  0.014   -0.056  1.000
Buying group
  (1=y, 0=n)       0.292    0.005 -0.129  0.100  0.019    0.354  0.011 1.000
Sunday hours
  (1=y, 0=n)       0.090   -0.072 -0.135  0.184  0.097    0.252 -0.209 0.194
Number
  of evenings      0.144   -0.064 -0.066 -0.007  0.053    0.271 -0.198 0.172
Extraordinary
  Accessibility   -0.011   -0.063  0.074  0.150  0.023    0.052  0.123 0.089
                                   Extra
                  Sundays Evening Access.
Sales volume
  (total amount)
Sales velocity
  (sales/sq. ft.)
Profit margin
  on sales
Per capita
  income
Number of
  "52" firms
Store size
  (square feet)
Store age
  (years)
Buying group
  (1=y, 0=n)
Sunday hours
  (1=y, 0=n)       1.000
Number
  of evenings      0.365   1.000
Extraordinary
  Accessibility    0.041   0.020   1.000
                      Summary of Regression Analysis
              Dependent Variable: Performance as Sales Volume
             (Standardized Betas Reported with t-Ratio Below)
                                                      Model 2
                                        Model 1       Evening
Variable                              Sunday Sales     Sales
External control variables
  Resident affluence
  (Log of weighted per capita income)    0.051         0.023
                                        (0.776)       (0.363)
  Number of competitors                  0.114         0.104
  (SIC 52 category)                     (1.78 [+])    (1.61)
Internal control variables
  Size of store                          0.661         0.6777
  (square footage)                      (9.42 [***])  (9.47 [***])
  Age of store                           0.058         0.064
  (years)                               (0.884)       (0.982)
  Member in buying group                 0.375         0.352
  (1=yes: 0 = no)                       (2.05 [*])    (1.93 [+])
Store hour variables
  Open Sundays                          -0.178
  (I=yes: 0 = no)                      (-1.16)
  Number of evenings open                             -0.017
                                                     (-0.249)
  Extraordinary accessibility
                                         Model 3
                                      Extraordinary
Variable                              Accessibility
External control variables
  Resident affluence
  (Log of weighted per capita income)     0.036
                                         (0.572)
  Number of competitors                   0.108
  (SIC 52 category)                       (1.69 [+])
Internal control variables
  Size of store                           0.645
  (square footage)                       (9.36 [***])
  Age of store                            0.079
  (years)                                (1.23)
  Member in buying group                  0.361
  (1=yes: 0 = no)                        (1.99 [*])
Store hour variables
  Open Sundays
  (I=yes: 0 = no)
  Number of evenings open
  Extraordinary accessibility            -0.057
                                        (-8.880)
Constant              3.591        3.495        3.484
                    (19.6 [***]) (21.4 [***]) (21.4 [***])
[R.sup.2] (percent)  31.0         32.3         30.8
Adjusted [R.sup.2]
(percent)            29.6         30.9         29.3
df                    280          274          281
F                    21.0         21.8         20.9

(+.)p [less than].10

(*.)p [less than].05

(**.)p [less than].01

(***.)p [less than].001

                      Summary of Regression Analysis
             Dependent Variable: Performance as Sales Velocity
             (Standardized Betas Reported with t-Ratio Below)
                                                       Model 5
                                        Model 4        Evening
Variable                              Sunday Sales      Sales
External control variables
  Resident affluence                     0.008          0.008
  (Log of weighted per capita income)   (0.983)        (0.897)
  Number of competitors                  0.017          0.016
  (S.I.C. 52 category)                  (1.96 [+])     (1.89 [+])
Internal control variables
  Size of store                         -0.068         -0.066
  (square footage)                     (-7.15) [***]) (-6.73 [***])
  Age of store                           0.005          0.004
  (Years)                               (0.601)        (0.486)
  Member in buying group                 0.059          0.060
  (1=yes: 0 = no)                       (2.36 [*])     (2.41 [*])
Store hour variables
  Open Sundays                          -0.004
  (1=yes: 0 = no)                      (-0.194)
  Number of evenings open                               0.001
                                                       (0.136)
  Extraordinary accessibility
                                         Model 6
                                      Extraordinary
Variable                              Accessibility
External control variables
  Resident affluence                      0.009
  (Log of weighted per capita income)    (1.09)
  Number of competitors                   0.017
  (S.I.C. 52 category)                   (1.97 [*])
Internal control variables
  Size of store                          -0.069
  (square footage)                      (-7.35 [***])
  Age of store                            0.007
  (Years)                                (0.789)
  Member in buying group                  0.061
  (I=yes: 0 = no)                        (2.47 [*])
Store hour variables
  Open Sundays
  (I=yes: 0 = no)
  Number of evenings open
  Extraordinary accessibility            -0.011
                                         (-1.25)
Constant                      0.155        0.151        0.151
                             (6.16 [***]) (6.78 [***]) (6.76 [***])
[R.sup.2] (percent)          17.4         15.9         17.9
Adjusted [R.sup.2] (percent) 15.6         14.1         16.1
df                           280           274          281
F                             9.84         8.63        10.2
(+.)p [less than] .10
(*.)p [less than] .05
(**.)p [less than] .01
(***.)p [less than] .001
                      Summary of Regression Analysis
              Dependent Variable: Performance as Sales Margin
             (Standardized Betas Reported with t-Ratio Below)
                                                    Model 8
                                        Model 7     Evening
Variable                              Sunday Sales   Sales
External control variables
  Resident affluence
  (Log of weighted per capita income)    -0.243      -0.265
                                        (-2.20 [*]) (-2.45 [*])
  Number of competitors
  (S.I.C. 52 category)                   -0.194      -0.205
                                        (-1.82 [+]) (-1.89 [+])
Internal control variables
  Size of store                          -0.013      -0.021
  (square footage)                      (-0.109)    (-0.176)
  Age of store                           -0.180      -0.153
  (years)                               (-1.64)     (-1.38)
  Member in buying group                 -0.457      -0.501
  (1= yes: 0= no)                       (-1.51)     (-1.65 [+])
Store hour variables
  Open Sundays                           -0.365
  (1= yes: 0= no)                       (-1.44)
  Number of evenings open                            -0.108
                                                    (-0.938)
  Extraordinary accessibility
                                         Model 9
                                      Extraordinary
Variable                              Accessibility
External control variables
  Resident affluence
  (Log of weighted per capita income)    -0.308
                                        (-2.87 [**])
  Number of competitors
  (S.I.C. 52 category)                   -0.210
                                         (1.98 [*])
Internal control variables
  Size of store                          -0.039
  (square footage)                      (-0.347)
  Age of store                           -0.172
  (years)                               (-1.60)
  Member in buying group                 -0.588
  (1= yes: 0= no)                       (-1.97 [+])
Store hour variables
  Open Sundays
  (1= yes: 0= no)
  Number of evenings open
                                          0.253
  Extraordinary accessibility            (2.34 [*])
Constant              4.375        4.157        4.222
                    (14.4 [***]) (15.3 [***]) (15.7 [***])
[R.sup.2] (percent)   6.9          6.2          7.9
Adjusted [R.sup.2]
(percent)             4.9          4.1          5.9
df                   271           266          272
F                     3.37 [*]     2.92 [*]     3.89 [*]
(+.)p [less than].10
(*.)p [less than].05
(**.)p [less than].01
(***.)p [less than].001
                       Combined Frequency of Sunday,
                 Evening, and Extraordinary Accessibility
                         (percent in parentheses)
            Not Open Sundays                   Open Sundays
                   No          Low      High        No        Low      High
                Evenings     Evenings Evenings   Evenings   Evenings Evenings
High Extra.        9                     2          15         8        19
  Access.        (3.03)                (0.67)     (5.05)     (2.69)   (6.40)
Low Extra.         39           4        9                     8        73
  Access.       (13.10)       (1.35)  (11.00)                (2.69)  (24.60)
No Extra.          27                    3          29         8        44
  Access.        (9.09)                (1.01)     (9.76)     (2.69)   (14.8)
          Summary of Regression Analysis for Interaction Effects
     Dependent Variable: Performance as Sales Volume, Velocity, Margin
             (Standardized Betas Reported with t-Ratio Below)
                                                Model 11      Model 12
                                   Model 10      Sales         Sales
Variable                         Sales Volume   Velocity       Margin
External control variables
  Resident affluence (Log of
  weighted per capita income)       0.062         0.011        -0.279
                                   (0.914)       (1.18)       (-2.46 [**])
  Number of competitors
  (S.I.C. 52 category)              0.122         0.018        -0.214
                                  (-1.87 [+])    (2.04 [+])    (1.97 [*])
Internal control variables
  Size of store                     0.682        -0.067        -0.014
  (square footage)                 (9.33 [***]) (-6.66 [***]) (-0.113)
  Age of store                      0.060         0.005        -0.231
  (years)                          (0.901)       (0.605)      (-2.05 [*])
  Member in buying group            0.368         0.060        -0.549
  (1= yes: 0= no)                  (2.01 [*])    (2.39 [*])   (-1.80 [+])
Store hour variables
  Open Sundays                     -0.141         0.005        -0.086
  (1=yes: 0 = no)                 (-0.796)       (0.203)      (-0.293)
  Number of evenings open          -0.102        -0.018        -0.366
                                  (-0.676)      (-0.866)      (-1.47)
  Frequency of midnight openings    0.071         0.011         0.574
                                   (0.411)       (0.454)       (1.98 [*])
Interaction variables
  Sundays x Evenings                                 0.161        0.027
                                                    (0.940)      (1.17)
  Sundays x Extraordinary accessibility             -0.092       -0.019
                                                   (-0.492)     (-0.746)
  Evenings x Extraordinary accessibility             0.235        0.032
                                                    (1.35)       (1.35)
  Sundays x Evenings x Extraordinary accessibility  -0.400       -0.052
                                                   (-2.11 [*])  (-1.99 [*])
Constant                                             3.537        0.142
                                                   (17.5 [***])  (5.12 [***])
[R.sup.2] (percent)                                 34.8         18.4
Adjusted [R.sup.2] (percent)                        31.9         14.7
df                                                  267          267
F                                                   11.9 [***]    5.01 [**]
Interaction variables
  Sundays x Evenings                                 0.344
                                                    (1.20)
  Sundays x Extraordinary accessibility             -0.324
                                                   (-1.04)
  Evenings x Extraordinary accessibility             0.456
                                                    (1.57)
  Sundays x Evenings x Extraordinary accessibility  -0.399
                                                   (-1.27)
Constant                                             4.186
                                                    12.5 [***])
[R.sup.2] (percent)                                  9.9
Adjusted [R.sup.2] (percent)                         5.7
df                                                   259
F                                                    2.38 [*]

(+.)p [less than] .10

(*.)p [less than] .05

(**.)p [less than] .01

(***.)p [less than] .001