Effects of Residential Mobility on the Ratio of Average House Floor Area to Average Household Size

Abstract

The aim of this article is to test the hypothesis that mobile or predominantly mobile societies have a lower ratio of average house floor area to average household size. The analysis is performed on a cross-cultural sample consisting of 11 mobile and 35 sedentary cases. The research hypothesis is supported by the data, and the result is significant in two ways. First, it contributes to general anthropological understanding of relationships between cultural variables. Second, it has implications for demographic reconstructions in archaeology as it provides more specific information for converting observed house floor areas into population size and average household size estimates.

Keywords

archaeology mobility housing population

Introduction

It may be argued that demographic variables are key variables in anthropology, especially in theories that attempt to explain the sociocultural evolution and adaptation of human societies (Baker & Sanders, 1972; Binford, 2001; Boserup, 1965, 1976; Carneiro, 1962, 1970, 1986; Cowgill, 1975; Dewar, 1984; Ember, 1963; Feinman, 2011; Johnson & Earle, 2000; Johnson, 1982; Kosse, 1990, 1994; Peregrine, Ember, & Ember, 2004a, 2004b; Read & LeBlanc, 2003; Wood, 1998), or theories where cultural process is viewed as an evolutionary process in a Darwinian sense (Boyd & Richerson, 1985; Henrich, 2004; Kline & Boyd, 2010; Richerson & Boyd, 2005; Shennan, 2000, 2001, 2002). Given the theoretical importance of demography, archaeologists are developing and refining various methods of measuring and tracking population parameters of human societies through time (Chamberlain, 2006; Hassan, 1978; Schacht, 1980, 1981; Steele, 2010).

Broadly speaking, these efforts have been focused on extracting demographic information from two general classes of archaeological data: skeletal data from funerary contexts and settlement data on various scales (individual house remains, intrasettlement patterns of house remains, or regional distributions of settlements). Skeletal demography lies mainly in the domain of biological and physical anthropology, while archaeologists and cultural anthropologists develop methods for estimating demographic parameters from settlement data.

One of the most important lines of research regarding the domain of settlement data has focused on relating the house floor area to population and household size (Brown, 1987; Casselberry, 1974; Cook & Heizer, 1965; Dohm, 1990; Kolb, 1985; Kramer, 1982; LeBlanc, 1971; Naroll, 1962; Wiessner, 1974), as house remains are in most cases relatively easy to recognize and define in the archaeological record. As the term household can refer to different things (Wilk & Rathje, 1982), it is important to clarify that in this article, it will denote a residential group—people who live in the same house. In any case, the main idea behind previous research has been to determine whether house floor area is a correlate of population size, and if it is—to develop a method which uses this relation by converting house floor area into population estimate.

Naroll (1962) was the first to approach this problem through cross-cultural comparison to determine the ratio of total house floor area to population size. He found a strong relationship between total house floor area and population size (Pearson’s r = .88) and calculated that the slope of the line relating these two variables was 10 m²/person (Naroll, 1962). This empirically derived cross-cultural constant was the key for transforming house floor areas into population estimates. Casselberry (1974) used the same cross-cultural methods with a more specific aim—to calculate the conversion constant only for societies with multifamily households. Casselbery’s empirical estimate of the slope was 6 m²/person. Brown’s (1987) study was the latest in this line of research and the most sophisticated in methodological terms. It also had the largest sample size. Although Naroll tried to relate total population size to total house floor area, Brown tested for the relationship between average household size (AHS) and average house floor area (AHFA). Brown found that the relationship between AHS and AHFA was complex and could not be described with a single regression equation. For this reason, Brown proposed a second-best solution—to calculate the average value of AHFA to AHS ratio (AVRAT) on the cross-cultural level, and to use this value as a conversion constant. Brown found that the cross-cultural average value of AVRAT is 6 m²/person.

The approach taken by Wiessner (1974) was somewhat different. Wiessner’s approach was inspired by Nordbeck’s (1971) article on urban allometric growth. Wiessner (1974) proposed that differently organized settlements require different kinds of equations relating floor area to population. Like Naroll, Wiessner considered the total population size, but unlike Naroll, she used the total settlement area in calculations. The central point of her argument is that the equation relating the settlement area and population must take into account the dimensionality of population distribution in space. For example, hunting camps are often built in such way that dwellings form a circular pattern around the center of the settlement which is devoid of houses. This implies that population is distributed along a single dimension as dwellings are arranged in a circular configuration around the settlement center. Settlement area is, however, a two-dimensional concept. This means that equation which relates the population to settlement area should have this form A = a × P², where A is the settlement area, P is the total population size, and a is a constant. Population size (P) must be raised to the second degree to match the dimensionality of A. Following this logic, Wiessner argued that the equation for villages with rectangular plan (dwellings distributed over the whole area of the settlement) should have the exponent of 1 (as population size now matches the dimensionality of settlement area), whereas the equation for cities with multistorey buildings should have the exponent of 2/3, to compensate for the fact that the population in cities is distributed in three dimensions. It should be noted that an important idea emerges from Wiessner’s (1974) study—organizational factors may influence the relationship between house floor area and population size.

The present study also builds on the idea that different factors may influence the relationship between floor area and population. In principle, it follows Casselberry’s (1974) approach in trying to arrive at a more specific—factor dependent—estimate of conversion formulas. The theory of architectural design (McGuire & Schiffer, 1983) may offer insight about which factors may be important for consideration. Residential mobility is one of the key factors in this theory. Reasoning from this theoretical position would lead to the expectation that mobile or predominantly mobile societies should have lower AVRAT value than sedentary or predominantly sedentary societies. The logic behind this hypothesis is as follows: mobile people tend to make a compromise between different factors and this usually results in building dwellings in a more expedient manner than sedentary people. The most obvious reason is a trade-off between dwelling construction and maintenance costs. From this perspective, it makes sense for mobile people to reduce construction costs by minimizing the average allocation of square meters per person. This is especially so for mobile people in warm and temperate climates who tend to spend more time performing activities outdoors. But this does not have to be the only reason. As mobile people tend to invest less in terms of effort and resources in house construction, their dwellings are usually built from lighter materials (Binford, 1990; Diehl, 1992), which may result in low thermal insulation. In such situations, it would make more sense to have a more “crowded” dwelling. Likewise, in cold climates it is logical to minimize the dwelling space as much as possible to have more efficient heating. It is interesting to note that in such cases AVRAT values would be low because (rather than in spite) of the fact that most of the time would be spent inside the house. For this kind of reasons, it is logical to expect that mobile people will tend to have lower AVRAT than sedentary people.

Therefore, the aim of the present study is to test the hypothesis that residential mobility is inversely correlated with AVRAT and to explore the implications of the results for demographic reconstruction in archaeology.

Data and Method

The data on AVRAT are taken from Brown (1987) as it is the largest cross-cultural sample known to the author where this kind of data is available. The data on residential mobility come from the Standard Cross-Cultural Sample (SCCS; Murdock & White, 1969) for all cases except for 15 cases where data on mobility are taken from the Ethnographic Atlas (EA; Murdock, 1967). Data are presented in Table 1. The total sample size is 46 cases. Some cases from Brown’s study (Highland Scots and Tzeltal) could not be included in the analysis because there was no data on residence patterns available to the author.

Table 1.

Societies in the Sample

Society	AVRAT (m²/person; Brown, 1987)	Mobility^a
Aleut^b	6	2
Amhara^b	6.04	2
Andamanese^b	2.48	1
Aymara^b	1.49	2
Bellacoola^b	22.5	2
Bemba^b	3.6	2
Cagaba^c	6.3	2
Chukchee^b	4.65	1
Copper Eskimo^b	2.58	1
Cubeo (Tucano)^b	3.64	2
Eyak^b	1.9	2
Ganda^b	18.50	2
Garo^b	12.17	2
Hausa^b	4.91	2
Iban^b	16.61	2
Ifugao^b	3.33	2
Ila^c	3.4	2
Iroquois^c	3.5	2
Jivaro^b	6	2
Kanuri^c	8.1	2
Kapauku^b	2.1	2
Kazak^b	9.6	1
Khasi^c	11.8	2
Klamath^b	3.48	1
Koreans^b	10.42	2
Lapps^b	2.74	1
Lau^c	7.1	2
Masai^b	4.22	1
Makiritare^c	5.6	2
Ojibwa^c	1.4	1
Ona^b	1	1
Pawnee^b	6.55	2
Ramkokamekra^c	3.3	1
Mbuti (Pygmies)^b	.3	1
Pomo (Eastern)^b	3.57	2
Serbs^c	8.4	2
Sinhalese^c	12.2	2
Tapirape^c	2.9	2
Tarahumara^c	6	2
Tikopia^b	4.1	2
Tiv^b	7.22	2
Tlingit^c	9.8	2
Tonga^c	3	2
Trukese^b	2.8	2
Wolof ^b	7.41	2
Yanomamo^b	5.12	2

Codes for mobility: 1—mobile or predominantly mobile, 2—sedentary or predominantly sedentary.

Data on residential mobility from SCCS (Murdock & White, 1969); variable Fixity of Residence. Modes migratory, seminomadic-fixed then migratory, rotating among 2+ fixed were recoded as 1, whereas modes semisedentary-fixed core, some migratory, impermanent-periodically moved, permanent were recoded as 2.

Data on residential mobility from Ethnographic Atlas (Murdock, 1967); variable Settlement Pattern. Modes nomadic or fully migratory and seminomadic were recoded as 1, whereas modes semisedentary, dispersed family homesteads, compact and relatively permanent, and complex were recoded as 2.

The fixity of residence variable from the SCCS was split along the median and recoded into a new dichotomous variable with two modes: (a) “mobile and predominantly mobile” (b) “sedentary and predominantly sedentary communities.” I will refer to this new variable as mobility. For cases which are not present in the SCCS, information about mobility was extracted from the Settlement pattern variable from the EA (see notes in Table 1 for the details of the recoding procedure).

After the original variables had been recoded, there were 11 mobile or predominantly mobile cases and 35 sedentary or predominantly sedentary cases in the sample. The practical reason behind this recoding is that archaeologists are rarely in position to make fine distinctions in degree and type of residential mobility, so this crude dichotomy may be more appropriate if the archaeological implications of the results are to be considered.

Descriptive statistics, histograms and boxplots for AVRAT were calculated and constructed for both groups (mobile and sedentary). To formally test the research hypothesis, a t-test and Mann-Whitney U-test were performed with AVRAT as dependent, and mobility as an independent variable. Both mobile and sedentary AVRAT distributions are skewed: mobile category skewness = 1.715, SE = .661; sedentary category skewness = 1.589, SE = .398; skewness statistics calculated with SPSS 16 (for the interpretation of this particular skewness statistic, see Tabachnick & Fidell, 2007, pp. 79-83). The rule of thumb is to consider skewness values that are two times larger than corresponding standard errors as indicative of significant asymmetry. For this reason, the t-test was performed on logarithmically transformed AVRAT values. When logarithmic transformation is applied, skewness values are as follows: mobile category skewness = -1.079, SE = .661; sedentary category skewness = .105, SE = .398. These values are acceptable for the t-test. It should be emphasized that conclusions about means of transformed distributions apply to medians of untransformed distributions (Tabachnick & Fidell, 2007, p. 87)

Results

Figure 1 shows boxplots whereas Figure 2 shows histograms of the AVRAT variable for both categories of the mobility variable. The mean value of AVRAT in the mobile category is 3.25 m²/person, the median is 2.74 m²/person, and the standard deviation is 2.48 m²/person. The mean value of AVRAT in the sedentary category is 6.97 m²/person, the median is 6 m²/person, and the standard deviation is 4.82 m²/person.

Figure 1.

Boxplots for AVRAT

Figure 2.

Histograms for AVRAT

The t-test results suggest that the difference in AVRAT between mobile and sedentary societies is statistically significant at the 0.001 level (Levene’s test: F = .656, p = .422; t = -3.474, df = 44, one-tailed p = .0006). Results of the Mann-Whitney test also suggest that there is a statistically significant difference between mobile and sedentary societies (Mann-Whitney U = 76, one-tailed p = .0015).

Discussion and Conclusion

The results of the formal statistical analysis support the hypothesis that sedentary societies on average have higher AVRAT than mobile societies. However, it is clear from the distribution of AVRAT that mobility alone cannot account for the entire variation in the group of sedentary societies (Figure 2). The situation in the mobile group is not problematic—there is only one extreme outlier, but most cases tend to cluster very tightly around the mean (Figure 1). However, it should be kept in mind that the sample size of mobile societies is low (11 cases). In contrast, the variability of AVRAT is rather high in the sedentary group. At present, the factors which cause this variability in the sedentary category are unknown.

What are the implications of these results for archaeological research? First, it is evident that it would be better to use different AVRAT values for communities with different degrees of mobility when attempting population size and AHS reconstruction. For mobile groups, the average AVRAT value is 3.25 m²/person, so this value should be preferred to the cross-cultural average of 6 m²/person calculated by Brown (1987).

Although there is no dilemma regarding whether the group specific AVRAT value is better than the cross-cultural average for mobile groups, the situation is rather more complicated for sedentary societies. If the goal is to reconstruct the settlement population from data on house remains, it does not seem to make a great difference whether the group-specific AVRAT of 6.97 (≈ 7) m²/person or the cross-cultural average of 6 m²/person is used. This point can be demonstrated through a simple exercise. In Table 2 data on AHS, AHFA, and population size from six sedentary societies available from the literature are presented. The last two columns show the reconstructed population size using Brown’s cross-cultural constant and the sedentary group-specific mean value, respectively. Even though the average error in total population estimate is lower when the group-specific value is used (paired samples t-test, t = -5.984, df = 5, one-tailed p = .001), the orders of magnitude of the population estimates are the same. This is why it does not matter much which of the two estimates is used for sedentary societies—archaeologists usually aim to arrive at rough approximations of population size rather than an exact count of people (e.g., Drennan & Dai, 2010). The true problem when sedentary communities are concerned lies in the fact that the variation around the mean AVRAT is considerable. This might create considerable errors in population size estimation, and lead to even greater problems with the reconstruction of AHS.

Table 2.

Population Size Reconstruction With Two Different Conversion Constants

Settlement	Reference	AHS	AHFA (m²)	N of households	Observed population size	Estimated population size (AVRAT = 6 m²/p)	Estimated population size (AVRAT = 7 m²/p)
Pinosava	(Kojić, 1973, pp.120, Table 187)	5.63	36.74	16	90	97.98	83.99
San Juan la Laguna	(Kolb, 1985)	5.42	40.98	24	130	163.92	140.50
Santa Catarina Palopo	(Kolb, 1985)	4.55	32.24	11	50	59.11	50.66
Baghestan	(Blanton, 1994)^a	6.21	31.75	24	149	127.00	108.86
Hasanabad	(Blanton, 1994)^a	14.08	74.00	13	183	160.33	137.43
Darnaj	(Blanton, 1994)^a	9.20	133.5	10	92	222.50	190.71

Calculations of AHS and AHFA are based on data from Appendices in Blanton (1994); there are other communities in Blanton’s sample, but data on household size, house floor area or both, is often lacking for houses from these communities.

AHS reconstruction in the group of sedentary cases is problematic as archaeologists are often interested to infer the type of household from its size. Even with good AHS estimates it is difficult, if not impossible, for archaeologists to do more than discriminate between single family and multifamily households. With incorrect AHS estimates even this simple distinction becomes difficult to make. For example, one could imagine a hypothetical society with an AVRAT value close to 12 m²/person, a high AHFA value (e.g., around 60 m²), and the actual AHS of around five persons. If the sedentary group-specific mean AVRAT value (≈ 7 m²/person) was used for AHS reconstruction, one would arrive at a value of 8.6 persons as an AHS estimate. This estimation would lead the archaeologist to conclude that average household in this hypothetical society consisted of some sort of extended families. This would obviously be an erroneous conclusion as the actual AHS value is closer to AHS for single (probably nuclear) family households.

To summarize, this study found that residential mobility has a significant effect on the ratio of average house floor area to average household size. As expected, this ratio is on average lower in societies which are mobile or predominantly mobile. By using separate conversion constants for mobile and sedentary societies, it is possible to estimate population size from house remains more accurately. This also holds for estimating the average household size, with the cautionary note that estimating average household size within the sedentary group may still be associated with a potentially large error. More research is needed to identify factors which cause variability within the group of sedentary societies and to make more specific empirical generalizations regarding the conversion constant.

At present, one can only speculate about which factors lie behind this variability. For example, one potential factor may be population nucleation judging by results presented by Dohm (1990). The structure of dwellings (e.g., number of rooms, room specialization, and spatial configuration) may also have an influence on the AVRAT—Blanton’s (1994) research has shown that structural attributes of houses are often related to social attributes of households.

The general conclusion is that the potential value of future research aiming to identify additional variables that explain variation in AVRAT would be two-fold. It would produce new knowledge about the interconnections between anthropologically relevant variables and material culture (housing and the use of space), and it would advance archaeologists’ ability to reconstruct demographic parameters.

Footnotes

Acknowledgements

The author wishes to thank the editor and the anonymous reviewers for useful comments and suggestions. The author is especially grateful to David Orton for providing help with language and style corrections.

The responsibility for errors, shortcomings, and omissions is exclusively the author’s.

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was undertaken as a part of the project No. 177008 funded by the Ministry of Science and Technological Development of the Republic of Serbia.

References

Baker

P. T.

Sanders

W. T.

(1972). Demographic Studies in anthropology. Annual Review of Anthropology, 1, 151-178.

Binford

L. R.

(1990). Mobility, housing, and environment: A comparative study. Journal of Anthropological Research, 46, 119-152.

Binford

L. R.

(2001). Constructing frames of reference: An analytical method for the archaeological use of hunter-gatherer and environmental data sets. Berkeley, CA: University of California Press.

Blanton

R. E.

(1994). Houses and households: A comparative study. New York, NY: Plenum.

Boserup

(1965). The conditions of agricultural growth. Chicago, IL: Aldine.

Boserup

(1976). Environment, population, and technology in primitive societies. Population and Development Review, 2, 21-36.

Boyd

Richerson

P. J.

(1985). Culture and the evolutionary process. Chicago, IL: University of Chicago Press.

Brown

B. M.

(1987). Population estimation from floor area: A restudy of “Naroll’s constant.” Cross-Cultural Research, 21, 1-49.

Carneiro

R. L.

(1962). Scale analysis as an instrument for the study of cultural evolution. Southwestern Journal of Anthropology, 18, 149-169.

10.

Carneiro

R. L.

(1970). A theory of the origin of the state. Science, 169, 733-738.

11.

Carneiro

R. L.

(1986). On the relationship between size of population and complexity of social organization. Journal of Anthropological Research, 42, 355-364.

12.

Casselberry

S. E.

(1974). Further refinement of formulae for determining population from floor area. World Archaeology, 6, 117-122.

13.

Chamberlain

A. T.

(2006). Demography in archaeology. Cambridge, UK: Cambridge University Press.

14.

Cook

S. F.

Heizer

R. F.

(1965). The quantitative approach to the relation between population and settlement size. Berkeley, CA: University of California Press.

15.

Cowgill

G. L.

(1975). Causes and consequences of ancient and modern population changes. American Anthropologist, 77, 505-525.

16.

Dewar

R. E.

(1984). Environmental productivity, population regulation, and carrying capacity. American Anthropologist, 86, 601-614.

17.

Diehl

M. W.

(1992). Architecture as a material correlate of mobility strategies: Some implications for archeological interpretation. Cross-Cultural Research, 26, 1-35.

18.

Dohm

(1990). Effect of population nucleation on house size for pueblos in the American Southwest. Journal of Anthropological Archaeology, 9, 201-239.

19.

Drennan

R. D.

Dai

(2010). Chiefdoms and states in the Yuncheng Basin and the Chifeng region: A comparative analysis of settlement systems in North China. Journal of Anthropological Archaeology, 29, 455-468.

20.

Ember

(1963). The relationship between economic and political development in nonindustrialized societies. Ethnology, 2, 228-248.

21.

Feinman

G. M.

(2011). Size, complexity, and organizational variation: A comparative approach. Cross-Cultural Research, 45, 37-58.

22.

Hassan

F. A.

(1978). Demographic archaeology. In Schiffer

M. B.

(Ed.), Advances in archaeological method and theory (Vol. 1, pp. 49-103). New York, NY: Academic.

23.

Henrich

(2004). Demography and cultural evolution: How adaptive cultural processes can produce maladaptive losses: The Tasmanian case. American Antiquity, 69, 197-214.

24.

Johnson

A. W.

Earle

(2000). The evolution of human societies: From foraging group to agrarian state. Stanford, CA: Stanford University Press.

25.

Johnson

G. A.

(1982). Organizational structure and scalar stress. In Renfrew

Rowlands

Segraves

(Eds.), Theory and explanation in archaeology (pp. 389-421). New York, NY: Academic.

26.

Kline

M. A.

Boyd

(2010). Population size predicts technological complexity in Oceania. Proceedings of the Royal Society, B series, Biological sciences, 277, 2559-2564.

27.

Kojić

(1973). Seoska arhitektura i rurizam: teorija i elementi. [Rural architecture and rural settlement planning: theory and elements]. Beograd, Serbia: Građevinska knjiga.

28.

Kolb

C. C.

(1985). Demographic estimates in archaeology: Contributions from ethnoarchaeology on Mesoamerican peasants. Current Anthropology, 26, 581-599.

29.

Kosse

(1990). Group size and societal complexity: Thresholds in the long-term memory. Journal of Anthropological Archaeology, 9, 275-303.

30.

Kosse

(1994). The evolution of large, complex groups: A hypothesis. Journal of Anthropological Archaeology, 13, 35-50.

31.

Kramer

(1982). Ethnographic households and archaeological interpretation: A case from Iranian Kurdistan. American Behavioral Scientist, 25, 663-675.

32.

LeBlanc

(1971). An addition to Naroll’s suggested floor area and settlement population relationship. American Antiquity, 36, 210-211.

33.

McGuire

R. H.

Schiffer

M. B.

(1983). A theory of architectural design. Journal of Anthropological Archaeology, 2, 277-303.

34.

Murdock

G. P.

(1967). Ethnographic atlas. Pittsburgh, PA: University of Pittsburgh Press.

35.

Murdock

G. P.

White

D. R.

(1969). Standard cross-cultural sample. Ethnology, 8, 329-369.

36.

Naroll

(1962). Floor area and settlement population. American Antiquity, 27, 587-589.

37.

Nordbeck

(1971). Urban allometric growth. Geografiska Annaler, 53, 54-67.

38.

Peregrine

P. N.

Ember

C. R.

Ember

(2004a). Universal patterns in cultural evolution: An empirical analysis using Guttman scaling. American Anthropologist, 106, 145-149.

39.

Peregrine

P. N.

Ember

C. R.

(2004b). Predicting the future state of the world using archaeological data: An exercise in archaeomancy. Cross-Cultural Research, 38, 133-146.

40.

Read

D. Â. W.

LeBlanc

S. Â. A.

(2003). Population growth, carrying capacity, and conflict. Current Anthropology, 44, 59-85.

41.

Richerson

P. J.

Boyd

(2005). Not by genes alone: How culture transformed human evolution. Chicago, IL: University of Chicago Press.

42.

Schacht

R. M.

(1980). Two models of population growth. American Anthropologist, 82, 782-798.

43.

Schacht

R. M.

(1981). Estimating past population trends. Annual Review of Anthropology, 10, 119-140.

44.

Shennan

(2000). Population, culture history, and the dynamics of culture change. Current Anthropology, 41, 811-835.

45.

Shennan

(2001). Demography and cultural innovation: A model and its implications for the emergence of modern human culture. Cambridge Archaeological Journal, 11, 5-16.

46.

Shennan

(2002). Genes, memes and human history: Darwinian archaeology and cultural evolution. London, UK: Thames & Hudson.

47.

Steele

(2010). Radiocarbon dates as data: Quantitative strategies for estimating colonization front speeds and event densities. Journal of Archaeological Science, 37, 2017-2030.

48.

Tabachnick

B. G.

Fidell

L. S.

(2007). Using multivariate statistics. London, UK: SAGE.

49.

Wiessner

(1974). A functional estimator of population from floor area. American Antiquity, 39, 343-350.

50.

Wilk

R. R.

Rathje

W. L.

(1982). Household archaeology. American Behavioral Scientist, 25, 617-639.

51.

Wood

J. W.

(1998). A theory of preindustrial population dynamics: Demography, economy, and well-being in Malthusian systems. Current Anthropology, 39, 99-135.